1: %TCIDATA{Version=4.00.0.2321}

2: %TCIDATA{LaTeXparent=0,0,Advances2.tex}

3:                       

4: %TCIDATA{ChildDefaults=chapter:1,page:1}

5: 

6: 

7: \chapter[Other Inequalities]{Other Inequalities in Inner Product Spaces}

8: 

9: \label{ch6}

10: 

11: \section{Bounds for the Distance to Finite-Dimensional Subspaces}

12: 

13: \subsection{Introduction}

14: 

15: Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ be an inner

16: product space over the real or complex number field $\mathbb{K}$, $\left\{

17: y_{1},\dots ,y_{n}\right\} $ a subset of $H$ and $G\left( y_{1},\dots

18: ,y_{n}\right) $ the \textit{Gram matrix} of $\left\{ y_{1},\dots

19: ,y_{n}\right\} $ where $\left( i,j\right) -$entry is $\left\langle

20: y_{i},y_{j}\right\rangle .$ The determinant of $G\left( y_{1},\dots

21: ,y_{n}\right) $ is called the \textit{Gram determinant} of $\left\{

22: y_{1},\dots ,y_{n}\right\} $ and is denoted by $\Gamma \left( y_{1},\dots

23: ,y_{n}\right) .$ Thus,%

24: \begin{equation*}

25: \Gamma \left( y_{1},\dots ,y_{n}\right) =\left\vert 

26: \begin{array}{c}

27: \left\langle y_{1},y_{1}\right\rangle \;\left\langle

28: y_{1},y_{2}\right\rangle \;\cdots \;\left\langle y_{1},y_{n}\right\rangle 

29: \\ 

30: \left\langle y_{2},y_{1}\right\rangle \;\left\langle

31: y_{2},y_{2}\right\rangle \;\cdots \;\left\langle y_{2},y_{n}\right\rangle 

32: \\ 

33: \cdots \cdots \cdots \cdots \cdots  \\ 

34: \left\langle y_{n},y_{1}\right\rangle \;\left\langle

35: y_{n},y_{2}\right\rangle \;\cdots \;\left\langle y_{n},y_{n}\right\rangle 

36: \end{array}%

37: \right\vert .

38: \end{equation*}

39: 

40: Following \cite[p. 129 -- 133]{DExx}, we state here some general results for

41: the Gram determinant that will be used in the sequel.

42: 

43: \begin{enumerate}

44: \item Let $\left\{ x_{1},\dots ,x_{n}\right\} \subset H.$ Then $\Gamma

45: \left( x_{1},\dots ,x_{n}\right) \neq 0$ if and only if $\left\{ x_{1},\dots

46: ,x_{n}\right\} $ is linearly independent;

47: 

48: \item Let $M=span\left\{ x_{1},\dots ,x_{n}\right\} $ be $n-$dimensional in $%

49: H,$ i.e., $\{x_{1},\dots ,$ $x_{n}\}$ is linearly independent. Then for each 

50: $x\in H,$ the distance $d\left( x,M\right) $ from $x$ to the linear subspace 

51: $H$ has the representations%

52: \begin{equation}

53: d^{2}\left( x,M\right) =\frac{\Gamma \left( x_{1},\dots ,x_{n},x\right) }{%

54: \Gamma \left( x_{1},\dots ,x_{n}\right) }  \label{upper1.1}

55: \end{equation}%

56: and%

57: \begin{equation}

58: d^{2}\left( x,M\right) =\left\Vert x\right\Vert ^{2}-\beta ^{T}G^{-1}\beta ,

59: \label{upper1.2}

60: \end{equation}%

61: where $G=G\left( x_{1},\dots ,x_{n}\right) ,$ $G^{-1}$ is the inverse matrix

62: of $G$ and 

63: \begin{equation*}

64: \beta ^{T}=\left( \left\langle x,x_{1}\right\rangle ,\left\langle

65: x,x_{2}\right\rangle ,\dots ,\left\langle x,x_{n}\right\rangle \right) ,

66: \end{equation*}%

67: denotes the transpose of the column vector $\beta .$

68: 

69: Moreover, one has the simpler representation%

70: \begin{equation}

71: d^{2}\left( x,M\right) =\left\{ 

72: \begin{array}{ll}

73: \left\Vert x\right\Vert ^{2}-\frac{\left( \sum_{i=1}^{n}\left\vert

74: \left\langle x,x_{i}\right\rangle \right\vert ^{2}\right) ^{2}}{\left\Vert

75: \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert ^{2}} & 

76: \text{if \ }x\notin M^{\perp }, \\ 

77: &  \\ 

78: \left\Vert x\right\Vert ^{2} & \text{if \ }x\in M^{\perp },%

79: \end{array}%

80: \right.  \label{upper1.3}

81: \end{equation}%

82: where $M^{\perp }$ denotes the orthogonal complement of $M.$

83: 

84: \item Let $\left\{ x_{1},\dots ,x_{n}\right\} $ be a set of nonzero vectors

85: in $H.$ Then%

86: \begin{equation}

87: 0\leq \Gamma \left( x_{1},\dots ,x_{n}\right) \leq \left\Vert

88: x_{1}\right\Vert ^{2}\left\Vert x_{2}\right\Vert ^{2}\cdots \left\Vert

89: x_{n}\right\Vert ^{2}.  \label{upper1.4}

90: \end{equation}%

91: The equality holds on the left (respectively right) side of (\ref{upper1.4})

92: if and only if $\left\{ x_{1},\dots ,x_{n}\right\} $ is linearly dependent

93: (respectively orthogonal). The first inequality in (\ref{upper1.4}) is known

94: in the literature as \textit{Gram's inequality} while the second one is

95: known as \textit{Hadamard's inequality.}

96: 

97: \item If $\left\{ x_{1},\dots ,x_{n}\right\} $ is an orthonormal set in $H,$

98: i.e., $\left\langle x_{i},x_{j}\right\rangle =\delta _{ij},$ $i,j\in \left\{

99: 1,\dots ,n\right\} ,$ where $\delta _{ij}$ is Kronecker's delta, then%

100: \begin{equation}

101: d^{2}\left( x,M\right) =\left\Vert x\right\Vert

102: ^{2}-\sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

103: ^{2}.  \label{upper1.5}

104: \end{equation}

105: \end{enumerate}

106: 

107: The following inequalities which involve Gram determinants may be stated as

108: well \cite[p. 597]{MPFxx}:%

109: \begin{equation}

110: \frac{\Gamma \left( x_{1},\dots ,x_{n}\right) }{\Gamma \left( x_{1},\dots

111: ,x_{k}\right) }\leq \frac{\Gamma \left( x_{2},\dots ,x_{n}\right) }{\Gamma

112: \left( x_{1},\dots ,x_{k}\right) }\leq \cdots \leq \Gamma \left(

113: x_{k+1},\dots ,x_{n}\right) ,  \label{upper1.6}

114: \end{equation}%

115: \begin{equation}

116: \Gamma \left( x_{1},\dots ,x_{n}\right) \leq \Gamma \left( x_{1},\dots

117: ,x_{k}\right) \Gamma \left( x_{k+1},\dots ,x_{n}\right)  \label{upper1.7}

118: \end{equation}%

119: and%

120: \begin{multline}

121: \quad \Gamma ^{\frac{1}{2}}\left( x_{1}+y_{1},x_{2},\dots ,x_{n}\right)

122: \label{upper1.8} \\

123: \leq \Gamma ^{\frac{1}{2}}\left( x_{1},x_{2},\dots ,x_{n}\right) +\Gamma ^{%

124: \frac{1}{2}}\left( y_{1},x_{2},\dots ,x_{n}\right) .\quad

125: \end{multline}

126: 

127: The main aim of this section is to point out some upper bounds for the

128: distance $d\left( x,M\right) $ in terms of the linearly independent vectors $%

129: \left\{ x_{1},\dots ,x_{n}\right\} $ that span $M$ and $x\notin M^{\perp },$

130: where $M^{\perp }$ is the orthogonal complement of $M$ in the inner product

131: space $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $.

132: 

133: As a by-product of this endeavour, some refinements of the generalisations

134: for Bessel's inequality due to several authors including: Boas, Bellman and

135: Bombieri are obtained. Refinements for the well known Hadamard's inequality

136: for Gram determinants are also derived.

137: 

138: \subsection{Upper Bounds for $d\left( x,M\right) $}

139: 

140: The following result may be stated \cite{SILV1xx}.

141: 

142: \begin{theorem}[Dragomir, 2005]

143: \label{uppert2.1}Let $\left\{ x_{1},\dots ,x_{n}\right\} $ be a linearly

144: independent system of vectors in $H$ and $M:=span\left\{ x_{1},\dots

145: ,x_{n}\right\} .$ If $x\notin M^{\perp },$ then%

146: \begin{equation}

147: d^{2}\left( x,M\right) <\frac{\left\Vert x\right\Vert

148: ^{2}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}-\sum_{i=1}^{n}\left\vert

149: \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{n}\left\Vert

150: x_{i}\right\Vert ^{2}}  \label{upper2.1}

151: \end{equation}%

152: or, equivalently,%

153: \begin{multline}

154: \Gamma \left( x_{1},\dots ,x_{n},x\right)  \label{upper2.2} \\

155: <\frac{\left\Vert x\right\Vert ^{2}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

156: ^{2}-\sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

157: ^{2}}{\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}}\cdot \Gamma \left(

158: x_{1},\dots ,x_{n}\right) .

159: \end{multline}

160: \end{theorem}

161: 

162: \begin{proof}

163: If we use the Cauchy-Bunyakovsky-Schwarz type inequality%

164: \begin{equation}

165: \left\Vert \sum_{i=1}^{n}\alpha _{i}y_{i}\right\Vert ^{2}\leq

166: \sum_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\sum_{i=1}^{n}\left\Vert

167: y_{i}\right\Vert ^{2},  \label{upper2.3}

168: \end{equation}%

169: that can be easily deduced from the obvious identity%

170: \begin{equation}

171: \sum_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\sum_{i=1}^{n}\left\Vert

172: y_{i}\right\Vert ^{2}-\left\Vert \sum_{i=1}^{n}\alpha _{i}y_{i}\right\Vert

173: ^{2}=\frac{1}{2}\sum_{i,j=1}^{n}\left\Vert \overline{\alpha _{i}}x_{j}-%

174: \overline{\alpha _{j}}x_{i}\right\Vert ^{2},  \label{upper2.4}

175: \end{equation}%

176: we can state that%

177: \begin{equation}

178: \left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert

179: ^{2}\leq \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

180: \right\vert ^{2}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}.

181: \label{upper2.5}

182: \end{equation}%

183: Note that the equality case holds in (\ref{upper2.5}) if and only if, by (%

184: \ref{upper2.4}), 

185: \begin{equation}

186: \overline{\left\langle x,x_{i}\right\rangle }x_{j}=\overline{\left\langle

187: x,x_{i}\right\rangle }x_{i}  \label{upper2.6}

188: \end{equation}%

189: for each $i,j\in \left\{ 1,\dots ,n\right\} .$

190: 

191: Utilising the expression (\ref{upper1.3}) of the distance $d\left(

192: x,M\right) $, we have%

193: \begin{equation}

194: d^{2}\left( x,M\right) =\left\Vert x\right\Vert ^{2}-\frac{%

195: \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

196: ^{2}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}}{\left\Vert

197: \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert ^{2}}\cdot 

198: \frac{\sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

199: ^{2}}{\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}}.  \label{upper2.7}

200: \end{equation}%

201: Since $\left\{ x_{1},\dots ,x_{n}\right\} $ are linearly independent, hence (%

202: \ref{upper2.6}) cannot be achieved and then we have strict inequality in (%

203: \ref{upper2.5}).

204: 

205: Finally, on using (\ref{upper2.5}) and (\ref{upper2.7}) we get the desired

206: result (\ref{upper2.1}).

207: \end{proof}

208: 

209: \begin{remark}

210: \label{upperr2.2}It is known that (see (\ref{upper1.4})) if not all $\left\{

211: x_{1},\dots ,x_{n}\right\} $ are orthogonal on each other, then the

212: following result, which is well known in the literature as Hadamard's

213: inequality holds:%

214: \begin{equation}

215: \Gamma \left( x_{1},\dots ,x_{n}\right) <\left\Vert x_{1}\right\Vert

216: ^{2}\left\Vert x_{2}\right\Vert ^{2}\cdots \left\Vert x_{n}\right\Vert ^{2}.

217: \label{upper2.8}

218: \end{equation}%

219: Utilising the inequality (\ref{upper2.2}), we may write successively:%

220: \begin{align*}

221: \Gamma \left( x_{1},x_{2}\right) & \leq \frac{\left\Vert x_{1}\right\Vert

222: ^{2}\left\Vert x_{2}\right\Vert ^{2}-\left\vert \left\langle

223: x_{2},x_{1}\right\rangle \right\vert ^{2}}{\left\Vert x_{1}\right\Vert ^{2}}%

224: \left\Vert x_{1}\right\Vert ^{2}\leq \left\Vert x_{1}\right\Vert

225: ^{2}\left\Vert x_{2}\right\Vert ^{2}, \\

226: \Gamma \left( x_{1},x_{2},x_{3}\right) & <\frac{\left\Vert x_{3}\right\Vert

227: ^{2}\sum_{i=1}^{2}\left\Vert x_{i}\right\Vert ^{2}-\sum_{i=1}^{2}\left\vert

228: \left\langle x_{3},x_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{2}\left%

229: \Vert x_{i}\right\Vert ^{2}}\Gamma \left( x_{1},x_{2}\right)  \\

230: & \leq \left\Vert x_{3}\right\Vert ^{2}\Gamma \left( x_{1},x_{2}\right)  \\

231: & \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots \cdots

232: \cdots \cdots \cdots \cdots  \\

233: \Gamma \left( x_{1},\dots ,x_{n-1},x_{n}\right) & <\frac{\left\Vert

234: x_{n}\right\Vert ^{2}\sum_{i=1}^{n-1}\left\Vert x_{i}\right\Vert

235: ^{2}-\sum_{i=1}^{n-1}\left\vert \left\langle x_{n},x_{i}\right\rangle

236: \right\vert ^{2}}{\sum_{i=1}^{n-1}\left\Vert x_{i}\right\Vert ^{2}} \\

237: & \qquad \qquad \times \Gamma \left( x_{1},\dots ,x_{n-1}\right)  \\

238: & \leq \left\Vert x_{n}\right\Vert ^{2}\Gamma \left( x_{1},\dots

239: ,x_{n-1}\right) .

240: \end{align*}%

241: Multiplying the above inequalities, we deduce%

242: \begin{align}

243: & \Gamma \left( x_{1},\dots ,x_{n-1},x_{n}\right)   \label{upper2.9} \\

244: & <\left\Vert x_{1}\right\Vert ^{2}\prod_{k=2}^{n}\left( \left\Vert

245: x_{k}\right\Vert ^{2}-\frac{1}{\sum_{i=1}^{k-1}\left\Vert x_{i}\right\Vert

246: ^{2}}\sum_{i=1}^{k-1}\left\vert \left\langle x_{k},x_{i}\right\rangle

247: \right\vert ^{2}\right)   \notag \\

248: & \leq \prod_{j=1}^{n}\left\Vert x_{j}\right\Vert ^{2},  \notag

249: \end{align}%

250: valid for a system of $n\geq 2$ linearly independent vectors which are not

251: orthogonal on each other.

252: \end{remark}

253: 

254: In \cite{DRA1xx}, the author has obtained the following inequality.

255: 

256: \begin{lemma}[Dragomir, 2004]

257: \label{upperl2.3}Let $z_{1},\dots ,z_{n}\in H$ and $\alpha _{1},\dots

258: ,\alpha _{n}\in \mathbb{K}$. Then one has the inequalities:%

259: \begin{equation}

260: \left\Vert \sum_{i=1}^{n}\alpha _{i}z_{i}\right\Vert ^{2}  \label{upper2.10}

261: \end{equation}%

262: \begin{multline*}

263: \leq \left\{ 

264: \begin{array}{l}

265: \max\limits_{1\leq i\leq n}\left\vert \alpha _{i}\right\vert

266: ^{2}\sum\limits_{i=1}^{n}\left\Vert z_{i}\right\Vert ^{2}; \\ 

267: \\ 

268: \left( \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2\alpha

269: }\right) ^{\frac{1}{\alpha }}\left( \sum\limits_{i=1}^{n}\left\Vert

270: z_{i}\right\Vert ^{2\beta }\right) ^{\frac{1}{p}} \\ 

271: \hfill \text{where \ }\alpha >1,\ \frac{1}{\alpha }+\frac{1}{\beta }=1; \\ 

272: \\ 

273: \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert

274: ^{2}\max\limits_{1\leq i\leq n}\left\Vert z_{i}\right\Vert ^{2};%

275: \end{array}%

276: \right. \\

277: +\left\{ 

278: \begin{array}{l}

279: \max\limits_{1\leq i\neq j\leq n}\left\{ \left\vert \alpha _{i}\alpha

280: _{j}\right\vert \right\} \sum\limits_{1\leq i\neq j\leq n}\left\vert

281: \left\langle z_{i},z_{j}\right\rangle \right\vert ; \\ 

282: \\ 

283: \left[ \left( \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert

284: ^{\gamma }\right) ^{2}-\sum\limits_{i=1}^{n}\left\vert \alpha

285: _{i}\right\vert ^{2\gamma }\right] ^{\frac{1}{\gamma }}\left(

286: \sum\limits_{1\leq i\neq j\leq n}\left\vert \left\langle

287: z_{i},z_{j}\right\rangle \right\vert ^{\delta }\right) ^{\frac{1}{\delta }}

288: \\ 

289: \hfill \text{where \ }\gamma >1,\ \frac{1}{\gamma }+\frac{1}{\delta }=1; \\ 

290: \\ 

291: \left[ \left( \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert \right)

292: ^{2}-\sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\right]

293: \max\limits_{1\leq i\neq j\leq n}\left\vert \left\langle

294: z_{i},z_{j}\right\rangle \right\vert ;%

295: \end{array}%

296: \right.

297: \end{multline*}%

298: where any term in the first branch can be combined with each term from the

299: second branch giving 9 possible combinations.

300: \end{lemma}

301: 

302: Out of these, we select the following ones that are of relevance for further

303: consideration:%

304: \begin{align}

305: & \left\Vert \sum_{i=1}^{n}\alpha _{i}z_{i}\right\Vert ^{2}

306: \label{upper2.11} \\

307: & \leq \max\limits_{1\leq i\leq n}\left\Vert z_{i}\right\Vert

308: ^{2}\sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}  \notag \\

309: & \qquad \qquad +\max\limits_{1\leq i<j\leq n}\left\vert \left\langle

310: z_{i},z_{j}\right\rangle \right\vert \left[ \left(

311: \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert \right)

312: ^{2}-\sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\right] 

313: \notag \\

314: & \leq \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\left(

315: \max\limits_{1\leq i\leq n}\left\Vert z_{i}\right\Vert ^{2}+\left(

316: n-1\right) \max\limits_{1\leq i<j\leq n}\left\vert \left\langle

317: z_{i},z_{j}\right\rangle \right\vert \right)  \notag

318: \end{align}%

319: and%

320: \begin{align}

321: & \left\Vert \sum_{i=1}^{n}\alpha _{i}z_{i}\right\Vert ^{2}

322: \label{upper2.12} \\

323: & \leq \max\limits_{1\leq i\leq n}\left\Vert z_{i}\right\Vert

324: ^{2}\sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}+\left[

325: \left( \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\right)

326: ^{2}-\sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{4}\right]

327: ^{1/2}  \notag \\

328: & \qquad \qquad \qquad \qquad \qquad \times \left( \sum\limits_{1\leq i\neq

329: j\leq n}\left\vert \left\langle z_{i},z_{j}\right\rangle \right\vert

330: ^{2}\right) ^{\frac{1}{2}}  \notag \\

331: & \leq \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\left[

332: \max\limits_{1\leq i\leq n}\left\Vert z_{i}\right\Vert ^{2}+\left(

333: \sum\limits_{1\leq i\neq j\leq n}\left\vert \left\langle

334: z_{i},z_{j}\right\rangle \right\vert ^{2}\right) ^{\frac{1}{2}}\right] . 

335: \notag

336: \end{align}%

337: Note that the last inequality in (\ref{upper2.11}) follows by the fact that%

338: \begin{equation*}

339: \left( \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert \right)

340: ^{2}\leq n\sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2},

341: \end{equation*}%

342: while the last inequality in (\ref{upper2.12}) is obvious.

343: 

344: Utilising the above inequalities (\ref{upper2.11}) and (\ref{upper2.12})

345: which provide alternatives to the Cauchy-Bunyakovsky-Schwarz inequality (\ref%

346: {upper2.3}), we can state the following results \cite{SILV1xx}.

347: 

348: \begin{theorem}[Dragomir, 2005]

349: \label{uppert2.4}Let $\left\{ x_{1},\dots ,x_{n}\right\} ,$ $M$ and $x$ be

350: as in Theorem \ref{uppert2.1}. Then%

351: \begin{multline}

352: d^{2}\left( x,M\right)   \label{upper2.13} \\

353: \leq \frac{\left\Vert x\right\Vert ^{2}\left[ \max\limits_{1\leq i\leq

354: n}\left\Vert x_{i}\right\Vert ^{2}+\left( \sum\limits_{1\leq i\neq j\leq

355: n}\left\vert \left\langle x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{%

356: \frac{1}{2}}\right] -\sum\limits_{i=1}^{n}\left\vert \left\langle

357: x,x_{i}\right\rangle \right\vert ^{2}}{\max\limits_{1\leq i\leq n}\left\Vert

358: x_{i}\right\Vert ^{2}+\left( \sum\limits_{1\leq i\neq j\leq n}\left\vert

359: \left\langle x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{\frac{1}{2}}}

360: \end{multline}%

361: or, equivalently,%

362: \begin{multline}

363: \Gamma \left( x_{1},\dots ,x_{n},x\right)   \label{upper2.14} \\

364: \leq \frac{\left\Vert x\right\Vert ^{2}\left[ \max\limits_{1\leq i\leq

365: n}\left\Vert x_{i}\right\Vert ^{2}+\left( \sum\limits_{1\leq i\neq j\leq

366: n}\left\vert \left\langle x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{%

367: \frac{1}{2}}\right] -\sum\limits_{i=1}^{n}\left\vert \left\langle

368: x,x_{i}\right\rangle \right\vert ^{2}}{\max\limits_{1\leq i\leq n}\left\Vert

369: x_{i}\right\Vert ^{2}+\left( \sum\limits_{1\leq i\neq j\leq n}\left\vert

370: \left\langle x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{\frac{1}{2}}}

371: \\

372: \times \Gamma \left( x_{1},\dots ,x_{n}\right) .

373: \end{multline}

374: \end{theorem}

375: 

376: \begin{proof}

377: Utilising the inequality (\ref{upper2.12}) for $\alpha _{i}=\left\langle

378: x,x_{i}\right\rangle $ and $z_{i}=x_{i},$ $i\in \left\{ 1,\dots ,n\right\} ,$

379: we can write:%

380: \begin{multline}

381: \left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert

382: ^{2}  \label{upper2.15} \\

383: \leq \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

384: ^{2}\left[ \max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert

385: ^{2}+\left( \sum_{1\leq i\neq j\leq n}\left\vert \left\langle

386: x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{\frac{1}{2}}\right]

387: \end{multline}%

388: for any $x\in H.$

389: 

390: Now, since, by the representation formula (\ref{upper1.3})%

391: \begin{equation}

392: d^{2}\left( x,M\right) =\left\Vert x\right\Vert ^{2}-\frac{%

393: \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{%

394: \left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert

395: ^{2}}\cdot \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

396: \right\vert ^{2},  \label{upper2.16}

397: \end{equation}%

398: for $x\notin M^{\perp },$ hence, by (\ref{upper2.15}) and (\ref{upper2.16})

399: we deduce the desired result (\ref{upper2.13}).

400: \end{proof}

401: 

402: \begin{remark}

403: \label{upperr2.5}In 1941, R.P. Boas \cite{BOxx} and in 1944, R. Bellman \cite%

404: {BExx}, independent of each other, proved the following generalisation of

405: Bessel's inequality:%

406: \begin{equation}

407: \sum_{i=1}^{n}\left\vert \left\langle y,y_{i}\right\rangle \right\vert

408: ^{2}\leq \left\Vert y\right\Vert ^{2}\left[ \max\limits_{1\leq i\leq

409: n}\left\Vert y_{i}\right\Vert ^{2}+\left( \sum_{1\leq i\neq j\leq

410: n}\left\vert \left\langle y_{i},y_{j}\right\rangle \right\vert ^{2}\right) ^{%

411: \frac{1}{2}}\right] ,  \label{upper2.17}

412: \end{equation}%

413: provided $y$ and $y_{i}$ $\left( i\in \left\{ 1,\dots ,n\right\} \right) $

414: are arbitrary vectors in the inner product space $\left( H;\left\langle

415: \cdot ,\cdot \right\rangle \right) .$ If $\left\{ y_{i}\right\} _{i\in

416: \left\{ 1,\dots ,n\right\} }$ are orthonormal, then (\ref{upper2.17})

417: reduces to Bessel's inequality.

418: 

419: In this respect, one may see (\ref{upper2.13}) as a refinement of the

420: Boas-Bellman result (\ref{upper2.17}).

421: \end{remark}

422: 

423: \begin{remark}

424: \label{upperr2.6}On making use of a similar argument to that utilised in

425: Remark \ref{upperr2.2}, one can obtain the following refinement of the

426: Hadamard inequality:%

427: \begin{align}

428: & \Gamma \left( x_{1},\dots ,x_{n}\right)  \label{upper2.18} \\

429: & \leq \left\Vert x_{1}\right\Vert ^{2}  \notag \\

430: & \times \prod_{k=2}^{n}\left( \left\Vert x_{k}\right\Vert ^{2}-\frac{%

431: \sum\limits_{i=1}^{k-1}\left\vert \left\langle x_{k},x_{i}\right\rangle

432: \right\vert ^{2}}{\max\limits_{1\leq i\leq k-1}\left\Vert x_{i}\right\Vert

433: ^{2}+\left( \sum\limits_{1\leq i\neq j\leq k-1}\left\vert \left\langle

434: x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{\frac{1}{2}}}\right) 

435: \notag \\

436: & \leq \prod_{j=1}^{n}\left\Vert x_{j}\right\Vert ^{2}.  \notag

437: \end{align}

438: \end{remark}

439: 

440: Further on, if we choose $\alpha _{i}=\left\langle x,x_{i}\right\rangle ,$ $%

441: z_{i}=x_{i},$ $i\in \left\{ 1,\dots ,n\right\} $ in (\ref{upper2.11}), then

442: we may state the inequality%

443: \begin{multline}

444: \left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert

445: ^{2}  \label{upper2.19} \\

446: \leq \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

447: ^{2}\left( \max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert

448: ^{2}+\left( n-1\right) \max_{1\leq i\neq j\leq n}\left\vert \left\langle

449: x_{i},x_{j}\right\rangle \right\vert \right) .

450: \end{multline}%

451: Utilising (\ref{upper2.19}) and (\ref{upper2.16}) we may state the following

452: result as well \cite{SILV1xx}:

453: 

454: \begin{theorem}[Dragomir, 2005]

455: \label{uppert2.7}Let $\left\{ x_{1},\dots ,x_{n}\right\} ,$ $M$ and $x$ be

456: as in Theorem \ref{uppert2.1}. Then%

457: \begin{multline}

458: d^{2}\left( x,M\right)   \label{upper2.20} \\

459: \leq \frac{\left\Vert x\right\Vert ^{2}\left[ \max\limits_{1\leq i\leq

460: n}\left\Vert x_{i}\right\Vert ^{2}+\left( n-1\right) \max\limits_{1\leq

461: i\neq j\leq n}\left\vert \left\langle x_{i},x_{j}\right\rangle \right\vert %

462: \right] -\sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

463: \right\vert ^{2}}{\max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert

464: ^{2}+\left( n-1\right) \max\limits_{1\leq i\neq j\leq n}\left\vert

465: \left\langle x_{i},x_{j}\right\rangle \right\vert }

466: \end{multline}%

467: or, equivalently,%

468: \begin{multline}

469: \Gamma \left( x_{1},\dots ,x_{n},x\right)   \label{upper2.21} \\

470: \leq \frac{\left\Vert x\right\Vert ^{2}\left[ \max\limits_{1\leq i\leq

471: n}\left\Vert x_{i}\right\Vert ^{2}+\left( n-1\right) \max\limits_{1\leq

472: i\neq j\leq n}\left\vert \left\langle x_{i},x_{j}\right\rangle \right\vert %

473: \right] -\sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

474: \right\vert ^{2}}{\max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert

475: ^{2}+\left( n-1\right) \max\limits_{1\leq i\neq j\leq n}\left\vert

476: \left\langle x_{i},x_{j}\right\rangle \right\vert } \\

477: \times \Gamma \left( x_{1},\dots ,x_{n}\right) .

478: \end{multline}

479: \end{theorem}

480: 

481: \begin{remark}

482: \label{upperr2.8}The above result (\ref{upper2.20}) provides a refinement

483: for the following generalisation of Bessel's inequality:%

484: \begin{equation}

485: \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

486: ^{2}\leq \left\Vert x\right\Vert ^{2}\left[ \max\limits_{1\leq i\leq

487: n}\left\Vert x_{i}\right\Vert ^{2}+\left( n-1\right) \max\limits_{1\leq

488: i\neq j\leq n}\left\vert \left\langle x_{i},x_{j}\right\rangle \right\vert %

489: \right] ,  \label{upper2.22}

490: \end{equation}%

491: obtained by the author in \cite{DRA1xx}.

492: 

493: One can also provide the corresponding refinement of Hadamard's inequality (%

494: \ref{upper1.4}) on using (\ref{upper2.21}), i.e., 

495: \begin{align}

496: & \Gamma \left( x_{1},\dots ,x_{n}\right)  \label{upper2.23} \\

497: & \leq \left\Vert x_{1}\right\Vert ^{2}  \notag \\

498: & \times \prod_{k=2}^{n}\left( \left\Vert x_{k}\right\Vert ^{2}-\frac{%

499: \sum\limits_{i=1}^{k-1}\left\vert \left\langle x_{k},x_{i}\right\rangle

500: \right\vert ^{2}}{\max\limits_{1\leq i\leq k-1}\left\Vert x_{i}\right\Vert

501: ^{2}+\left( k-2\right) \max\limits_{1\leq i\neq j\leq k-1}\left\vert

502: \left\langle x_{i},x_{j}\right\rangle \right\vert }\right)  \notag \\

503: & \leq \prod_{j=1}^{n}\left\Vert x_{j}\right\Vert ^{2}.  \notag

504: \end{align}

505: \end{remark}

506: 

507: \subsection{Other Upper Bounds for $d\left( x,M\right) $}

508: 

509: In \cite[p. 140]{SSDxx} the author obtained the following inequality that is

510: similar to the Cauchy-Bunyakovsky-Schwarz result.

511: 

512: \begin{lemma}[Dragomir, 2004]

513: \label{upperl3.1}Let $z_{1},\dots ,z_{n}\in H$ and $\alpha _{1},\dots

514: ,\alpha _{n}\in \mathbb{K}$. Then one has the inequalities:%

515: \begin{align}

516: \left\Vert \sum_{i=1}^{n}\alpha _{i}z_{i}\right\Vert ^{2}& \leq

517: \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert

518: ^{2}\sum\limits_{j=1}^{n}\left\vert \left\langle z_{i},z_{j}\right\rangle

519: \right\vert  \label{upper3.1} \\

520: & \leq \left\{ 

521: \begin{array}{l}

522: \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert

523: ^{2}\max\limits_{1\leq i\leq n}\left[ \sum\limits_{j=1}^{n}\left\vert

524: \left\langle z_{i},z_{j}\right\rangle \right\vert \right] ; \\ 

525: \\ 

526: \left( \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2p}\right) ^{%

527: \frac{1}{p}}\left( \sum\limits_{i=1}^{n}\left(

528: \sum\limits_{j=1}^{n}\left\vert \left\langle z_{i},z_{j}\right\rangle

529: \right\vert \right) ^{q}\right) ^{\frac{1}{q}} \\ 

530: \hfill \text{where \ }p>1,\ \frac{1}{p}+\frac{1}{q}=1; \\ 

531: \\ 

532: \max\limits_{1\leq i\leq n}\left\vert \alpha _{i}\right\vert

533: ^{2}\sum\limits_{i,j=1}^{n}\left\vert \left\langle z_{i},z_{j}\right\rangle

534: \right\vert .%

535: \end{array}%

536: \right.  \notag

537: \end{align}

538: \end{lemma}

539: 

540: We can state and prove now another upper bound for the distance $d\left(

541: x,M\right) $ as follows \cite{SILV1xx}.

542: 

543: \begin{theorem}[Dragomir, 2005]

544: \label{uppert3.2}Let $\left\{ x_{1},\dots ,x_{n}\right\} ,$ $M$ and $x$ be

545: as in Theorem \ref{uppert2.1}. Then%

546: \begin{equation}

547: d^{2}\left( x,M\right) \leq \frac{\left\Vert x\right\Vert

548: ^{2}\max\limits_{1\leq i\leq n}\left[ \sum\limits_{j=1}^{n}\left\vert

549: \left\langle x_{i},x_{j}\right\rangle \right\vert \right] -\sum%

550: \limits_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

551: ^{2}}{\max\limits_{1\leq i\leq n}\left[ \sum\limits_{j=1}^{n}\left\vert

552: \left\langle x_{i},x_{j}\right\rangle \right\vert \right] }  \label{upper3.2}

553: \end{equation}%

554: or, equivalently,%

555: \begin{multline}

556: \Gamma \left( x_{1},\dots ,x_{n},x\right)  \label{upper3.3} \\

557: \leq \frac{\left\Vert x\right\Vert ^{2}\max\limits_{1\leq i\leq n}\left[

558: \sum\limits_{j=1}^{n}\left\vert \left\langle x_{i},x_{j}\right\rangle

559: \right\vert \right] -\sum\limits_{i=1}^{n}\left\vert \left\langle

560: x,x_{i}\right\rangle \right\vert ^{2}}{\max\limits_{1\leq i\leq n}\left[

561: \sum\limits_{j=1}^{n}\left\vert \left\langle x_{i},x_{j}\right\rangle

562: \right\vert \right] }\cdot \Gamma \left( x_{1},\dots ,x_{n}\right) .

563: \end{multline}

564: \end{theorem}

565: 

566: \begin{proof}

567: Utilising the first branch in (\ref{upper3.1}) we may state that%

568: \begin{equation}

569: \left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert

570: ^{2}\leq \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

571: \right\vert ^{2}\max\limits_{1\leq i\leq n}\left[ \sum\limits_{j=1}^{n}\left%

572: \vert \left\langle x_{i},x_{j}\right\rangle \right\vert \right]

573: \label{upper3.3a}

574: \end{equation}%

575: for any $x\in H.$

576: 

577: Now, since, by the representation formula (\ref{upper1.3}) we have%

578: \begin{equation}

579: d^{2}\left( x,M\right) =\left\Vert x\right\Vert ^{2}-\frac{%

580: \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{%

581: \left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert

582: ^{2}}\cdot \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

583: \right\vert ^{2},  \label{upper3.4}

584: \end{equation}%

585: for $x\notin M^{\perp },$ hence, by (\ref{upper3.3a}) and (\ref{upper3.4})

586: we deduce the desired result (\ref{upper3.2}).

587: \end{proof}

588: 

589: \begin{remark}

590: \label{upperr3.3}In 1971, E. Bombieri \cite{BOMxx} proved the following

591: generalisation of Bessel's inequality, however not stated in the general

592: form for inner products. The general version can be found for instance in 

593: \cite[p. 394]{MPFxx}. It reads as follows: if $y,y_{1},\dots ,y_{n}$ are

594: vectors in the inner product space $\left( H;\left\langle \cdot ,\cdot

595: \right\rangle \right) ,$ then%

596: \begin{equation}

597: \sum_{i=1}^{n}\left\vert \left\langle y,y_{i}\right\rangle \right\vert

598: ^{2}\leq \left\Vert y\right\Vert ^{2}\max\limits_{1\leq i\leq n}\left\{

599: \sum\limits_{j=1}^{n}\left\vert \left\langle y_{i},y_{j}\right\rangle

600: \right\vert \right\} .  \label{upper3.5}

601: \end{equation}%

602: Obviously, when $\left\{ y_{1},\dots ,y_{n}\right\} $ are orthonormal, the

603: inequality (\ref{upper3.5}) produces Bessel's inequality.

604: 

605: In this respect, we may regard our result (\ref{upper3.2}) as a refinement

606: of the Bombieri inequality (\ref{upper3.5}).

607: \end{remark}

608: 

609: \begin{remark}

610: \label{upperr3.4}On making use of a similar argument to that in Remark \ref%

611: {upperr2.2}, we obtain the following refinement for the Hadamard inequality:%

612: \begin{align}

613: \Gamma \left( x_{1},\dots ,x_{n}\right) & \leq \left\Vert x_{1}\right\Vert

614: ^{2}\prod_{k=2}^{n}\left[ \left\Vert x_{k}\right\Vert ^{2}-\frac{%

615: \sum\limits_{i=1}^{k-1}\left\vert \left\langle x_{k},x_{i}\right\rangle

616: \right\vert ^{2}}{\max\limits_{1\leq i\leq k-1}\left[ \sum%

617: \limits_{j=1}^{k-1}\left\vert \left\langle x_{i},x_{j}\right\rangle

618: \right\vert \right] }\right]  \label{upper3.6} \\

619: & \leq \prod_{j=1}^{n}\left\Vert x_{j}\right\Vert ^{2}.  \notag

620: \end{align}

621: \end{remark}

622: 

623: Another different Cauchy-Bunyakovsky-Schwarz type inequality is incorporated

624: in the following lemma \cite{DRA5xx}.

625: 

626: \begin{lemma}[Dragomir, 2004]

627: \label{upperl3.4}Let $z_{1},\dots ,z_{n}\in H$ and $\alpha _{1},\dots

628: ,\alpha _{n}\in \mathbb{K}$. Then 

629: \begin{equation}

630: \left\Vert \sum_{i=1}^{n}\alpha _{i}z_{i}\right\Vert ^{2}\leq \left(

631: \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{p}\right) ^{\frac{2%

632: }{p}}\left( \sum\limits_{i,j=1}^{n}\left\vert \left\langle

633: z_{i},z_{j}\right\rangle \right\vert ^{q}\right) ^{\frac{1}{q}}

634: \label{upper3.7}

635: \end{equation}%

636: for $p>1,$ $\frac{1}{p}+\frac{1}{q}=1.$

637: 

638: If in (\ref{upper3.7}) we choose $p=q=2,$ then we get%

639: \begin{equation}

640: \left\Vert \sum_{i=1}^{n}\alpha _{i}z_{i}\right\Vert ^{2}\leq

641: \sum\limits_{i=1}^{n}\left\vert \alpha _{i}\right\vert ^{2}\left(

642: \sum\limits_{i,j=1}^{n}\left\vert \left\langle z_{i},z_{j}\right\rangle

643: \right\vert ^{2}\right) ^{\frac{1}{2}}.  \label{upper3.8}

644: \end{equation}

645: \end{lemma}

646: 

647: Based on (\ref{upper3.8}), we can state the following result that provides

648: yet another upper bound for the distance $d\left( x,M\right) $ \cite{SILV1xx}%

649: .

650: 

651: \begin{theorem}[Dragomir, 2005]

652: \label{uppert3.5}Let $\left\{ x_{1},\dots ,x_{n}\right\} ,$ $M$ and $x$ be

653: as in Theorem \ref{uppert2.1}. Then%

654: \begin{equation}

655: d^{2}\left( x,M\right) \leq \frac{\left\Vert x\right\Vert ^{2}\left(

656: \sum\limits_{i,j=1}^{n}\left\vert \left\langle x_{i},x_{j}\right\rangle

657: \right\vert ^{2}\right) ^{\frac{1}{2}}-\sum\limits_{i=1}^{n}\left\vert

658: \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{\left(

659: \sum\limits_{i,j=1}^{n}\left\vert \left\langle x_{i},x_{j}\right\rangle

660: \right\vert ^{2}\right) ^{\frac{1}{2}}}  \label{upper3.9}

661: \end{equation}%

662: or, equivalently,%

663: \begin{multline}

664: \Gamma \left( x_{1},\dots ,x_{n},x\right)  \label{upper3.10} \\

665: \leq \frac{\left\Vert x\right\Vert ^{2}\left(

666: \sum\limits_{i,j=1}^{n}\left\vert \left\langle x_{i},x_{j}\right\rangle

667: \right\vert ^{2}\right) ^{\frac{1}{2}}-\sum\limits_{i=1}^{n}\left\vert

668: \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{\left(

669: \sum\limits_{i,j=1}^{n}\left\vert \left\langle x_{i},x_{j}\right\rangle

670: \right\vert ^{2}\right) ^{\frac{1}{2}}}\cdot \Gamma \left( x_{1},\dots

671: ,x_{n}\right) .

672: \end{multline}

673: \end{theorem}

674: 

675: Similar comments apply related to Hadamard's inequality. We omit the details.

676: 

677: \subsection{Some Conditional Bounds}

678: 

679: In the recent paper \cite{DRA4xx}, the author has established the following

680: reverse of the Bessel inequality.

681: 

682: Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ be an inner

683: product space over the real or complex number field $\mathbb{K}$, $\left\{

684: e_{i}\right\} _{i\in I}$ a finite family of orthonormal vectors in $H,$ $%

685: \varphi _{i},\phi _{i}\in \mathbb{K}$, $i\in I$ and $x\in H.$ If%

686: \begin{equation}

687: \func{Re}\left\langle \sum_{i\in I}\phi _{i}e_{i}-x,x-\sum_{i\in I}\varphi

688: _{i}e_{i}\right\rangle \geq 0  \label{upper4.1}

689: \end{equation}%

690: or, equivalently,%

691: \begin{equation}

692: \left\Vert x-\sum_{i\in I}\frac{\varphi _{i}+\phi _{i}}{2}e_{i}\right\Vert

693: \leq \frac{1}{2}\left( \sum_{i\in I}\left\vert \phi _{i}-\varphi

694: _{i}\right\vert ^{2}\right) ^{\frac{1}{2}},  \label{upper4.2}

695: \end{equation}%

696: then%

697: \begin{equation}

698: \left( 0\leq \right) \left\Vert x\right\Vert ^{2}-\sum_{i\in I}\left\vert

699: \left\langle x,e_{i}\right\rangle \right\vert ^{2}\leq \frac{1}{4}\sum_{i\in

700: I}\left\vert \phi _{i}-\varphi _{i}\right\vert ^{2}.  \label{upper4.3}

701: \end{equation}%

702: The constant $\frac{1}{4}$ is best possible in the sense that it cannot be

703: replaced by a smaller constant \cite{SILV1xx}.

704: 

705: \begin{theorem}[Dragomir, 2005]

706: \label{uppert4.1}Let $\left\{ x_{1},\dots x_{n}\right\} $ be a linearly

707: independent system of vectors in $H$ and $M:=span\left\{ x_{1},\dots

708: x_{n}\right\} .$ If $\gamma _{i},$ $\Gamma _{i}\in \mathbb{K}$, $i\in

709: \left\{ 1,\dots ,n\right\} $ and $x\in H\backslash M^{\perp }$ is such that%

710: \begin{equation}

711: \func{Re}\left\langle \sum_{i=1}^{n}\Gamma

712: _{i}x_{i}-x,x-\sum_{i=1}^{n}\gamma _{i}x_{i}\right\rangle \geq 0,

713: \label{upper4.4}

714: \end{equation}%

715: then we have the bound%

716: \begin{equation}

717: d^{2}\left( x,M\right) \leq \frac{1}{4}\left\Vert \sum_{i=1}^{n}\left(

718: \Gamma _{i}-\gamma _{i}\right) x_{i}\right\Vert ^{2}  \label{upper4.5}

719: \end{equation}%

720: or, equivalently,%

721: \begin{equation}

722: \Gamma \left( x_{1},\dots ,x_{n},x\right) \leq \frac{1}{4}\left\Vert

723: \sum_{i=1}^{n}\left( \Gamma _{i}-\gamma _{i}\right) x_{i}\right\Vert

724: ^{2}\Gamma \left( x_{1},\dots ,x_{n}\right) .  \label{upper4.6}

725: \end{equation}

726: \end{theorem}

727: 

728: \begin{proof}

729: It is easy to see that in an inner product space for any $x,z,Z\in H$ one has%

730: \begin{equation*}

731: \left\Vert x-\frac{z+Z}{2}\right\Vert ^{2}-\frac{1}{4}\left\Vert

732: Z-z\right\Vert ^{2}=\func{Re}\left\langle Z-x,x-z\right\rangle ,

733: \end{equation*}%

734: therefore, the condition (\ref{upper4.4}) is actually equivalent to%

735: \begin{equation}

736: \left\Vert x-\sum_{i=1}^{n}\frac{\Gamma _{i}+\gamma _{i}}{2}x_{i}\right\Vert

737: ^{2}\leq \frac{1}{4}\left\Vert \sum_{i=1}^{n}\left( \Gamma _{i}-\gamma

738: _{i}\right) x_{i}\right\Vert ^{2}.  \label{upper4.7}

739: \end{equation}%

740: Now, obviously,%

741: \begin{equation}

742: d^{2}\left( x,M\right) =\inf_{y\in M}\left\Vert x-y\right\Vert ^{2}\leq

743: \left\Vert x-\sum_{i=1}^{n}\frac{\Gamma _{i}+\gamma _{i}}{2}x_{i}\right\Vert

744: ^{2}  \label{upper4.8}

745: \end{equation}%

746: and thus, by (\ref{upper4.7}) and (\ref{upper4.8}) we deduce (\ref{upper4.5}%

747: ).

748: 

749: The last inequality is obvious by the representation (\ref{upper1.2}).

750: \end{proof}

751: 

752: \begin{remark}

753: \label{upperr4.2}Utilising various Cauchy-Bunyakovsky-Schwarz type

754: inequalities we may obtain more convenient (although coarser) bounds for $%

755: d^{2}\left( x,M\right) .$ For instance, if we use the inequality (\ref%

756: {upper2.11}) we can state the inequality:%

757: \begin{multline*}

758: \left\Vert \sum_{i=1}^{n}\left( \Gamma _{i}-\gamma _{i}\right)

759: x_{i}\right\Vert ^{2} \\

760: \leq \sum_{i=1}^{n}\left\vert \Gamma _{i}-\gamma _{i}\right\vert ^{2}\left(

761: \max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert ^{2}+\left(

762: n-1\right) \max\limits_{1\leq i<j\leq n}\left\vert \left\langle

763: x_{i},x_{j}\right\rangle \right\vert \right) ,

764: \end{multline*}%

765: giving the bound:%

766: \begin{multline}

767: d^{2}\left( x,M\right) \leq \frac{1}{4}\sum_{i=1}^{n}\left\vert \Gamma

768: _{i}-\gamma _{i}\right\vert ^{2}  \label{upper4.9} \\

769: \times \left[ \max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert

770: ^{2}+\left( n-1\right) \max\limits_{1\leq i<j\leq n}\left\vert \left\langle

771: x_{i},x_{j}\right\rangle \right\vert \right] ,

772: \end{multline}%

773: provided (\ref{upper4.4}) holds true.

774: 

775: Obviously, if $\left\{ x_{1},\dots ,x_{n}\right\} $ is an orthonormal family

776: in $H,$ then from (\ref{upper4.9}) we deduce the reverse of Bessel's

777: inequality incorporated in (\ref{upper4.3}).

778: 

779: If we use the inequality (\ref{upper2.12}), then we can state the inequality%

780: \begin{multline*}

781: \left\Vert \sum_{i=1}^{n}\left( \Gamma _{i}-\gamma _{i}\right)

782: x_{i}\right\Vert ^{2} \\

783: \leq \sum_{i=1}^{n}\left\vert \Gamma _{i}-\gamma _{i}\right\vert ^{2}\left[

784: \max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert ^{2}+\left(

785: \sum\limits_{1\leq i\neq j\leq n}\left\vert \left\langle

786: x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{\frac{1}{2}}\right] ,

787: \end{multline*}%

788: giving the bound%

789: \begin{multline}

790: d^{2}\left( x,M\right) \leq \frac{1}{4}\sum_{i=1}^{n}\left\vert \Gamma

791: _{i}-\gamma _{i}\right\vert ^{2}  \label{upper4.10} \\

792: \times \left[ \max\limits_{1\leq i\leq n}\left\Vert x_{i}\right\Vert

793: ^{2}+\left( \sum\limits_{1\leq i\neq j\leq n}\left\vert \left\langle

794: x_{i},x_{j}\right\rangle \right\vert ^{2}\right) ^{\frac{1}{2}}\right] ,

795: \end{multline}%

796: provided (\ref{upper4.4}) holds true.

797: 

798: In this case, when one assumes that $\left\{ x_{1},\dots ,x_{n}\right\} $ is

799: an orthonormal family of vectors, then (\ref{upper4.10}) reduces to (\ref%

800: {upper4.3}) as well.

801: 

802: Finally, on utilising the first branch of the inequality (\ref{upper3.1}),

803: we can state that%

804: \begin{equation}

805: d^{2}\left( x,M\right) \leq \frac{1}{4}\sum_{i=1}^{n}\left\vert \Gamma

806: _{i}-\gamma _{i}\right\vert ^{2}\max\limits_{1\leq i\leq n}\left[

807: \sum_{j=1}^{n}\left\vert \left\langle x_{i},x_{j}\right\rangle \right\vert %

808: \right] ,  \label{upper4.11}

809: \end{equation}%

810: provided (\ref{upper4.4}) holds true.

811: 

812: This inequality is also a generalisation of (\ref{upper4.3}).

813: \end{remark}

814: 

815: \section{Reversing the CBS Inequality for Sequences}

816: 

817: \subsection{Introduction}

818: 

819: Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ be an inner

820: product space over the real or complex number field $\mathbb{K}$. One of the

821: most important inequalities in inner product spaces with numerous

822: applications, is the \textit{Schwarz inequality} 

823: \begin{equation}

824: \left\vert \left\langle x,y\right\rangle \right\vert ^{2}\leq \left\Vert

825: x\right\Vert ^{2}\left\Vert y\right\Vert ^{2},\ \ \ \ x,y\in H

826: \label{revcbs1.1}

827: \end{equation}%

828: or, equivalently, 

829: \begin{equation}

830: \left\vert \left\langle x,y\right\rangle \right\vert \leq \left\Vert

831: x\right\Vert \left\Vert y\right\Vert ,\ \ \ \ x,y\in H.  \label{revcbs1.2}

832: \end{equation}%

833: The case of equality holds iff there exists a scalar $\alpha \in \mathbb{K}$

834: such that $x=\alpha y.$

835: 

836: By a \textit{multiplicative reverse} of the Schwarz inequality we understand

837: an inequality of the form%

838: \begin{equation}

839: \left( 1\leq \right) \frac{\left\Vert x\right\Vert \left\Vert y\right\Vert }{%

840: \left\vert \left\langle x,y\right\rangle \right\vert }\leq k_{1}\text{ \ or

841: \ }\left( 1\leq \right) \frac{\left\Vert x\right\Vert ^{2}\left\Vert

842: y\right\Vert ^{2}}{\left\vert \left\langle x,y\right\rangle \right\vert ^{2}}%

843: \leq k_{2}  \label{revcbs1.3}

844: \end{equation}%

845: with appropriate $k_{1}$ and $k_{2}$ and under various assumptions for the

846: vectors $x$ and $y,$ while by an \textit{additive reverse} we understand an

847: inequality of the form%

848: \begin{align}

849: (0& \leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

850: \left\langle x,y\right\rangle \right\vert \leq h_{1}\text{ \ or \ }

851: \label{revcbs1.4} \\

852: (0& \leq )\left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert

853: ^{2}-\left\vert \left\langle x,y\right\rangle \right\vert ^{2}\leq h_{2}. 

854: \notag

855: \end{align}

856: 

857: Similar definition apply when $\left\vert \left\langle x,y\right\rangle

858: \right\vert $ is replaced by $\func{Re}\left\langle x,y\right\rangle $ or $%

859: \left\vert \func{Re}\left\langle x,y\right\rangle \right\vert .$

860: 

861: The following recent reverses for the Schwarz inequality hold (see for

862: instance the monograph on line \cite[p. 20]{SSDxx}).

863: 

864: \begin{theorem}[Dragomir, 2004]

865: \label{revcbst1.1}Let $\left( H;\left\langle \cdot ,\cdot \right\rangle

866: \right) $ be an inner product space over the real or complex number field $%

867: \mathbb{K}$. If $x,y\in H$ and $r>0$ are such that 

868: \begin{equation}

869: \left\Vert x-y\right\Vert \leq r<\left\Vert y\right\Vert ,  \label{revcbs1.5}

870: \end{equation}%

871: then we have the following multiplicative reverse of the Schwarz inequality%

872: \begin{equation}

873: \left( 1\leq \right) \frac{\left\Vert x\right\Vert \left\Vert y\right\Vert }{%

874: \left\vert \left\langle x,y\right\rangle \right\vert }\leq \frac{\left\Vert

875: x\right\Vert \left\Vert y\right\Vert }{\func{Re}\left\langle

876: x,y\right\rangle }\leq \frac{\left\Vert y\right\Vert }{\sqrt{\left\Vert

877: y\right\Vert ^{2}-r^{2}}}  \label{revcbs1.6}

878: \end{equation}%

879: and the subsequent additive reverses%

880: \begin{align}

881: (0& \leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

882: \left\langle x,y\right\rangle \right\vert \leq \left\Vert x\right\Vert

883: \left\Vert y\right\Vert -\func{Re}\left\langle x,y\right\rangle

884: \label{revcbs1.7} \\

885: & \leq \frac{r^{2}}{\sqrt{\left\Vert y\right\Vert ^{2}-r^{2}}\left(

886: \left\Vert y\right\Vert +\sqrt{\left\Vert y\right\Vert ^{2}-r^{2}}\right) }%

887: \func{Re}\left\langle x,y\right\rangle  \notag

888: \end{align}%

889: and%

890: \begin{align}

891: (0& \leq )\left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert

892: ^{2}-\left\vert \left\langle x,y\right\rangle \right\vert ^{2}

893: \label{revcbs1.8} \\

894: & \leq \left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert ^{2}-\left[ \func{%

895: Re}\left\langle x,y\right\rangle \right] ^{2}  \notag \\

896: & \leq r^{2}\left\Vert x\right\Vert ^{2}.  \notag

897: \end{align}%

898: All the above inequalities are sharp.

899: \end{theorem}

900: 

901: Other additive reverses of the quadratic Schwarz's inequality are

902: incorporated in the following result \cite[p. 18-19]{SSDxx}.

903: 

904: \begin{theorem}[Dragomir, 2004]

905: \label{revcbst1.2}Let $x,y\in H$ and $a,A\in \mathbb{K}$. If%

906: \begin{equation}

907: \func{Re}\left\langle Ay-x,x-ay\right\rangle \geq 0  \label{revcbs1.9}

908: \end{equation}%

909: or, equivalently,%

910: \begin{equation}

911: \left\Vert x-\frac{a+A}{2}\cdot y\right\Vert \leq \frac{1}{2}\left\vert

912: A-a\right\vert \left\Vert y\right\Vert ,  \label{revcbs1.10}

913: \end{equation}%

914: then%

915: \begin{align}

916: (0& \leq )\left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert

917: ^{2}-\left\vert \left\langle x,y\right\rangle \right\vert ^{2}

918: \label{revcbs1.11} \\

919: & \leq \frac{1}{4}\left\vert A-a\right\vert ^{2}\left\Vert y\right\Vert

920: ^{4}-\left\{ 

921: \begin{array}{l}

922: \left\vert \frac{A+a}{2}\left\Vert y\right\Vert ^{2}-\left\langle

923: x,y\right\rangle \right\vert ^{2} \\ 

924: \\ 

925: \left\Vert y\right\Vert ^{2}\func{Re}\left\langle Ay-x,x-ay\right\rangle%

926: \end{array}%

927: \right.  \notag \\

928: & \leq \frac{1}{4}\left\vert A-a\right\vert ^{2}\left\Vert y\right\Vert ^{4}.

929: \notag

930: \end{align}%

931: The constant $\frac{1}{4}$ is best possible in all inequalities.

932: \end{theorem}

933: 

934: If one were to assume more about the complex numbers $A$ and $a,$ then one

935: may state the following result as well \cite[p. 21-23]{SSDxx}.

936: 

937: \begin{theorem}[Dragomir, 2004]

938: \label{revcbst1.3}With the assumptions of Theorem \ref{revcbst1.2} and, if

939: in addition, $\func{Re}\left( A\bar{a}\right) >0,$ then%

940: \begin{equation}

941: \left\Vert x\right\Vert \left\Vert y\right\Vert \leq \frac{1}{2}\cdot \frac{%

942: \func{Re}\left[ \left( \bar{A}+\bar{a}\right) \left\langle x,y\right\rangle %

943: \right] }{\sqrt{\func{Re}\left( A\bar{a}\right) }}\leq \frac{1}{2}\cdot 

944: \frac{\left\vert A+a\right\vert }{\sqrt{\func{Re}\left( A\bar{a}\right) }}%

945: \left\vert \left\langle x,y\right\rangle \right\vert ,  \label{revcbs1.12}

946: \end{equation}%

947: \begin{align}

948: (0& \leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}%

949: \left\langle x,y\right\rangle  \label{revcbs1.13} \\

950: & \leq \frac{1}{2}\cdot \frac{\func{Re}\left[ \left( \bar{A}+\bar{a}-2\sqrt{%

951: \func{Re}\left( A\bar{a}\right) }\right) \left\langle x,y\right\rangle %

952: \right] }{\sqrt{\func{Re}\left( A\bar{a}\right) }}  \notag

953: \end{align}%

954: and%

955: \begin{equation}

956: (0\leq )\left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert ^{2}-\left\vert

957: \left\langle x,y\right\rangle \right\vert ^{2}\leq \frac{1}{4}\cdot \frac{%

958: \left\vert A-a\right\vert ^{2}}{\func{Re}\left( A\bar{a}\right) }\left\vert

959: \left\langle x,y\right\rangle \right\vert ^{2}.  \label{revcbs1.14}

960: \end{equation}%

961: The constants $\frac{1}{2}$ and $\frac{1}{4}$ are best possible.

962: \end{theorem}

963: 

964: \begin{remark}

965: \label{revcbsr1.4}If $A=M,$ $a=m$ and $M\geq m>0,$ then (\ref{revcbs1.12})

966: and (\ref{revcbs1.13}) may be written in a more convenient form as%

967: \begin{equation}

968: \left\Vert x\right\Vert \left\Vert y\right\Vert \leq \frac{M+m}{2\sqrt{mM}}%

969: \func{Re}\left\langle x,y\right\rangle  \label{revcbs1.15}

970: \end{equation}%

971: and%

972: \begin{equation}

973: (0\leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}%

974: \left\langle x,y\right\rangle \leq \frac{\left( \sqrt{M}-\sqrt{m}\right) ^{2}%

975: }{2\sqrt{mM}}\func{Re}\left\langle x,y\right\rangle .  \label{revcbs1.16}

976: \end{equation}%

977: Here the constant $\frac{1}{2}$ is sharp in both inequalities.

978: \end{remark}

979: 

980: In this section several reverses for the Cauchy-Bunyakovsky-Schwarz (CBS)

981: inequality for sequences of vectors in Hilbert spaces are obtained.

982: Applications for bounding the distance to a finite-dimensional subspace and

983: in reversing the generalised triangle inequality are also given.

984: 

985: \subsection{Reverses of the $\left( CBS\right) -$Inequality for Two

986: Sequences in $\ell _{\mathbf{p}}^{2}\left( K\right) $}

987: 

988: Let $\left( K,\left\langle \cdot ,\cdot \right\rangle \right) $ be a Hilbert

989: space over $\mathbb{K}$, $p_{i}\geq 0,$ $i\in \mathbb{N}$ with $%

990: \sum_{i=1}^{\infty }p_{i}=1.$ Consider $\ell _{\mathbf{p}}^{2}\left(

991: K\right) $ as the space%

992: \begin{equation*}

993: \ell _{\mathbf{p}}^{2}\left( K\right) :=\left\{ x=\left( x_{i}\right) _{i\in 

994: \mathbb{N}}\left\vert x_{i}\in K,\ i\in \mathbb{N}\text{ \ and \ }%

995: \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}<\infty \right.

996: \right\} .

997: \end{equation*}%

998: It is well known that $\ell _{\mathbf{p}}^{2}\left( K\right) $ endowed with

999: the inner product%

1000: \begin{equation*}

1001: \left\langle x,y\right\rangle _{\mathbf{p}}:=\sum_{i=1}^{\infty

1002: }p_{i}\left\langle x_{i},y_{i}\right\rangle

1003: \end{equation*}%

1004: is a Hilbert space over $\mathbb{K}$. The norm $\left\Vert \cdot \right\Vert

1005: _{\mathbf{p}}$ of $\ell _{\mathbf{p}}^{2}\left( K\right) $ is given by%

1006: \begin{equation*}

1007: \left\Vert x\right\Vert _{\mathbf{p}}:=\left( \sum_{i=1}^{\infty

1008: }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{1}{2}}.

1009: \end{equation*}%

1010: If $x,y\in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ then the following

1011: Cauchy-Bunyakovsky-Schwarz $\left( CBS\right) $ inequality holds true%

1012: \begin{equation}

1013: \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\sum_{i=1}^{\infty

1014: }p_{i}\left\Vert y_{i}\right\Vert ^{2}\geq \left\vert \sum_{i=1}^{\infty

1015: }p_{i}\left\langle x_{i},y_{i}\right\rangle \right\vert ^{2}

1016: \label{revcbs2.1}

1017: \end{equation}%

1018: with equality iff there exists a $\lambda \in \mathbb{K}$ such that $%

1019: x_{i}=\lambda y_{i}$ for each $i\in \mathbb{N}$.

1020: 

1021: This is an obvious consequence of the Schwarz inequality (\ref{revcbs1.1})

1022: written for the inner product $\left\langle \cdot ,\cdot \right\rangle _{%

1023: \mathbf{p}}$ defined on $\ell _{\mathbf{p}}^{2}\left( K\right) .$

1024: 

1025: The following proposition may be stated \cite{DRA3xx}.

1026: 

1027: \begin{proposition}

1028: \label{revcbsp2.1}Let $x,y\in \ell _{\mathbf{p}}^{2}\left( K\right) $ and $%

1029: r>0.$ Assume that%

1030: \begin{equation}

1031: \left\Vert x_{i}-y_{i}\right\Vert \leq r<\left\Vert y_{i}\right\Vert \text{

1032: \ for each \ }i\in \mathbb{N}\text{.}  \label{revcbs2.2}

1033: \end{equation}%

1034: Then we have the inequality%

1035: \begin{align}

1036: (1& \leq )\frac{\left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1037: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

1038: 1}{2}}}{\left\vert \sum_{i=1}^{\infty }p_{i}\left\langle

1039: x_{i},y_{i}\right\rangle \right\vert }  \label{revcbs2.3} \\

1040: & \leq \frac{\left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1041: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

1042: 1}{2}}}{\sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle

1043: x_{i},y_{i}\right\rangle }  \notag \\

1044: & \leq \frac{\left( \sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert

1045: ^{2}\right) ^{\frac{1}{2}}}{\sqrt{\sum_{i=1}^{\infty }p_{i}\left\Vert

1046: y_{i}\right\Vert ^{2}-r^{2}}},  \notag

1047: \end{align}%

1048: \begin{align}

1049: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1050: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

1051: 1}{2}}-\left\vert \sum_{i=1}^{\infty }p_{i}\left\langle

1052: x_{i},y_{i}\right\rangle \right\vert  \label{revcbs2.4} \\

1053: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1054: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

1055: 1}{2}}-\sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle

1056: x_{i},y_{i}\right\rangle  \notag \\

1057: & \leq \frac{r^{2}\cdot \sum\limits_{i=1}^{\infty }p_{i}\func{Re}%

1058: \left\langle x_{i},y_{i}\right\rangle }{\sqrt{\sum\limits_{i=1}^{\infty

1059: }p_{i}\left\Vert y_{i}\right\Vert ^{2}-r^{2}}\left[ \left(

1060: \sum\limits_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{%

1061: \frac{1}{2}}+\sqrt{\sum\limits_{i=1}^{\infty }p_{i}\left\Vert

1062: y_{i}\right\Vert ^{2}-r^{2}}\right] }  \notag

1063: \end{align}%

1064: and%

1065: \begin{align}

1066: (0& \leq )\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1067: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}-\left\vert

1068: \sum_{i=1}^{\infty }p_{i}\left\langle x_{i},y_{i}\right\rangle \right\vert

1069: ^{2}  \label{revcbs2.5} \\

1070: & \leq \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1071: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}-\left[

1072: \sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle x_{i},y_{i}\right\rangle %

1073: \right] ^{2}  \notag \\

1074: & \leq r^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}. 

1075: \notag

1076: \end{align}

1077: \end{proposition}

1078: 

1079: \begin{proof}

1080: From (\ref{revcbs2.2}), we have%

1081: \begin{equation*}

1082: \left\Vert x-y\right\Vert _{\mathbf{p}}^{2}=\sum_{i=1}^{\infty

1083: }p_{i}\left\Vert x_{i}-y_{i}\right\Vert ^{2}\leq r^{2}\sum_{i=1}^{\infty

1084: }p_{i}\leq \sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert

1085: ^{2}=\left\Vert y\right\Vert _{\mathbf{p}}^{2},

1086: \end{equation*}%

1087: giving $\left\Vert x-y\right\Vert _{\mathbf{p}}\leq r\leq \left\Vert

1088: y\right\Vert _{\mathbf{p}}.$ Applying Theorem \ref{revcbst1.1} for $\ell _{%

1089: \mathbf{p}}^{2}\left( K\right) $ and $\left\langle \cdot ,\cdot

1090: \right\rangle _{\mathbf{p}},$ we deduce the desired inequality.

1091: \end{proof}

1092: 

1093: The following proposition holds \cite{DRA3xx}.

1094: 

1095: \begin{proposition}

1096: \label{revcbsp2.2}Let $x,y\in \ell _{\mathbf{p}}^{2}\left( K\right) $ and $%

1097: a,A\in \mathbb{K}$. If%

1098: \begin{equation}

1099: \func{Re}\left\langle Ay_{i}-x_{i},x_{i}-ay_{i}\right\rangle \geq 0\quad 

1100: \text{for each }i\in \mathbb{N}  \label{revcbs2.6}

1101: \end{equation}%

1102: or, equivalently,%

1103: \begin{equation}

1104: \left\Vert x_{i}-\frac{a+A}{2}y_{i}\right\Vert \leq \frac{1}{2}\left\vert

1105: A-a\right\vert \left\Vert y_{i}\right\Vert \quad \text{for each }i\in 

1106: \mathbb{N}  \label{revcbs2.7}

1107: \end{equation}%

1108: then%

1109: \begin{align}

1110: (0& \leq )\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1111: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}-\left\vert

1112: \sum_{i=1}^{\infty }p_{i}\left\langle x_{i},y_{i}\right\rangle \right\vert

1113: ^{2}  \label{revcbs2.8} \\

1114: & \leq \frac{1}{4}\left\vert A-a\right\vert ^{2}\left( \sum_{i=1}^{\infty

1115: }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{2}  \notag \\

1116: & \quad -\left\{ 

1117: \begin{array}{c}

1118: \left\vert \frac{A+a}{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert

1119: ^{2}-\sum_{i=1}^{\infty }p_{i}\left\langle x_{i},y_{i}\right\rangle

1120: \right\vert ^{2} \\ 

1121: \\ 

1122: \sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\sum_{i=1}^{\infty

1123: }p_{i}\func{Re}\left\langle Ay_{i}-x_{i},x_{i}-ay_{i}\right\rangle%

1124: \end{array}%

1125: \right.  \notag \\

1126: & \leq \frac{1}{4}\left\vert A-a\right\vert ^{2}\left( \sum_{i=1}^{\infty

1127: }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{2}.  \notag

1128: \end{align}

1129: \end{proposition}

1130: 

1131: The proof follows by Theorem \ref{revcbst1.2}, we omit the details.

1132: 

1133: Finally, on using Theorem \ref{revcbst1.3}, we may state \cite{DRA3xx}:

1134: 

1135: \begin{proposition}

1136: \label{revcbsp2.3}Assume that $x,y,a$ and $A$ are as in Proposition \ref%

1137: {revcbsp2.2}. Moreover, if $\func{Re}\left( A\bar{a}\right) >0,$ then we

1138: have the inequality:%

1139: \begin{align}

1140: & \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1141: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

1142: 1}{2}}  \label{revcbs2.9} \\

1143: & \leq \frac{1}{2}\cdot \frac{\func{Re}\left[ \left( \bar{A}+\bar{a}\right)

1144: \sum_{i=1}^{\infty }p_{i}\left\langle x_{i},y_{i}\right\rangle \right] }{%

1145: \sqrt{\func{Re}\left( A\bar{a}\right) }}  \notag \\

1146: & \leq \frac{1}{2}\cdot \frac{\left\vert A-a\right\vert }{\sqrt{\func{Re}%

1147: \left( A\bar{a}\right) }}\left\vert \sum_{i=1}^{\infty }p_{i}\left\langle

1148: x_{i},y_{i}\right\rangle \right\vert ,  \notag

1149: \end{align}%

1150: \begin{align}

1151: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1152: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

1153: 1}{2}}-\sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle

1154: x_{i},y_{i}\right\rangle  \label{revcbs2.10} \\

1155: & \leq \frac{1}{2}\cdot \frac{\func{Re}\left[ \left( \bar{A}+\bar{a}-2\sqrt{%

1156: \func{Re}\left( A\bar{a}\right) }\right) \sum_{i=1}^{\infty

1157: }p_{i}\left\langle x_{i},y_{i}\right\rangle \right] }{\sqrt{\func{Re}\left( A%

1158: \bar{a}\right) }}  \notag

1159: \end{align}%

1160: and%

1161: \begin{align}

1162: (0& \leq )\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

1163: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}-\left\vert

1164: \sum_{i=1}^{\infty }p_{i}\left\langle x_{i},y_{i}\right\rangle \right\vert

1165: ^{2}  \label{revcbs2.11} \\

1166: & \leq \frac{1}{4}\cdot \frac{\left\vert A-a\right\vert ^{2}}{\func{Re}%

1167: \left( A\bar{a}\right) }\left\vert \sum_{i=1}^{\infty }p_{i}\left\langle

1168: x_{i},y_{i}\right\rangle \right\vert ^{2}.  \notag

1169: \end{align}

1170: \end{proposition}

1171: 

1172: \subsection{Reverses of the $\left( CBS\right) -$Inequality for Mixed

1173: Sequences}

1174: 

1175: Let $\left( K,\left\langle \cdot ,\cdot \right\rangle \right) $ be a Hilbert

1176: space over $\mathbb{K}$ and for $p_{i}\geq 0,$ $i\in \mathbb{N}$ with $%

1177: \sum_{i=1}^{\infty }p_{i}=1,$ and $\ell _{\mathbf{p}}^{2}\left( K\right) $

1178: the Hilbert space defined in the previous section.

1179: 

1180: If%

1181: \begin{equation*}

1182: \alpha \in \ell _{\mathbf{p}}^{2}\left( \mathbb{K}\right) :=\left\{ \alpha

1183: =\left( \alpha _{i}\right) _{i\in \mathbb{N}}\left\vert \alpha _{i}\in 

1184: \mathbb{K},\ i\in \mathbb{N}\text{ \ and \ }\sum_{i=1}^{\infty

1185: }p_{i}\left\vert \alpha _{i}\right\vert ^{2}<\infty \right. \right\}

1186: \end{equation*}%

1187: and $x\in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ then the following

1188: Cauchy-Bunyakovsky-Schwarz $\left( CBS\right) $ inequality holds true:%

1189: \begin{equation}

1190: \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1191: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\geq \left\Vert

1192: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2},  \label{revcbs3.1}

1193: \end{equation}%

1194: with equality if and only if there exists a vector $v\in K$ such that $x_{i}=%

1195: \overline{\alpha _{i}}v$ for any $i\in \mathbb{N}$.

1196: 

1197: The inequality (\ref{revcbs3.1}) follows by the obvious identity%

1198: \begin{multline*}

1199: \qquad \sum_{i=1}^{n}p_{i}\left\vert \alpha _{i}\right\vert

1200: ^{2}\sum_{i=1}^{n}p_{i}\left\Vert x_{i}\right\Vert ^{2}-\left\Vert

1201: \sum_{i=1}^{n}p_{i}\alpha _{i}x_{i}\right\Vert ^{2} \\

1202: =\frac{1}{2}\sum_{i=1}^{n}\sum_{j=1}^{n}p_{i}p_{j}\left\Vert \overline{%

1203: \alpha _{i}}x_{j}-\overline{\alpha _{j}}x_{i}\right\Vert ^{2},\qquad

1204: \end{multline*}%

1205: for any $n\in \mathbb{N}$, $n\geq 1.$

1206: 

1207: In the following we establish some reverses of the $\left( CBS\right) -$%

1208: inequality in some of its various equivalent forms that will be specified

1209: where they occur \cite{DRA3xx}.

1210: 

1211: \begin{theorem}[Dragomir, 2005]

1212: \label{revcbst3.1}Let $\alpha \in \ell _{\mathbf{p}}^{2}\left( \mathbb{K}%

1213: \right) ,$ $x\in \ell _{\mathbf{p}}^{2}\left( K\right) $ and $a\in K,$ $r>0$

1214: such that $\left\Vert a\right\Vert >r.$ If the following condition holds%

1215: \begin{equation}

1216: \left\Vert x_{i}-\overline{\alpha _{i}}a\right\Vert \leq r\left\vert \alpha

1217: _{i}\right\vert \quad \text{for each }i\in \mathbb{N},  \label{revcbs3.2}

1218: \end{equation}%

1219: (note that if $\alpha _{i}\neq 0$ for any $i\in \mathbb{N}$, then the

1220: condition (\ref{revcbs3.2}) is equivalent to%

1221: \begin{equation}

1222: \left\Vert \frac{x_{i}}{\overline{\alpha _{i}}}-a\right\Vert \leq r\quad 

1223: \text{for each }i\in \mathbb{N}),  \label{revcbs3.3}

1224: \end{equation}%

1225: then we have the following inequalities 

1226: \begin{align}

1227: \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1228: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1229: 1}{2}}& \leq \frac{1}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}}\func{Re}%

1230: \left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},a\right\rangle

1231: \label{revcbs3.4} \\

1232: & \leq \frac{\left\Vert a\right\Vert }{\sqrt{\left\Vert a\right\Vert

1233: ^{2}-r^{2}}}\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ;

1234: \notag

1235: \end{align}%

1236: \begin{align}

1237: 0& \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1238: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1239: 1}{2}}-\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert

1240: \label{revcbs3.5} \\

1241: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1242: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1243: 1}{2}}-\func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},\frac{%

1244: a}{\left\Vert a\right\Vert }\right\rangle  \notag

1245: \end{align}%

1246: \begin{align}

1247: & \leq \frac{r^{2}}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\left(

1248: \left\Vert a\right\Vert +\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\right) }%

1249: \func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},\frac{a}{%

1250: \left\Vert a\right\Vert }\right\rangle  \notag \\

1251: & \leq \frac{r^{2}}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\left(

1252: \left\Vert a\right\Vert +\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\right) }%

1253: \left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ;  \notag

1254: \end{align}%

1255: \begin{align}

1256: \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1257: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}& \leq \frac{1}{%

1258: \left\Vert a\right\Vert ^{2}-r^{2}}\left[ \func{Re}\left\langle

1259: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},a\right\rangle \right] ^{2}

1260: \label{revcbs3.6} \\

1261: & \leq \frac{\left\Vert a\right\Vert ^{2}}{\left\Vert a\right\Vert ^{2}-r^{2}%

1262: }\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \notag

1263: \end{align}%

1264: and%

1265: \begin{align}

1266: 0& \leq \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1267: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}-\left\Vert

1268: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \label{revcbs3.7}

1269: \\

1270: & \leq \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1271: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}-\left[ \func{Re%

1272: }\left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},\frac{a}{\left\Vert

1273: a\right\Vert }\right\rangle \right] ^{2}  \notag \\

1274: & \leq \frac{r^{2}}{\left\Vert a\right\Vert ^{2}\left( \left\Vert

1275: a\right\Vert ^{2}-r^{2}\right) }\left[ \func{Re}\left\langle

1276: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},a\right\rangle \right] ^{2}  \notag

1277: \\

1278: & \leq \frac{r^{2}}{\left\Vert a\right\Vert ^{2}-r^{2}}\left\Vert

1279: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}.  \notag

1280: \end{align}%

1281: All the inequalities in (\ref{revcbs3.4}) -- (\ref{revcbs3.7}) are sharp.

1282: \end{theorem}

1283: 

1284: \begin{proof}

1285: From (\ref{revcbs3.2}) we deduce%

1286: \begin{equation*}

1287: \left\Vert x_{i}\right\Vert ^{2}-2\func{Re}\left\langle x_{i},\overline{%

1288: \alpha _{i}}a\right\rangle +\left\vert \alpha _{i}\right\vert ^{2}\left\Vert

1289: a\right\Vert ^{2}\leq \left\vert \alpha _{i}\right\vert ^{2}r^{2}

1290: \end{equation*}%

1291: for any $i\in \mathbb{N}$, which is clearly equivalent to%

1292: \begin{equation}

1293: \left\Vert x_{i}\right\Vert ^{2}+\left( \left\Vert a\right\Vert

1294: ^{2}-r^{2}\right) \left\vert \alpha _{i}\right\vert ^{2}\leq 2\func{Re}%

1295: \left\langle \alpha _{i}x_{i},a\right\rangle  \label{revcbs3.8}

1296: \end{equation}%

1297: for each $i\in \mathbb{N}$.

1298: 

1299: If we multiply (\ref{revcbs3.8}) by $p_{i}\geq 0$ and sum over $i\in \mathbb{%

1300: N}$, then we deduce%

1301: \begin{equation}

1302: \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}+\left( \left\Vert

1303: a\right\Vert ^{2}-r^{2}\right) \sum_{i=1}^{\infty }p_{i}\left\vert \alpha

1304: _{i}\right\vert ^{2}\leq 2\func{Re}\left\langle \sum_{i=1}^{\infty

1305: }p_{i}\alpha _{i}x_{i},a\right\rangle .  \label{revcbs3.9}

1306: \end{equation}%

1307: Now, dividing (\ref{revcbs3.9}) by $\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}%

1308: >0$ we get%

1309: \begin{multline}

1310: \frac{1}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}}\sum_{i=1}^{\infty

1311: }p_{i}\left\Vert x_{i}\right\Vert ^{2}+\sqrt{\left\Vert a\right\Vert

1312: ^{2}-r^{2}}\sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert ^{2}

1313: \label{revcbs3.10} \\

1314: \leq \frac{2}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}}\func{Re}%

1315: \left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},a\right\rangle .

1316: \end{multline}%

1317: On the other hand, by the elementary inequality%

1318: \begin{equation*}

1319: \frac{1}{\alpha }p+\alpha q\geq 2\sqrt{pq},\qquad \alpha >0,\ p,q\geq 0,

1320: \end{equation*}%

1321: we can state that:%

1322: \begin{multline}

1323: 2\left[ \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1324: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right] ^{\frac{%

1325: 1}{2}}  \label{revcbs3.11} \\

1326: \leq \frac{1}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}}\sum_{i=1}^{\infty

1327: }p_{i}\left\Vert x_{i}\right\Vert ^{2}+\sqrt{\left\Vert a\right\Vert

1328: ^{2}-r^{2}}\sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert ^{2}.

1329: \end{multline}%

1330: Making use of (\ref{revcbs3.10}) and (\ref{revcbs3.11}), we deduce the first

1331: part of (\ref{revcbs3.4}).

1332: 

1333: The second part is obvious by Schwarz's inequality%

1334: \begin{equation*}

1335: \func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

1336: _{i}x_{i},a\right\rangle \leq \left\Vert \sum_{i=1}^{\infty }p_{i}\alpha

1337: _{i}x_{i}\right\Vert \left\Vert a\right\Vert .

1338: \end{equation*}

1339: 

1340: If $p_{1}=1,$ $x_{1}=x,$ $\alpha _{1}=1$ and $p_{i}=0,$ $\alpha _{i}=0,$ $%

1341: x_{i}=0$ for $i\geq 2,$ then from (\ref{revcbs3.4}) we deduce the inequality%

1342: \begin{equation*}

1343: \left\Vert x\right\Vert \leq \frac{1}{\sqrt{\left\Vert a\right\Vert

1344: ^{2}-r^{2}}}\func{Re}\left\langle x,a\right\rangle \leq \frac{\left\Vert

1345: x\right\Vert \left\Vert a\right\Vert }{\sqrt{\left\Vert a\right\Vert

1346: ^{2}-r^{2}}}

1347: \end{equation*}%

1348: provided $\left\Vert x-a\right\Vert \leq r<\left\Vert a\right\Vert ,$ $%

1349: x,a\in K.$ The sharpness of this inequality has been shown in \cite[p. 20]%

1350: {SSDxx}, and we omit the details.

1351: 

1352: The other inequalities are obvious consequences of (\ref{revcbs3.4}) and we

1353: omit the details.

1354: \end{proof}

1355: 

1356: The following corollary may be stated \cite{DRA3xx}.

1357: 

1358: \begin{corollary}

1359: \label{revcbsc3.2}Let $\alpha \in \ell _{\mathbf{p}}^{2}\left( \mathbb{K}%

1360: \right) ,$ $x\in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ $e\in H,$ $%

1361: \left\Vert e\right\Vert =1$ and $\varphi ,\phi \in \mathbb{K}$ with $\func{Re%

1362: }\left( \phi \bar{\varphi}\right) >0.$ If%

1363: \begin{equation}

1364: \left\Vert x_{i}-\overline{\alpha _{i}}\cdot \frac{\varphi +\phi }{2}\cdot

1365: e\right\Vert \leq \frac{1}{2}\left\vert \phi -\varphi \right\vert \left\vert

1366: \alpha _{i}\right\vert  \label{revcbs3.12}

1367: \end{equation}%

1368: for each $i\in \mathbb{N}$, or, equivalently%

1369: \begin{equation}

1370: \func{Re}\left\langle \phi \overline{\alpha _{i}}e-x_{i},x_{i}-\varphi 

1371: \overline{\alpha _{i}}e\right\rangle \geq 0  \label{revcbs3.13}

1372: \end{equation}%

1373: for each $i\in \mathbb{N}$, (note that, if $\alpha _{i}\neq 0$ for any $i\in 

1374: \mathbb{N}$, then (\ref{revcbs3.12}) is equivalent to%

1375: \begin{equation}

1376: \left\Vert \frac{x_{i}}{\overline{\alpha _{i}}}-\frac{\varphi +\phi }{2}%

1377: \cdot e\right\Vert \leq \frac{1}{2}\left\vert \phi -\varphi \right\vert

1378: \label{revcbs3.14}

1379: \end{equation}%

1380: for each $i\in \mathbb{N}$ and (\ref{revcbs3.13}) is equivalent to%

1381: \begin{equation*}

1382: \func{Re}\left\langle \phi e-\frac{x_{i}}{\overline{\alpha _{i}}},\frac{x_{i}%

1383: }{\overline{\alpha _{i}}}-\varphi e\right\rangle \geq 0

1384: \end{equation*}%

1385: for each $i\in \mathbb{N}$), then the following reverses of the $\left(

1386: CBS\right) -$inequality are valid:%

1387: \begin{align}

1388: \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1389: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1390: 1}{2}}& \leq \frac{\func{Re}\left[ \left( \bar{\phi}+\bar{\varphi}\right)

1391: \left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},e\right\rangle \right]

1392: }{2\left[ \func{Re}\left( \phi \overline{\varphi }\right) \right] ^{\frac{1}{%

1393: 2}}}  \label{revcbs3.15} \\

1394: & \leq \frac{1}{2}\cdot \frac{\left\vert \varphi +\phi \right\vert }{\left[ 

1395: \func{Re}\left( \phi \overline{\varphi }\right) \right] ^{\frac{1}{2}}}%

1396: \left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ;  \notag

1397: \end{align}%

1398: \begin{align}

1399: 0& \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1400: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1401: 1}{2}}-\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert

1402: \label{revcbs3.16} \\

1403: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1404: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1405: 1}{2}}  \notag \\

1406: & \qquad -\func{Re}\left[ \frac{\bar{\phi}+\bar{\varphi}}{\left\vert \varphi

1407: +\phi \right\vert }\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

1408: _{i}x_{i},e\right\rangle \right]  \notag \\

1409: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{2\sqrt{\func{Re}%

1410: \left( \phi \varphi \right) }\left( \left\vert \varphi +\phi \right\vert +2%

1411: \sqrt{\func{Re}\left( \phi \overline{\varphi }\right) }\right) }  \notag \\

1412: & \qquad \times \func{Re}\left[ \frac{\bar{\phi}+\bar{\varphi}}{\left\vert

1413: \varphi +\phi \right\vert }\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

1414: _{i}x_{i},e\right\rangle \right]  \notag \\

1415: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{2\sqrt{\func{Re}%

1416: \left( \phi \varphi \right) }\left( \left\vert \varphi +\phi \right\vert +2%

1417: \sqrt{\func{Re}\left( \phi \overline{\varphi }\right) }\right) }\left\Vert

1418: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ;  \notag

1419: \end{align}%

1420: \begin{align}

1421: & \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1422: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}

1423: \label{revcbs3.17} \\

1424: & \leq \frac{1}{4\func{Re}\left( \phi \bar{\varphi}\right) }\left[ \func{Re}%

1425: \left\{ \left( \bar{\phi}+\bar{\varphi}\right) \left\langle

1426: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},e\right\rangle \right\} \right]

1427: ^{2}  \notag \\

1428: & \leq \frac{1}{4}\cdot \frac{\left\vert \varphi +\phi \right\vert ^{2}}{%

1429: \func{Re}\left( \phi \bar{\varphi}\right) }\left\Vert \sum_{i=1}^{\infty

1430: }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \notag

1431: \end{align}%

1432: and%

1433: \begin{align}

1434: 0& \leq \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1435: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}-\left\Vert

1436: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \label{revcbs3.18}

1437: \\

1438: & \leq \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1439: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}  \notag \\

1440: & \qquad -\left[ \func{Re}\left\{ \frac{\bar{\phi}+\bar{\varphi}}{\left\vert

1441: \varphi +\phi \right\vert }\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

1442: _{i}x_{i},e\right\rangle \right\} \right] ^{2}  \notag \\

1443: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{4\left\vert \phi

1444: +\varphi \right\vert ^{2}\func{Re}\left( \phi \bar{\varphi}\right) }\left\{ 

1445: \func{Re}\left[ \left( \bar{\phi}+\bar{\varphi}\right) \left\langle

1446: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},e\right\rangle \right] \right\}

1447: ^{2}  \notag \\

1448: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{4\func{Re}\left(

1449: \phi \bar{\varphi}\right) }\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha

1450: _{i}x_{i}\right\Vert ^{2}.  \notag

1451: \end{align}%

1452: All the inequalities in (\ref{revcbs3.15}) -- (\ref{revcbs3.18}) are sharp.

1453: \end{corollary}

1454: 

1455: \begin{remark}

1456: \label{revcbsr2.3}We remark that if $M\geq m>0$ and for $\alpha \in \ell _{%

1457: \mathbf{p}}^{2}\left( \mathbb{K}\right) ,$ $x\in \ell _{\mathbf{p}%

1458: }^{2}\left( K\right) ,$ $e\in H$ with $\left\Vert e\right\Vert =1,$ one

1459: would assume that either%

1460: \begin{equation}

1461: \left\Vert \frac{x_{i}}{\overline{\alpha _{i}}}-\frac{M+m}{2}\cdot

1462: e\right\Vert \leq \frac{1}{2}\left( M-m\right)  \label{revcbs3.19}

1463: \end{equation}%

1464: for each $i\in \mathbb{N}$, or, equivalently%

1465: \begin{equation}

1466: \func{Re}\left\langle Me-\frac{x_{i}}{\overline{\alpha _{i}}},\frac{x_{i}}{%

1467: \overline{\alpha _{i}}}-me\right\rangle \geq 0  \label{revcbs3.20}

1468: \end{equation}%

1469: for each $i\in \mathbb{N}$, then the following, much simpler reverses of the 

1470: $\left( CBS\right) -$ inequality may be stated:%

1471: \begin{align}

1472: \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1473: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1474: 1}{2}}& \leq \frac{M+m}{2\sqrt{mM}}\func{Re}\left\langle \sum_{i=1}^{\infty

1475: }p_{i}\alpha _{i}x_{i},e\right\rangle  \label{revcbs3.21} \\

1476: & \leq \frac{M+m}{2\sqrt{mM}}\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha

1477: _{i}x_{i}\right\Vert ;  \notag

1478: \end{align}%

1479: \begin{align}

1480: 0& \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1481: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1482: 1}{2}}-\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert

1483: \label{revcbs3.22} \\

1484: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1485: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

1486: 1}{2}}-\func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

1487: _{i}x_{i},e\right\rangle  \notag \\

1488: & \leq \frac{\left( M-m\right) ^{2}}{2\left( \sqrt{M}+\sqrt{m}\right) ^{2}%

1489: \sqrt{mM}}\func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

1490: _{i}x_{i},e\right\rangle  \notag \\

1491: & \leq \frac{\left( M-m\right) ^{2}}{2\left( \sqrt{M}+\sqrt{m}\right) ^{2}%

1492: \sqrt{mM}}\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ; 

1493: \notag

1494: \end{align}%

1495: \begin{align}

1496: & \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1497: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}-\left\Vert

1498: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \label{revcbs3.23}

1499: \\

1500: & \leq \frac{\left( M+m\right) ^{2}}{4mM}\left[ \func{Re}\left\langle

1501: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},e\right\rangle \right] ^{2}  \notag

1502: \\

1503: & \leq \frac{\left( M+m\right) ^{2}}{4mM}\left\Vert \sum_{i=1}^{\infty

1504: }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \notag

1505: \end{align}%

1506: and%

1507: \begin{equation}

1508: 0\leq \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1509: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}-\left\Vert

1510: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \label{revcbs3.24}

1511: \end{equation}%

1512: \begin{align*}

1513: & \leq \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

1514: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}-\left[ \func{Re%

1515: }\left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},e\right\rangle %

1516: \right] ^{2} \\

1517: & \leq \frac{\left( M-m\right) ^{2}}{4mM}\left[ \func{Re}\left\langle

1518: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},e\right\rangle \right] ^{2} \\

1519: & \leq \frac{\left( M-m\right) ^{2}}{4mM}\left\Vert \sum_{i=1}^{\infty

1520: }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}.

1521: \end{align*}

1522: \end{remark}

1523: 

1524: \subsection{Reverses for the Generalised Triangle Inequality}

1525: 

1526: In 1966, J.B. Diaz and F.T. Metcalf \cite{DMxx} proved the following reverse

1527: of the generalised triangle inequality holding in an inner product space $%

1528: \left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ over the real or

1529: complex number field $\mathbb{K}$:%

1530: \begin{equation}

1531: r\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert \leq \left\Vert

1532: \sum_{i=1}^{n}x_{i}\right\Vert  \label{revcbs4.1}

1533: \end{equation}%

1534: provided the vectors $x_{1},\dots ,x_{n}\in H\backslash \left\{ 0\right\} $

1535: satisfy the assumption%

1536: \begin{equation}

1537: 0\leq r\leq \frac{\func{Re}\left\langle x_{i},a\right\rangle }{\left\Vert

1538: x_{i}\right\Vert },  \label{revcbs4.2}

1539: \end{equation}%

1540: where $a\in H$ and $\left\Vert a\right\Vert =1.$

1541: 

1542: In an attempt to diversify the assumptions for which such reverse results

1543: hold, the author pointed out in \cite{DRA2xx} that%

1544: \begin{equation}

1545: \sqrt{1-\rho ^{2}}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert \leq \left\Vert

1546: \sum_{i=1}^{n}x_{i}\right\Vert ,  \label{revcbs4.3}

1547: \end{equation}%

1548: where the vectors $x_{i,}i\in \left\{ 1,\dots ,n\right\} $ satisfy the

1549: condition%

1550: \begin{equation}

1551: \left\Vert x_{i}-a\right\Vert \leq \rho ,\qquad i\in \left\{ 1,\dots

1552: ,n\right\}  \label{revcbs4.4}

1553: \end{equation}%

1554: where $a\in H,$ $\left\Vert a\right\Vert =1$ and $\rho \in \left( 0,1\right)

1555: .$

1556: 

1557: If, for $M\geq m>0,$ the vectors $x_{i}\in H,$ $i\in \left\{ 1,\dots

1558: ,n\right\} $ verify either%

1559: \begin{equation}

1560: \func{Re}\left\langle Ma-x_{i},x_{i}-ma\right\rangle \geq 0,\qquad i\in

1561: \left\{ 1,\dots ,n\right\} ,  \label{revcbs4.5}

1562: \end{equation}%

1563: or, equivalently,%

1564: \begin{equation}

1565: \left\Vert x_{i}-\frac{M+m}{2}\cdot a\right\Vert \leq \frac{1}{2}\left(

1566: M-m\right) ,\qquad i\in \left\{ 1,\dots ,n\right\} ,  \label{revcbs4.6}

1567: \end{equation}%

1568: where $a\in H,$ $\left\Vert a\right\Vert =1$, then the following reverse of

1569: the generalised triangle inequality may be stated as well \cite{DRA2xx}%

1570: \begin{equation}

1571: \frac{2\sqrt{mM}}{M+m}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert \leq

1572: \left\Vert \sum_{i=1}^{n}x_{i}\right\Vert .  \label{revcbs4.7}

1573: \end{equation}

1574: 

1575: Note that the inequalities (\ref{revcbs4.1}), (\ref{revcbs4.3}), and (\ref%

1576: {revcbs4.7}) are sharp; necessary and sufficient equality conditions were

1577: provided (see \cite{DMxx} and \cite{DRA2xx}).

1578: 

1579: It is obvious, from Theorem \ref{revcbst3.1}, that, if%

1580: \begin{equation}

1581: \left\Vert x_{i}-a\right\Vert \leq r,\quad \text{for }\quad i\in \left\{

1582: 1,\dots ,n\right\} ,  \label{revcbs4.8}

1583: \end{equation}%

1584: where $\left\Vert a\right\Vert >r,$ $a\in H$ and $x_{i}\in H$, $i\in \left\{

1585: 1,\dots ,n\right\} ,$ then one can state the inequalities%

1586: \begin{align}

1587: \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert & \leq \sqrt{n}\left(

1588: \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{1}{2}}

1589: \label{revcbs4.9} \\

1590: & \leq \frac{1}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}}\func{Re}%

1591: \left\langle \sum_{i=1}^{n}x_{i},a\right\rangle  \notag \\

1592: & \leq \frac{\left\Vert a\right\Vert }{\sqrt{\left\Vert a\right\Vert

1593: ^{2}-r^{2}}}\left\Vert \sum_{i=1}^{n}x_{i}\right\Vert  \notag

1594: \end{align}%

1595: and%

1596: \begin{align}

1597: 0& \leq \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert -\left\Vert

1598: \sum_{i=1}^{n}x_{i}\right\Vert  \label{revcbs4.10} \\

1599: & \leq \sqrt{n}\left( \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}\right)

1600: ^{\frac{1}{2}}-\left\Vert \sum_{i=1}^{n}x_{i}\right\Vert  \notag \\

1601: & \leq \sqrt{n}\left( \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}\right)

1602: ^{\frac{1}{2}}-\func{Re}\left\langle \sum_{i=1}^{n}x_{i},\frac{a}{\left\Vert

1603: a\right\Vert }\right\rangle  \notag \\

1604: & \leq \frac{r^{2}}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\left(

1605: \left\Vert a\right\Vert +\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\right) }%

1606: \func{Re}\left\langle \sum_{i=1}^{n}x_{i},\frac{a}{\left\Vert a\right\Vert }%

1607: \right\rangle  \notag \\

1608: & \leq \frac{r^{2}}{\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\left(

1609: \left\Vert a\right\Vert +\sqrt{\left\Vert a\right\Vert ^{2}-r^{2}}\right) }%

1610: \left\Vert \sum_{i=1}^{n}x_{i}\right\Vert .  \notag

1611: \end{align}%

1612: We note that for $\left\Vert a\right\Vert =1$ and $r\in \left( 0,1\right) ,$

1613: the inequality (\ref{revcbs3.9}) becomes%

1614: \begin{equation}

1615: \sqrt{1-r^{2}}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert \leq \sqrt{\left(

1616: 1-r^{2}\right) n}\left( \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

1617: ^{2}\right) ^{\frac{1}{2}}  \label{revcbs4.11}

1618: \end{equation}%

1619: \begin{equation*}

1620: \leq \func{Re}\left\langle \sum_{i=1}^{n}x_{i},a\right\rangle \leq

1621: \left\Vert \sum_{i=1}^{n}x_{i}\right\Vert

1622: \end{equation*}%

1623: which is a refinement of (\ref{revcbs4.3}).

1624: 

1625: With the same assumptions for $a$ and $r,$ we have from (\ref{revcbs4.10})

1626: the following additive reverse of the generalised triangle inequality:%

1627: \begin{align}

1628: 0& \leq \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert -\left\Vert

1629: \sum_{i=1}^{n}x_{i}\right\Vert  \label{revcbs4.12} \\

1630: & \leq \frac{r^{2}}{\sqrt{1-r^{2}}\left( 1+\sqrt{1-r^{2}}\right) }\func{Re}%

1631: \left\langle \sum_{i=1}^{n}x_{i},a\right\rangle  \notag \\

1632: & \leq \frac{r^{2}}{\sqrt{1-r^{2}}\left( 1+\sqrt{1-r^{2}}\right) }\left\Vert

1633: \sum_{i=1}^{n}x_{i}\right\Vert .  \notag

1634: \end{align}

1635: 

1636: We can obtain the following reverses of the generalised triangle inequality

1637: from Corollary \ref{revcbsc3.2} when the assumptions are in terms of complex

1638: numbers $\phi $ and $\varphi :$

1639: 

1640: If $\varphi ,\phi \in \mathbb{K}$ with $\func{Re}\left( \phi \bar{\varphi}%

1641: \right) >0$ and $x_{i}\in H,$ $i\in \left\{ 1,\dots ,n\right\} ,$ $e\in H,$ $%

1642: \left\Vert e\right\Vert =1$ are such that%

1643: \begin{equation}

1644: \left\Vert x_{i}-\frac{\varphi +\phi }{2}e\right\Vert \leq \frac{1}{2}%

1645: \left\vert \phi -\varphi \right\vert \text{ \ for each \ }i\in \left\{

1646: 1,\dots ,n\right\} ,  \label{revcbs4.13}

1647: \end{equation}%

1648: or, equivalently,%

1649: \begin{equation*}

1650: \func{Re}\left\langle \phi e-x_{i},x_{i}-\varphi e\right\rangle \geq 0\text{

1651: \ for each \ }i\in \left\{ 1,\dots ,n\right\} ,

1652: \end{equation*}%

1653: then we have the following reverses of the generalised triangle inequality:%

1654: \begin{align}

1655: \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert & \leq \sqrt{n}\left(

1656: \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{1}{2}}

1657: \label{revcbs4.14} \\

1658: & \leq \frac{\func{Re}\left[ \left( \bar{\phi}+\bar{\varphi}\right)

1659: \left\langle \sum_{i=1}^{n}x_{i},e\right\rangle \right] }{2\sqrt{\func{Re}%

1660: \left( \phi \bar{\varphi}\right) }}  \notag \\

1661: & \leq \frac{1}{2}\cdot \frac{\left\vert \bar{\phi}+\bar{\varphi}\right\vert 

1662: }{\sqrt{\func{Re}\left( \phi \bar{\varphi}\right) }}\left\Vert

1663: \sum_{i=1}^{n}x_{i}\right\Vert  \notag

1664: \end{align}%

1665: and%

1666: \begin{align}

1667: 0& \leq \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert -\left\Vert

1668: \sum_{i=1}^{n}x_{i}\right\Vert  \label{revcbs4.15} \\

1669: & \leq \sqrt{n}\left( \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}\right)

1670: ^{\frac{1}{2}}-\left\Vert \sum_{i=1}^{n}x_{i}\right\Vert  \notag \\

1671: & \leq \sqrt{n}\left( \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}\right)

1672: ^{\frac{1}{2}}-\func{Re}\left[ \frac{\left\vert \bar{\phi}+\bar{\varphi}%

1673: \right\vert }{\sqrt{\func{Re}\left( \bar{\phi}\bar{\varphi}\right) }}%

1674: \left\langle \sum_{i=1}^{n}x_{i},e\right\rangle \right]  \notag \\

1675: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{2\sqrt{\func{Re}%

1676: \left( \phi \bar{\varphi}\right) }\left( \left\vert \phi +\varphi

1677: \right\vert +2\sqrt{\func{Re}\left( \phi \bar{\varphi}\right) }\right) } 

1678: \notag \\

1679: & \qquad \times \func{Re}\left[ \frac{\bar{\phi}+\bar{\varphi}}{\left\vert 

1680: \bar{\phi}+\bar{\varphi}\right\vert }\left\langle

1681: \sum_{i=1}^{n}x_{i},e\right\rangle \right]  \notag \\

1682: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{2\sqrt{\func{Re}%

1683: \left( \phi \bar{\varphi}\right) }\left( \left\vert \phi +\varphi

1684: \right\vert +2\sqrt{\func{Re}\left( \phi \bar{\varphi}\right) }\right) }%

1685: \left\Vert \sum_{i=1}^{n}x_{i}\right\Vert .  \notag

1686: \end{align}

1687: 

1688: Obviously (\ref{revcbs4.14}) for $\phi =M,$ $\varphi =m,$ $M\geq m>0$

1689: provides a refinement for (\ref{revcbs4.7}).

1690: 

1691: \subsection{Lower Bounds for the Distance to Finite-Dimensional Subspaces}

1692: 

1693: Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ be an inner

1694: product space over the real or complex number field $\mathbb{K}$, $\left\{

1695: y_{1},\dots ,y_{n}\right\} $ a subset of $H$ and $G\left( y_{1},\dots

1696: ,y_{n}\right) $ the \textit{Gram matrix} of $\left\{ y_{1},\dots

1697: ,y_{n}\right\} $ where $\left( i,j\right) -$entry is $\left\langle

1698: y_{i},y_{j}\right\rangle .$ The determinant of $G\left( y_{1},\dots

1699: ,y_{n}\right) $ is called the \textit{Gram determinant} of $\left\{

1700: y_{1},\dots ,y_{n}\right\} $ and is denoted by $\Gamma \left( y_{1},\dots

1701: ,y_{n}\right) .$

1702: 

1703: Following \cite[p. 129 -- 133]{DExx}, we state here some general results for

1704: the Gram determinant that will be used in the sequel:

1705: 

1706: \begin{enumerate}

1707: \item Let $\left\{ x_{1},\dots ,x_{n}\right\} \subset H.$ Then $\Gamma

1708: \left( x_{1},\dots ,x_{n}\right) \neq 0$ if and only if $\left\{ x_{1},\dots

1709: ,x_{n}\right\} $ is linearly independent;

1710: 

1711: \item Let $M=span\left\{ x_{1},\dots ,x_{n}\right\} $ be $n-$dimensional in $%

1712: H,$ i.e., $\{x_{1},\dots ,$ $x_{n}\}$ is linearly independent. Then for each 

1713: $x\in H,$ the distance $d\left( x,M\right) $ from $x$ to the linear subspace 

1714: $H$ has the representations%

1715: \begin{equation}

1716: d^{2}\left( x,M\right) =\frac{\Gamma \left( x_{1},\dots ,x_{n},x\right) }{%

1717: \Gamma \left( x_{1},\dots ,x_{n}\right) }  \label{revcbs5.2}

1718: \end{equation}%

1719: and%

1720: \begin{equation}

1721: d^{2}\left( x,M\right) =\left\{ 

1722: \begin{array}{ll}

1723: \left\Vert x\right\Vert ^{2}-\frac{\left( \sum_{i=1}^{n}\left\vert

1724: \left\langle x,x_{i}\right\rangle \right\vert ^{2}\right) ^{2}}{\left\Vert

1725: \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert ^{2}} & 

1726: \text{if \ }x\notin M^{\perp }, \\ 

1727: &  \\ 

1728: \left\Vert x\right\Vert ^{2} & \text{if \ }x\in M^{\perp },%

1729: \end{array}%

1730: \right.  \label{revcbs5.3}

1731: \end{equation}%

1732: where $M^{\perp }$ denotes the orthogonal complement of $M.$

1733: \end{enumerate}

1734: 

1735: The following result may be stated \cite{DRA3xx}.

1736: 

1737: \begin{proposition}

1738: \label{revcbsp5.1}Let $\left\{ x_{1},\dots ,x_{n}\right\} $ be a system of

1739: linearly independent vectors, $M=span\left\{ x_{1},\dots ,x_{n}\right\} ,$ $%

1740: x\in H\backslash M^{\perp },$ $a\in H,$ $r>0$ and $\left\Vert a\right\Vert

1741: >r.$ If%

1742: \begin{equation}

1743: \left\Vert x_{i}-\overline{\left\langle x,x_{i}\right\rangle }a\right\Vert

1744: \leq \left\vert \left\langle x,x_{i}\right\rangle \right\vert r\text{ \ for

1745: each \ }i\in \left\{ 1,\dots ,n\right\} ,  \label{revcbs5.6}

1746: \end{equation}%

1747: (note that if $\left\langle x,x_{i}\right\rangle \neq 0$ for each $i\in

1748: \left\{ 1,\dots ,n\right\} ,$ then (\ref{revcbs5.6}) can be written as%

1749: \begin{equation}

1750: \left\Vert \frac{x_{i}}{\overline{\left\langle x,x_{i}\right\rangle }}%

1751: -a\right\Vert \leq r\text{ \ for each \ }i\in \left\{ 1,\dots ,n\right\} ),

1752: \label{revcbs5.7}

1753: \end{equation}%

1754: then we have the inequality%

1755: \begin{align}

1756: d^{2}\left( x,M\right) & \geq \left\Vert x\right\Vert ^{2}-\frac{\left\Vert

1757: a\right\Vert ^{2}}{\left\Vert a\right\Vert ^{2}-r^{2}}\cdot \frac{%

1758: \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{%

1759: \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}}  \label{revcbs5.8} \\

1760: & \geq 0.  \notag

1761: \end{align}

1762: \end{proposition}

1763: 

1764: \begin{proof}

1765: Utilising (\ref{revcbs5.3}) we can state that%

1766: \begin{equation}

1767: d^{2}\left( x,M\right) =\left\Vert x\right\Vert ^{2}-\frac{%

1768: \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{%

1769: \left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle x_{i}\right\Vert

1770: ^{2}}\cdot \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

1771: \right\vert ^{2}.  \label{revcbs5.9}

1772: \end{equation}%

1773: Also, by the inequality (\ref{revcbs3.6}) applied for $\alpha

1774: _{i}=\left\langle x,x_{i}\right\rangle ,$ $p_{i}=\frac{1}{n},$ $i\in \left\{

1775: 1,\dots ,n\right\} ,$ we can state that%

1776: \begin{equation}

1777: \frac{\sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle \right\vert

1778: ^{2}}{\left\Vert \sum_{i=1}^{n}\left\langle x,x_{i}\right\rangle

1779: x_{i}\right\Vert ^{2}}\leq \frac{\left\Vert a\right\Vert ^{2}}{\left\Vert

1780: a\right\Vert ^{2}-r^{2}}\cdot \frac{1}{\sum_{i=1}^{n}\left\Vert

1781: x_{i}\right\Vert ^{2}}  \label{revcbs5.10}

1782: \end{equation}%

1783: provided the condition (\ref{revcbs5.7}) holds true.

1784: 

1785: Combining (\ref{revcbs5.9}) with (\ref{revcbs5.10}) we deduce the first

1786: inequality in (\ref{revcbs5.8}).

1787: 

1788: The last inequality is obvious since, by Schwarz's inequality%

1789: \begin{equation*}

1790: \left\Vert x\right\Vert ^{2}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

1791: ^{2}\geq \sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

1792: \right\vert ^{2}\geq \frac{\left\Vert a\right\Vert ^{2}}{\left\Vert

1793: a\right\Vert ^{2}-r^{2}}\sum_{i=1}^{n}\left\vert \left\langle

1794: x,x_{i}\right\rangle \right\vert ^{2}.

1795: \end{equation*}

1796: \end{proof}

1797: 

1798: \begin{remark}

1799: Utilising (\ref{revcbs5.2}), we can state the following result for Gram

1800: determinants%

1801: \begin{multline}

1802: \Gamma \left( x_{1},\dots ,x_{n},x\right)  \label{revcbs5.11} \\

1803: \geq \left[ \left\Vert x\right\Vert ^{2}-\frac{\left\Vert a\right\Vert ^{2}}{%

1804: \left\Vert a\right\Vert ^{2}-r^{2}}\cdot \frac{\sum_{i=1}^{n}\left\vert

1805: \left\langle x,x_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{n}\left\Vert

1806: x_{i}\right\Vert ^{2}}\right] \Gamma \left( x_{1},\dots ,x_{n}\right) \geq 0

1807: \end{multline}%

1808: for $x\notin M^{\perp }$ and $x,x_{i},a$ and $r$ are as in Proposition \ref%

1809: {revcbsp5.1}.

1810: \end{remark}

1811: 

1812: The following corollary of Proposition \ref{revcbsp5.1} may be stated as

1813: well \cite{DRA3xx}.

1814: 

1815: \begin{corollary}

1816: \label{revcbsc5.3}Let $\left\{ x_{1},\dots ,x_{n}\right\} $ be a system of

1817: linearly independent vectors, $M=span\left\{ x_{1},\dots ,x_{n}\right\} ,$ $%

1818: x\in H\backslash M^{\perp }$ \ and $\phi ,\varphi \in K$ with $\func{Re}%

1819: \left( \phi \bar{\varphi}\right) >0.$ If $e\in H,$ $\left\Vert e\right\Vert

1820: =1$ and%

1821: \begin{equation}

1822: \left\Vert x_{i}-\overline{\left\langle x,x_{i}\right\rangle }\cdot \frac{%

1823: \varphi +\phi }{2}e\right\Vert \leq \frac{1}{2}\left\vert \phi -\varphi

1824: \right\vert \left\vert \left\langle x,x_{i}\right\rangle \right\vert

1825: \label{revcbs5.12}

1826: \end{equation}%

1827: or, equivalently,%

1828: \begin{equation*}

1829: \func{Re}\left\langle \phi \overline{\cdot \left\langle x,x_{i}\right\rangle 

1830: }e-x_{i},x_{i}-\varphi \cdot \overline{\left\langle x,x_{i}\right\rangle }%

1831: e\right\rangle \geq 0,

1832: \end{equation*}%

1833: for each $i\in \left\{ 1,\dots ,n\right\} ,$ then%

1834: \begin{equation}

1835: d^{2}\left( x,M\right) \geq \left\Vert x\right\Vert ^{2}-\frac{1}{4}\cdot 

1836: \frac{\left\vert \varphi +\phi \right\vert ^{2}}{\func{Re}\left( \phi \bar{%

1837: \varphi}\right) }\cdot \frac{\sum_{i=1}^{n}\left\vert \left\langle

1838: x,x_{i}\right\rangle \right\vert ^{2}}{\sum_{i=1}^{n}\left\Vert

1839: x_{i}\right\Vert ^{2}}\geq 0,  \label{revcbs5.13}

1840: \end{equation}%

1841: or, equivalently,%

1842: \begin{multline}

1843: \Gamma \left( x_{1},\dots ,x_{n},x\right)  \label{revcbs5.14} \\

1844: \geq \left[ \left\Vert x\right\Vert ^{2}-\frac{1}{4}\cdot \frac{\left\vert

1845: \varphi +\phi \right\vert ^{2}}{\func{Re}\left( \phi \bar{\varphi}\right) }%

1846: \cdot \frac{\sum_{i=1}^{n}\left\vert \left\langle x,x_{i}\right\rangle

1847: \right\vert ^{2}}{\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}}\right]

1848: \Gamma \left( x_{1},\dots ,x_{n}\right) \geq 0.

1849: \end{multline}

1850: \end{corollary}

1851: 

1852: \subsection{Applications for Fourier Coefficients}

1853: 

1854: Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ be a Hilbert

1855: space over the real or complex number field $\mathbb{K}$ and $\left\{

1856: e_{i}\right\} _{i\in I}$ an \textit{orthornormal basis} for $H.$ Then (see

1857: for instance \cite[p. 54 -- 61]{DExx})

1858: 

1859: \begin{enumerate}

1860: \item[(i)] Every element $x\in H$ can be expanded in a \textit{Fourier

1861: series, }i.e.,%

1862: \begin{equation*}

1863: x=\sum_{i\in I}\left\langle x,e_{i}\right\rangle e_{i},

1864: \end{equation*}%

1865: where $\left\langle x,e_{i}\right\rangle ,$ $i\in I$ are the \textit{Fourier

1866: coefficients} of $x;$

1867: 

1868: \item[(ii)] (Parseval identity)%

1869: \begin{equation*}

1870: \left\Vert x\right\Vert ^{2}=\sum_{i\in I}\left\langle x,e_{i}\right\rangle

1871: e_{i},\qquad x\in H;

1872: \end{equation*}

1873: 

1874: \item[(iii)] (Extended Parseval identity)%

1875: \begin{equation*}

1876: \left\langle x,y\right\rangle =\sum_{i\in I}\left\langle

1877: x,e_{i}\right\rangle \left\langle e_{i},y\right\rangle ,\qquad x,y\in H;

1878: \end{equation*}

1879: 

1880: \item[(iv)] (Elements are uniquely determined by their Fourier coefficients)%

1881: \begin{equation*}

1882: \left\langle x,e_{i}\right\rangle =\left\langle y,e_{i}\right\rangle \text{

1883: \ for every }i\in I\text{ \ implies that }x=y.

1884: \end{equation*}

1885: \end{enumerate}

1886: 

1887: Now, we must remark that all the results can be stated for $K=\mathbb{K}$

1888: where $\mathbb{K}$ is the Hilbert space of complex (real) numbers endowed

1889: with the usual norm and inner product.

1890: 

1891: Therefore, we can state the following proposition \cite{DRA3xx}.

1892: 

1893: \begin{proposition}

1894: \label{revcbsp6.1}Let $\left( H;\left\langle \cdot ,\cdot \right\rangle

1895: \right) $ be a Hilbert space over $\mathbb{K}$ and $\left\{ e_{i}\right\}

1896: _{i\in I}$ an orthornormal base for $H.$ If $x,y\in H$ $\left( y\neq

1897: 0\right) ,$ $a\in \mathbb{K}$ $\left( \mathbb{C},\mathbb{R}\right) $ and $%

1898: r>0 $ such that $\left\vert a\right\vert >r$ and%

1899: \begin{equation}

1900: \left\vert \frac{\left\langle x,e_{i}\right\rangle }{\left\langle

1901: y,e_{i}\right\rangle }-a\right\vert \leq r\text{ \ for each \ }i\in I,

1902: \label{revcbs6.1}

1903: \end{equation}%

1904: then we have the following reverse of the Schwarz inequality%

1905: \begin{align}

1906: \left\Vert x\right\Vert \left\Vert y\right\Vert & \leq \frac{1}{\sqrt{%

1907: \left\vert a\right\vert ^{2}-r^{2}}}\func{Re}\left[ \bar{a}\cdot

1908: \left\langle x,y\right\rangle \right]  \label{revcbs6.2} \\

1909: & \leq \frac{\left\vert a\right\vert }{\sqrt{\left\vert a\right\vert

1910: ^{2}-r^{2}}}\left\vert \left\langle x,y\right\rangle \right\vert ;  \notag

1911: \end{align}%

1912: \begin{align}

1913: (0& \leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

1914: \left\langle x,y\right\rangle \right\vert  \label{revcbs6.3} \\

1915: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}\left[ 

1916: \frac{\bar{a}}{\left\vert a\right\vert }\cdot \left\langle x,y\right\rangle %

1917: \right]  \notag \\

1918: & \leq \frac{r^{2}}{\sqrt{\left\vert a\right\vert ^{2}-r^{2}}\left(

1919: \left\vert a\right\vert +\sqrt{\left\vert a\right\vert ^{2}-r^{2}}\right) }%

1920: \func{Re}\left[ \frac{\bar{a}}{\left\vert a\right\vert }\cdot \left\langle

1921: x,y\right\rangle \right]  \notag \\

1922: & \leq \frac{r^{2}}{\sqrt{\left\vert a\right\vert ^{2}-r^{2}}\left(

1923: \left\vert a\right\vert +\sqrt{\left\vert a\right\vert ^{2}-r^{2}}\right) }%

1924: \left\vert \left\langle x,y\right\rangle \right\vert ;  \notag

1925: \end{align}%

1926: \begin{align}

1927: \left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert ^{2}& \leq \frac{1}{%

1928: \left\vert a\right\vert ^{2}-r^{2}}\left( \func{Re}\left[ \bar{a}\cdot

1929: \left\langle x,y\right\rangle \right] \right) ^{2}  \label{revcbs6.4} \\

1930: & \leq \frac{\left\vert a\right\vert ^{2}}{\left\vert a\right\vert ^{2}-r^{2}%

1931: }\left\vert \left\langle x,y\right\rangle \right\vert ^{2}  \notag

1932: \end{align}%

1933: and%

1934: \begin{align}

1935: (0& \leq )\left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert

1936: ^{2}-\left\vert \left\langle x,y\right\rangle \right\vert ^{2}

1937: \label{revcbs6.5} \\

1938: & \leq \left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert ^{2}-\left( \func{%

1939: Re}\left[ \frac{\bar{a}}{\left\vert a\right\vert }\cdot \left\langle

1940: x,y\right\rangle \right] \right) ^{2}  \notag \\

1941: & \leq \frac{r^{2}}{\left\vert a\right\vert ^{2}\left( \left\vert

1942: a\right\vert ^{2}-r^{2}\right) }-\left( \func{Re}\left[ \frac{\bar{a}}{%

1943: \left\vert a\right\vert }\cdot \left\langle x,y\right\rangle \right] \right)

1944: ^{2}  \notag \\

1945: & \leq \frac{r^{2}}{\left\vert a\right\vert ^{2}-r^{2}}\left\vert

1946: \left\langle x,y\right\rangle \right\vert .  \notag

1947: \end{align}

1948: \end{proposition}

1949: 

1950: The proof is similar to the one in Theorem \ref{revcbst3.1}, when instead of 

1951: $x_{i}$ we take $\left\langle x,e_{i}\right\rangle ,$ instead of $\alpha

1952: _{i} $ we take $\left\langle e_{i},y\right\rangle ,$ $\left\Vert \cdot

1953: \right\Vert =\left\vert \cdot \right\vert ,$ $p_{i}=1,$ and we use the

1954: Parseval identities mentioned above in (ii) and (iii). We omit the details.

1955: 

1956: The following result may be stated as well \cite{DRA3xx}.

1957: 

1958: \begin{proposition}

1959: \label{revcbsp6.2}Let $\left( H;\left\langle \cdot ,\cdot \right\rangle

1960: \right) $ be a Hilbert space over $\mathbb{K}$ and $\left\{ e_{i}\right\}

1961: _{i\in I}$ an orthornormal base for $H.$ If $x,y\in H$ $\left( y\neq

1962: 0\right) ,$ $e,\varphi ,\phi \in \mathbb{K}$ with $\func{Re}\left( \phi \bar{%

1963: \varphi}\right) >0,$ $\left\vert e\right\vert =1$ and, either%

1964: \begin{equation}

1965: \left\vert \frac{\left\langle x,e_{i}\right\rangle }{\left\langle

1966: y,e_{i}\right\rangle }-\frac{\varphi +\phi }{2}\cdot e\right\vert \leq \frac{%

1967: 1}{2}\left\vert \phi -\varphi \right\vert  \label{revcbs6.6}

1968: \end{equation}%

1969: or, equivalently,%

1970: \begin{equation}

1971: \func{Re}\left[ \left( \phi e-\frac{\left\langle x,e_{i}\right\rangle }{%

1972: \left\langle y,e_{i}\right\rangle }\right) \left( \frac{\left\langle

1973: e_{i},x\right\rangle }{\left\langle e_{i},y\right\rangle }-\bar{\varphi}\bar{%

1974: e}\right) \right] \geq 0  \label{revcbs6.7}

1975: \end{equation}%

1976: for each $i\in I,$ then the following reverses of the Schwarz inequality

1977: hold:%

1978: \begin{equation}

1979: \left\Vert x\right\Vert \left\Vert y\right\Vert \leq \frac{\func{Re}\left[

1980: \left( \bar{\phi}+\bar{\varphi}\right) \bar{e}\left\langle x,y\right\rangle %

1981: \right] }{2\sqrt{\func{Re}\left( \phi \bar{\varphi}\right) }}\leq \frac{1}{2}%

1982: \cdot \frac{\left\vert \varphi +\phi \right\vert }{\sqrt{\func{Re}\left(

1983: \phi \bar{\varphi}\right) }}\left\vert \left\langle x,y\right\rangle

1984: \right\vert ,  \label{revcbs6.8}

1985: \end{equation}%

1986: \begin{align}

1987: (0& \leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

1988: \left\langle x,y\right\rangle \right\vert  \label{revcbs6.9} \\

1989: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}\left[ 

1990: \frac{\left( \bar{\phi}+\bar{\varphi}\right) \bar{e}}{\left\vert \varphi

1991: +\phi \right\vert }\left\langle x,y\right\rangle \right] \displaybreak 

1992: \notag \\

1993: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{2\sqrt{\func{Re}%

1994: \left( \phi \bar{\varphi}\right) }\left( \left\vert \varphi +\phi

1995: \right\vert +2\sqrt{\func{Re}\left( \phi \bar{\varphi}\right) }\right) } 

1996: \notag \\

1997: & \qquad \times \func{Re}\left[ \frac{\left( \bar{\phi}+\bar{\varphi}\right) 

1998: \bar{e}}{\left\vert \varphi +\phi \right\vert }\left\langle x,y\right\rangle %

1999: \right]  \notag \\

2000: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{2\sqrt{\func{Re}%

2001: \left( \phi \bar{\varphi}\right) }\left( \left\vert \varphi +\phi

2002: \right\vert +2\sqrt{\func{Re}\left( \phi \bar{\varphi}\right) }\right) }%

2003: \left\vert \left\langle x,y\right\rangle \right\vert  \notag

2004: \end{align}%

2005: and%

2006: \begin{align}

2007: (0& \leq )\left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert

2008: ^{2}-\left\vert \left\langle x,y\right\rangle \right\vert ^{2}

2009: \label{revcbs6.10} \\

2010: & \leq \left\Vert x\right\Vert ^{2}\left\Vert y\right\Vert ^{2}-\left\{ 

2011: \func{Re}\left[ \frac{\left( \bar{\phi}+\bar{\varphi}\right) \bar{e}}{%

2012: \left\vert \varphi +\phi \right\vert }\left\langle x,y\right\rangle \right]

2013: \right\} ^{2}  \notag \\

2014: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{4\left\vert \phi

2015: +\varphi \right\vert ^{2}\func{Re}\left( \phi \bar{\varphi}\right) }\left\{ 

2016: \func{Re}\left[ \left( \bar{\phi}+\bar{\varphi}\right) \bar{e}\left\langle

2017: x,y\right\rangle \right] \right\} ^{2}  \notag \\

2018: & \leq \frac{\left\vert \phi -\varphi \right\vert ^{2}}{4\func{Re}\left(

2019: \phi \bar{\varphi}\right) }\left\vert \left\langle x,y\right\rangle

2020: \right\vert ^{2}.  \notag

2021: \end{align}

2022: \end{proposition}

2023: 

2024: \begin{remark}

2025: \label{revcbsr5.3}If $\phi =M\geq m=\varphi >0,$ then one may state simpler

2026: inequalities from (\ref{revcbs6.8}) -- (\ref{revcbs6.10}). We omit the

2027: details.

2028: \end{remark}

2029: 

2030: \section{Other Reverses of the CBS Inequality}

2031: 

2032: \subsection{Introduction}

2033: 

2034: Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ be an inner

2035: product space over the real or complex number field $\mathbb{K}$.

2036: 

2037: The following reverses for the Schwarz inequality hold (see \cite{SSDR1}, or

2038: the monograph on line \cite[p. 27]{SSDxx}).

2039: 

2040: \begin{theorem}[Dragomir, 2004]

2041: \label{revt1}Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $

2042: be an inner product space over the real or complex number field $\mathbb{K}$%

2043: . If $x,a\in H$ and $r>0$ are such that%

2044: \begin{equation}

2045: x\in B\left( x,r\right) :=\left\{ z\in H|\left\Vert z-a\right\Vert \leq

2046: r\right\} ,  \label{rev1.5}

2047: \end{equation}%

2048: then we have the inequalities%

2049: \begin{align}

2050: \left( 0\leq \right) \left\Vert x\right\Vert \left\Vert a\right\Vert

2051: -\left\vert \left\langle x,a\right\rangle \right\vert & \leq \left\Vert

2052: x\right\Vert \left\Vert a\right\Vert -\left\vert \func{Re}\left\langle

2053: x,a\right\rangle \right\vert  \label{rev1.6} \\

2054: & \leq \left\Vert x\right\Vert \left\Vert a\right\Vert -\func{Re}%

2055: \left\langle x,a\right\rangle \leq \frac{1}{2}r^{2}.  \notag

2056: \end{align}%

2057: The constant $\frac{1}{2}$ is best possible in (\ref{rev1.5}) in the sense

2058: that it cannot be replaced by a smaller quantity.

2059: \end{theorem}

2060: 

2061: An additive version for the Schwarz inequality that may be more useful in

2062: applications is incorporated in \cite{SSDR1} (see also \cite[p. 28]{SSDxx}).

2063: 

2064: \begin{theorem}[Dragomir, 2004]

2065: \label{revt2}Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $

2066: be an inner product space over $\mathbb{K}$ and $x,y\in H$ and $\gamma

2067: ,\Gamma \in \mathbb{K}$ with $\Gamma \neq -\gamma $ and either%

2068: \begin{equation}

2069: \func{Re}\left\langle \Gamma y-x,x-\gamma y\right\rangle \geq 0,

2070: \label{rev1.7}

2071: \end{equation}%

2072: or, equivalently,%

2073: \begin{equation}

2074: \left\Vert x-\frac{\gamma +\Gamma }{2}y\right\Vert \leq \frac{1}{2}%

2075: \left\vert \Gamma -\gamma \right\vert \left\Vert y\right\Vert  \label{rev1.8}

2076: \end{equation}%

2077: holds. Then we have the inequalities%

2078: \begin{align}

2079: 0& \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

2080: \left\langle x,y\right\rangle \right\vert  \label{rev1.9} \\

2081: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert \func{Re}%

2082: \left[ \frac{\bar{\Gamma}+\bar{\gamma}}{\left\vert \Gamma +\gamma

2083: \right\vert }\cdot \left\langle x,y\right\rangle \right] \right\vert  \notag

2084: \\

2085: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}\left[ 

2086: \frac{\bar{\Gamma}+\bar{\gamma}}{\left\vert \Gamma +\gamma \right\vert }%

2087: \cdot \left\langle x,y\right\rangle \right]  \notag \\

2088: & \leq \frac{1}{4}\cdot \frac{\left\vert \Gamma -\gamma \right\vert ^{2}}{%

2089: \left\vert \Gamma +\gamma \right\vert }\left\Vert y\right\Vert ^{2}.  \notag

2090: \end{align}%

2091: The constant $\frac{1}{4}$ in the last inequality is best possible.

2092: \end{theorem}

2093: 

2094: We remark that a simpler version of the above result may be stated if one

2095: assumed that the scalars are real:

2096: 

2097: \begin{corollary}

2098: \label{revc1}If $M\geq m>0,$ and either%

2099: \begin{equation}

2100: \func{Re}\left\langle My-x,x-my\right\rangle \geq 0,  \label{rev1.10}

2101: \end{equation}%

2102: or, equivalently,%

2103: \begin{equation}

2104: \left\Vert x-\frac{m+M}{2}y\right\Vert \leq \frac{1}{2}\left( M-m\right)

2105: \left\Vert y\right\Vert  \label{rev1.11}

2106: \end{equation}%

2107: holds, then%

2108: \begin{align}

2109: 0& \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

2110: \left\langle x,y\right\rangle \right\vert  \label{rev1.12} \\

2111: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert \func{Re}%

2112: \left\langle x,y\right\rangle \right\vert  \notag \\

2113: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}%

2114: \left\langle x,y\right\rangle  \notag \\

2115: & \leq \frac{1}{4}\cdot \frac{\left( M-m\right) ^{2}}{M+m}\left\Vert

2116: y\right\Vert ^{2}.  \notag

2117: \end{align}%

2118: The constant $\frac{1}{4}$ is sharp.

2119: \end{corollary}

2120: 

2121: Now, let $\left( K,\left\langle \cdot ,\cdot \right\rangle \right) $ be a

2122: Hilbert space over $\mathbb{K}$, $p_{i}\geq 0,$ $i\in \mathbb{N}$ with $%

2123: \sum_{i=1}^{\infty }p_{i}=1.$ Consider $\ell _{\mathbf{p}}^{2}\left(

2124: K\right) $ as the space%

2125: \begin{equation*}

2126: \ell _{\mathbf{p}}^{2}\left( K\right) :=\left\{ x=\left( x_{i}\right)

2127: |x_{i}\in K,\ i\in \mathbb{N}\text{ \ and \ }\sum_{i=1}^{\infty

2128: }p_{i}\left\Vert x_{i}\right\Vert ^{2}<\infty \right\} .

2129: \end{equation*}%

2130: It is well known that $\ell _{\mathbf{p}}^{2}\left( K\right) $ endowed with

2131: the inner product%

2132: \begin{equation*}

2133: \left\langle x,y\right\rangle _{\mathbf{p}}:=\sum_{i=1}^{\infty

2134: }p_{i}\left\langle x_{i},y_{i}\right\rangle

2135: \end{equation*}%

2136: is a Hilbert space over $\mathbb{K}$. The norm $\left\Vert \cdot \right\Vert

2137: _{\mathbf{p}}$ of $\ell _{\mathbf{p}}^{2}\left( K\right) $ is given by%

2138: \begin{equation*}

2139: \left\Vert x\right\Vert _{\mathbf{p}}:=\left( \sum_{i=1}^{\infty

2140: }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{1}{2}}.

2141: \end{equation*}%

2142: If $x,y\in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ then the following

2143: Cauchy-Bunyakovsky-Schwarz (CBS) inequality holds true:%

2144: \begin{equation}

2145: \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\sum_{i=1}^{\infty

2146: }p_{i}\left\Vert y_{i}\right\Vert ^{2}\geq \left\vert \sum_{i=1}^{\infty

2147: }p_{i}\left\langle x_{i},y_{i}\right\rangle \right\vert ^{2}  \label{rev1.13}

2148: \end{equation}%

2149: with equality iff there exists a $\lambda \in \mathbb{K}$ such that $%

2150: x_{i}=\lambda y_{i}$ for each $i\in \mathbb{N}$.

2151: 

2152: If%

2153: \begin{equation*}

2154: \alpha \in \ell _{\mathbf{p}}^{2}\left( K\right) :=\left\{ \alpha =\left.

2155: \left( \alpha _{i}\right) _{i\in \mathbb{N}}\right\vert \alpha _{i}\in 

2156: \mathbb{K},\ i\in \mathbb{N}\text{ \ and \ }\sum_{i=1}^{\infty

2157: }p_{i}\left\vert \alpha _{i}\right\vert ^{2}<\infty \right\}

2158: \end{equation*}%

2159: and $x\in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ then the following

2160: (CBS)-type inequality is also valid:%

2161: \begin{equation}

2162: \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2163: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\geq \left\Vert

2164: \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert ^{2}  \label{rev1.14}

2165: \end{equation}%

2166: with equality if and only if there exists a vector $v\in K$ such that $x_{i}=%

2167: \overline{\alpha _{i}}v$ for each $i\in \mathbb{N}$.

2168: 

2169: In \cite{DRA3xx}, by the use of some preliminary results obtained in \cite%

2170: {SSDR0}, various reverses for the (CBS)-type inequalities (\ref{rev1.13})

2171: and (\ref{rev1.14}) for sequences of vectors in Hilbert spaces were

2172: obtained. Applications for bounding the distance to a finite-dimensional

2173: subspace and in reversing the generalised triangle inequality have also been

2174: provided.

2175: 

2176: The aim of the present section is to provide different results by employing

2177: some inequalities discovered in \cite{SSDR1}. Similar applications are

2178: pointed out.

2179: 

2180: \subsection{Reverses of the (CBS)-Inequality for Two Sequences in $\ell _{%

2181: \mathbf{p}}^{2}\left( K\right) $}

2182: 

2183: The following proposition may be stated \cite{5ab}.

2184: 

2185: \begin{proposition}

2186: \label{revp2.1}Let $x,y\in \ell _{\mathbf{p}}^{2}\left( K\right) $ and $r>0.$

2187: If%

2188: \begin{equation}

2189: \left\Vert x_{i}-y_{i}\right\Vert \leq r\text{ \ for each \ }i\in \mathbb{N}%

2190: \text{,}  \label{rev2.1}

2191: \end{equation}%

2192: then%

2193: \begin{align}

2194: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2195: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2196: 1}{2}}-\left\vert \sum_{i=1}^{\infty }p_{i}\left\langle

2197: x_{i},y_{i}\right\rangle \right\vert  \label{rev2.2} \\

2198: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2199: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2200: 1}{2}}-\left\vert \sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle

2201: x_{i},y_{i}\right\rangle \right\vert  \notag \\

2202: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2203: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2204: 1}{2}}-\sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle

2205: x_{i},y_{i}\right\rangle  \notag \\

2206: & \leq \frac{1}{2}r^{2}.  \notag

2207: \end{align}%

2208: The constant $\frac{1}{2}$ in front of $r^{2}$ is best possible in the sense

2209: that it cannot be replaced by a smaller quantity.

2210: \end{proposition}

2211: 

2212: \begin{proof}

2213: If (\ref{rev2.1}) holds true, then%

2214: \begin{equation*}

2215: \left\Vert x-y\right\Vert _{\mathbf{p}}^{2}=\sum_{i=1}^{\infty

2216: }p_{i}\left\Vert x_{i}-y_{i}\right\Vert ^{2}\leq r^{2}\sum_{i=1}^{\infty

2217: }p_{i}=r^{2}

2218: \end{equation*}%

2219: and thus $\left\Vert x-y\right\Vert _{\mathbf{p}}\leq r$.

2220: 

2221: Applying the inequality (\ref{rev1.6}) for the inner product $\left( \ell _{%

2222: \mathbf{p}}^{2}\left( K\right) ,\left\langle \cdot ,\cdot \right\rangle _{%

2223: \mathbf{p}}\right) ,$ we deduce the desired result (\ref{rev2.2}).

2224: 

2225: The sharpness of the constant follows by Theorem \ref{revt1} and we omit the

2226: details.

2227: \end{proof}

2228: 

2229: The following result may be stated as well \cite{5ab}.

2230: 

2231: \begin{proposition}

2232: \label{revp2.2}Let $x,y\in \ell _{\mathbf{p}}^{2}\left( K\right) $ and $%

2233: \gamma ,\Gamma \in \mathbb{K}$ with $\Gamma \neq -\gamma .$ If either%

2234: \begin{equation}

2235: \func{Re}\left\langle \Gamma y_{i}-x_{i},x_{i}-\gamma y_{i}\right\rangle

2236: \geq 0\ \text{\ for each}\ \ i\in \mathbb{N}  \label{rev2.3}

2237: \end{equation}%

2238: or, equivalently,%

2239: \begin{equation}

2240: \left\Vert x_{i}-\frac{\gamma +\Gamma }{2}y_{i}\right\Vert \leq \frac{1}{2}%

2241: \left\vert \Gamma -\gamma \right\vert \left\Vert y_{i}\right\Vert \ \text{\

2242: for each}\ \ i\in \mathbb{N}  \label{rev2.4}

2243: \end{equation}%

2244: holds, then:%

2245: \begin{align}

2246: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2247: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2248: 1}{2}}-\left\vert \sum_{i=1}^{\infty }p_{i}\left\langle

2249: x_{i},y_{i}\right\rangle \right\vert  \label{rev2.5} \\

2250: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2251: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2252: 1}{2}}  \notag \\

2253: & \qquad \qquad -\left\vert \func{Re}\left[ \frac{\bar{\Gamma}+\bar{\gamma}}{%

2254: \left\vert \Gamma +\gamma \right\vert }\sum_{i=1}^{\infty }p_{i}\left\langle

2255: x_{i},y_{i}\right\rangle \right] \right\vert  \notag \\

2256: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2257: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2258: 1}{2}}  \notag \\

2259: & \qquad \qquad -\func{Re}\left[ \frac{\bar{\Gamma}+\bar{\gamma}}{\left\vert

2260: \Gamma +\gamma \right\vert }\sum_{i=1}^{\infty }p_{i}\left\langle

2261: x_{i},y_{i}\right\rangle \right]  \notag \\

2262: & \leq \frac{1}{4}\cdot \frac{\left\vert \Gamma -\gamma \right\vert ^{2}}{%

2263: \left\vert \Gamma +\gamma \right\vert }\sum_{i=1}^{\infty }p_{i}\left\Vert

2264: y_{i}\right\Vert ^{2}.  \notag

2265: \end{align}%

2266: The constant $\frac{1}{4}$ is best possible in (\ref{rev2.5}).

2267: \end{proposition}

2268: 

2269: \begin{proof}

2270: Since, by (\ref{rev2.3}),%

2271: \begin{equation*}

2272: \func{Re}\left\langle \Gamma y-x,x-\gamma y\right\rangle _{\mathbf{p}%

2273: }=\sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle \Gamma

2274: y_{i}-x_{i},x_{i}-\gamma y_{i}\right\rangle \geq 0,

2275: \end{equation*}%

2276: hence, on applying the inequality (\ref{rev1.9}) for the Hilbert space 

2277: \newline

2278: $\left( \ell _{\mathbf{p}}^{2}\left( K\right) ,\left\langle \cdot ,\cdot

2279: \right\rangle _{\mathbf{p}}\right) ,$ we deduce the desired inequality (\ref%

2280: {rev2.5}).

2281: 

2282: The best constant follows by Theorem \ref{revt2} and we omit the details.

2283: \end{proof}

2284: 

2285: \begin{corollary}

2286: \label{revc2.3}If the conditions (\ref{rev2.3}) and (\ref{rev2.4}) hold for $%

2287: \Gamma =M,$ $\gamma =m$ with $M\geq m>0,$ then%

2288: \begin{align}

2289: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2290: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2291: 1}{2}}-\left\vert \sum_{i=1}^{\infty }p_{i}\left\langle

2292: x_{i},y_{i}\right\rangle \right\vert  \label{rev2.6} \\

2293: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2294: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2295: 1}{2}}-\left\vert \sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle

2296: x_{i},y_{i}\right\rangle \right\vert \displaybreak  \notag \\

2297: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert

2298: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert y_{i}\right\Vert ^{2}\right) ^{\frac{%

2299: 1}{2}}-\sum_{i=1}^{\infty }p_{i}\func{Re}\left\langle

2300: x_{i},y_{i}\right\rangle  \notag \\

2301: & \leq \frac{1}{4}\cdot \frac{\left( M-m\right) ^{2}}{M+m}\sum_{i=1}^{\infty

2302: }p_{i}\left\Vert y_{i}\right\Vert ^{2}.  \notag

2303: \end{align}%

2304: The constant $\frac{1}{4}$ is best possible.

2305: \end{corollary}

2306: 

2307: \subsection{Reverses of the (CBS)-Inequality for Mixed Sequences}

2308: 

2309: The following result holds \cite{5ab}:

2310: 

2311: \begin{theorem}[Dragomir, 2005]

2312: \label{revt3.1}Let $\alpha \in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ $%

2313: x\in \ell _{\mathbf{p}}^{2}\left( K\right) $ and $v\in K\backslash \left\{

2314: 0\right\} ,$ $r>0.$ If%

2315: \begin{equation}

2316: \left\Vert x_{i}-\overline{\alpha _{i}}v\right\Vert \leq r\left\vert \alpha

2317: _{i}\right\vert \ \text{\ for each}\ \ i\in \mathbb{N}  \label{rev3.1}

2318: \end{equation}%

2319: (note that if $\alpha _{i}\neq 0$ for any $i\in \mathbb{N}$, then the

2320: condition (\ref{rev3.1}) is equivalent to the simpler one%

2321: \begin{equation}

2322: \left\Vert \frac{x_{i}}{\overline{\alpha _{i}}}-v\right\Vert \leq r\ \text{\

2323: for each}\ \ i\in \mathbb{N}),  \label{rev3.2}

2324: \end{equation}%

2325: then%

2326: \begin{align}

2327: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2328: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2329: 1}{2}}-\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert 

2330: \label{rev3.3} \\

2331: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2332: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2333: 1}{2}}-\left\vert \left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},%

2334: \frac{v}{\left\Vert v\right\Vert }\right\rangle \right\vert   \notag \\

2335: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2336: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2337: 1}{2}}-\left\vert \func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2338: _{i}x_{i},\frac{v}{\left\Vert v\right\Vert }\right\rangle \right\vert  

2339: \notag \\

2340: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2341: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2342: 1}{2}}-\func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i},\frac{%

2343: v}{\left\Vert v\right\Vert }\right\rangle   \notag \\

2344: & \leq \frac{1}{2}\cdot \frac{r^{2}}{\left\Vert v\right\Vert }%

2345: \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert ^{2}.  \notag

2346: \end{align}%

2347: The constant $\frac{1}{2}$ is best possible in (\ref{rev3.3}).

2348: \end{theorem}

2349: 

2350: \begin{proof}

2351: From (\ref{rev3.1}) we deduce%

2352: \begin{equation*}

2353: \left\Vert x_{i}\right\Vert ^{2}-2\func{Re}\left\langle \alpha

2354: _{i}x_{i},v\right\rangle +\left\vert \alpha _{i}\right\vert ^{2}\left\Vert

2355: v\right\Vert ^{2}\leq r^{2}\left\vert \alpha _{i}\right\vert ^{2},

2356: \end{equation*}%

2357: which is clearly equivalent to%

2358: \begin{equation}

2359: \left\Vert x_{i}\right\Vert ^{2}+\left\vert \alpha _{i}\right\vert

2360: ^{2}\left\Vert v\right\Vert ^{2}\leq 2\func{Re}\left\langle \alpha

2361: _{i}x_{i},v\right\rangle +r^{2}\left\vert \alpha _{i}\right\vert ^{2}

2362: \label{rev3.4}

2363: \end{equation}%

2364: for each $i\in \mathbb{N}$.

2365: 

2366: If we multiply (\ref{rev3.4}) by $p_{i}\geq 0,$ $i\in \mathbb{N}$ and sum

2367: over $i\in \mathbb{N}$, then we deduce%

2368: \begin{multline}

2369: \qquad \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}+\left\Vert

2370: v\right\Vert ^{2}\sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2371: ^{2}  \label{rev3.5} \\

2372: \leq 2\func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2373: _{i}x_{i},v\right\rangle +r^{2}\sum_{i=1}^{\infty }p_{i}\left\vert \alpha

2374: _{i}\right\vert ^{2}.\qquad

2375: \end{multline}%

2376: Since, obviously%

2377: \begin{multline}

2378: \qquad 2\left\Vert v\right\Vert \left( \sum_{i=1}^{\infty }p_{i}\left\vert

2379: \alpha _{i}\right\vert ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert

2380: x_{i}\right\Vert ^{2}\right) ^{\frac{1}{2}}  \label{rev3.6} \\

2381: \leq \sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}+\left\Vert

2382: v\right\Vert ^{2}\sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2383: ^{2},\qquad

2384: \end{multline}%

2385: hence, by (\ref{rev3.5}) and (\ref{rev3.6}), we deduce%

2386: \begin{multline*}

2387: \qquad 2\left\Vert v\right\Vert \left( \sum_{i=1}^{\infty }p_{i}\left\vert

2388: \alpha _{i}\right\vert ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert

2389: x_{i}\right\Vert ^{2}\right) ^{\frac{1}{2}} \\

2390: \leq 2\func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2391: _{i}x_{i},v\right\rangle +r^{2}\sum_{i=1}^{\infty }p_{i}\left\vert \alpha

2392: _{i}\right\vert ^{2},\qquad

2393: \end{multline*}%

2394: which is clearly equivalent to the last inequality in (\ref{rev3.3}).

2395: 

2396: The other inequalities are obvious.

2397: 

2398: The best constant follows by Theorem \ref{revt1}.

2399: \end{proof}

2400: 

2401: The following corollary may be stated \cite{5ab}.

2402: 

2403: \begin{corollary}

2404: \label{revc3.1}Let $\alpha \in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ $%

2405: x\in \ell _{\mathbf{p}}^{2}\left( K\right) ,$ $e\in H,$ $\left\Vert

2406: e\right\Vert =1$ and $\gamma ,\Gamma \in \mathbb{K}$ with $\Gamma \neq

2407: -\gamma .$ If%

2408: \begin{equation}

2409: \left\Vert x_{i}-\overline{\alpha _{i}}\frac{\gamma +\Gamma }{2}\cdot

2410: e\right\Vert \leq \frac{1}{2}\left\vert \Gamma -\gamma \right\vert

2411: \left\vert \alpha _{i}\right\vert  \label{rev3.7}

2412: \end{equation}%

2413: for each $i\in \mathbb{N}$, or, equivalently,%

2414: \begin{equation}

2415: \func{Re}\left\langle \Gamma \overline{\alpha _{i}}e-x_{i},x_{i}-\gamma 

2416: \overline{\alpha _{i}}e\right\rangle  \label{rev3.8}

2417: \end{equation}%

2418: for each $i\in \mathbb{N}$ (note that, if $\alpha _{i}\neq 0$ for any $i\in 

2419: \mathbb{N}$, then (\ref{rev3.7}) is equivalent to%

2420: \begin{equation}

2421: \left\Vert \frac{x_{i}}{\overline{\alpha _{i}}}-\frac{\gamma +\Gamma }{2}%

2422: e\right\Vert \leq \frac{1}{2}\left\vert \Gamma -\gamma \right\vert

2423: \label{rev3.9}

2424: \end{equation}%

2425: for each $i\in \mathbb{N}$ and (\ref{rev3.8}) is equivalent to%

2426: \begin{equation}

2427: \func{Re}\left\langle \Gamma e-\frac{x_{i}}{\overline{\alpha _{i}}},\frac{%

2428: x_{i}}{\overline{\alpha _{i}}}-\gamma e\right\rangle \geq 0  \label{rev3.10}

2429: \end{equation}%

2430: for each $i\in \mathbb{N}$), then the following reverse of the

2431: (CBS)-inequality is valid:%

2432: \begin{align}

2433: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2434: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2435: 1}{2}}-\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert

2436: \label{rev3.11} \\

2437: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2438: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2439: 1}{2}}-\left\vert \left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2440: _{i}x_{i},e\right\rangle \right\vert  \notag \\

2441: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2442: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2443: 1}{2}}  \notag \\

2444: & \qquad \qquad -\left\vert \func{Re}\left[ \frac{\bar{\Gamma}+\bar{\gamma}}{%

2445: \left\vert \Gamma +\gamma \right\vert }\left\langle \sum_{i=1}^{\infty

2446: }p_{i}\alpha _{i}x_{i},e\right\rangle \right] \right\vert  \notag \\

2447: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2448: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2449: 1}{2}}  \notag \\

2450: & \qquad \qquad -\func{Re}\left[ \frac{\bar{\Gamma}+\bar{\gamma}}{\left\vert

2451: \Gamma +\gamma \right\vert }\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2452: _{i}x_{i},e\right\rangle \right]  \notag \\

2453: & \leq \frac{1}{4}\cdot \frac{\left\vert \Gamma -\gamma \right\vert ^{2}}{%

2454: \left\vert \Gamma +\gamma \right\vert }\sum_{i=1}^{\infty }p_{i}\left\vert

2455: \alpha _{i}\right\vert ^{2}.  \notag

2456: \end{align}%

2457: The constant $\frac{1}{4}$ is best possible.

2458: \end{corollary}

2459: 

2460: \begin{remark}

2461: If $M\geq m>0,$ $\alpha _{i}\neq 0$ and for $e$ as above, either%

2462: \begin{equation}

2463: \left\Vert \frac{x_{i}}{\overline{\alpha _{i}}}-\frac{M+m}{2}e\right\Vert

2464: \leq \frac{1}{2}\left( M-m\right) \ \text{\ for each}\ \ i\in \mathbb{N}

2465: \label{rev3.12}

2466: \end{equation}%

2467: or, equivalently,%

2468: \begin{equation*}

2469: \func{Re}\left\langle Me-\frac{x_{i}}{\overline{\alpha _{i}}},\frac{x_{i}}{%

2470: \overline{\alpha _{i}}}-me\right\rangle \geq 0\ \text{\ for each}\ \ i\in 

2471: \mathbb{N}

2472: \end{equation*}%

2473: holds, then%

2474: \begin{align*}

2475: (0& \leq )\left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2476: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2477: 1}{2}}-\left\Vert \sum_{i=1}^{\infty }p_{i}\alpha _{i}x_{i}\right\Vert %

2478: \displaybreak \\

2479: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2480: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2481: 1}{2}}-\left\vert \left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2482: _{i}x_{i},e\right\rangle \right\vert \\

2483: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2484: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2485: 1}{2}}-\left\vert \func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2486: _{i}x_{i},e\right\rangle \right\vert \\

2487: & \leq \left( \sum_{i=1}^{\infty }p_{i}\left\vert \alpha _{i}\right\vert

2488: ^{2}\sum_{i=1}^{\infty }p_{i}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{%

2489: 1}{2}}-\func{Re}\left\langle \sum_{i=1}^{\infty }p_{i}\alpha

2490: _{i}x_{i},e\right\rangle \\

2491: & \leq \frac{1}{4}\cdot \frac{\left( M-m\right) ^{2}}{M+m}\sum_{i=1}^{\infty

2492: }p_{i}\left\vert \alpha _{i}\right\vert ^{2}.

2493: \end{align*}%

2494: The constant $\frac{1}{4}$ is best possible.

2495: \end{remark}

2496: 

2497: \subsection{Reverses for the Generalised Triangle Inequality}

2498: 

2499: In 1966, Diaz and Metcalf \cite{DMxx} proved the following interesting

2500: reverse of the generalised triangle inequality:%

2501: \begin{equation}

2502: r\sum_{i=1}^{\infty }\left\Vert x_{i}\right\Vert \leq \left\Vert

2503: \sum_{i=1}^{\infty }x_{i}\right\Vert ,  \label{rev4.1}

2504: \end{equation}%

2505: provided the vectors $x_{1},\dots ,x_{n}\in H\backslash \left\{ 0\right\} $

2506: satisfy the assumption%

2507: \begin{equation}

2508: 0\leq r\leq \frac{\func{Re}\left\langle x_{i},a\right\rangle }{\left\Vert

2509: x_{i}\right\Vert },\qquad i\in \left\{ 1,\dots ,n\right\} ,  \label{rev4.2}

2510: \end{equation}%

2511: where $a\in H,$ $\left\Vert a\right\Vert =1$ and $\left( H;\left\langle

2512: \cdot ,\cdot \right\rangle \right) $ is a real or complex inner product

2513: space.

2514: 

2515: In an attempt to provide other sufficient conditions for (\ref{rev4.1}) to

2516: hold, the author pointed out in \cite{SEVER1} that%

2517: \begin{equation}

2518: \sqrt{1-\rho ^{2}}\sum_{i=1}^{\infty }\left\Vert x_{i}\right\Vert \leq

2519: \left\Vert \sum_{i=1}^{\infty }x_{i}\right\Vert  \label{rev4.3}

2520: \end{equation}%

2521: where the vectors $x_{i},$ $i\in \left\{ 1,\dots ,n\right\} $ satisfy the

2522: condition%

2523: \begin{equation}

2524: \left\Vert x_{i}-a\right\Vert \leq \rho ,\qquad i\in \left\{ 1,\dots

2525: ,n\right\} ,  \label{rev4.4}

2526: \end{equation}%

2527: where $r\in H,$ $\left\Vert a\right\Vert =1$ and $\rho \in \left( 0,1\right)

2528: .$

2529: 

2530: Following \cite{SEVER1}, if $M\geq m>0$ and the vectors $x_{i}\in H,$ $i\in

2531: \left\{ 1,\dots ,n\right\} $ verify either%

2532: \begin{equation}

2533: \func{Re}\left\langle Ma-x_{i},x_{i}-ma\right\rangle \geq 0,\qquad i\in

2534: \left\{ 1,\dots ,n\right\} ,  \label{rev4.5}

2535: \end{equation}%

2536: or, equivalently,%

2537: \begin{equation}

2538: \left\Vert x_{i}-\frac{M+m}{2}\cdot a\right\Vert \leq \frac{1}{2}\left(

2539: M-m\right) ,\qquad i\in \left\{ 1,\dots ,n\right\} ,  \label{rev4.6}

2540: \end{equation}%

2541: where $a\in H,$ $\left\Vert a\right\Vert =1,$ then%

2542: \begin{equation}

2543: \frac{2\sqrt{mM}}{M+m}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert \leq

2544: \left\Vert \sum_{i=1}^{n}x_{i}\right\Vert .  \label{rev4.7}

2545: \end{equation}%

2546: It is obvious from Theorem \ref{revt3.1}, that, if%

2547: \begin{equation}

2548: \left\Vert x_{i}-v\right\Vert \leq r,\qquad \text{for }\quad i\in \left\{

2549: 1,\dots ,n\right\} ,  \label{rev4.7.1}

2550: \end{equation}%

2551: where $x_{i}\in H,$ $i\in \left\{ 1,\dots ,n\right\} ,$ $v\in H\backslash

2552: \left\{ 0\right\} $ and $r>0,$ then we can state the inequality%

2553: \begin{align}

2554: (0& \leq )\left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2555: ^{2}\right) ^{\frac{1}{2}}-\left\Vert \frac{1}{n}\sum_{i=1}^{n}x_{i}\right%

2556: \Vert  \label{rev4.8} \\

2557: & \leq \left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2558: ^{2}\right) ^{\frac{1}{2}}-\left\vert \left\langle \frac{1}{n}%

2559: \sum_{i=1}^{n}x_{i},\frac{v}{\left\Vert v\right\Vert }\right\rangle

2560: \right\vert  \notag \\

2561: & \leq \left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2562: ^{2}\right) ^{\frac{1}{2}}-\left\vert \func{Re}\left\langle \frac{1}{n}%

2563: \sum_{i=1}^{n}x_{i},\frac{v}{\left\Vert v\right\Vert }\right\rangle

2564: \right\vert  \notag \\

2565: & \leq \left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2566: ^{2}\right) ^{\frac{1}{2}}-\func{Re}\left\langle \frac{1}{n}%

2567: \sum_{i=1}^{n}x_{i},\frac{v}{\left\Vert v\right\Vert }\right\rangle  \notag

2568: \\

2569: & \leq \frac{1}{2}\cdot \frac{r^{2}}{\left\Vert v\right\Vert }.  \notag

2570: \end{align}%

2571: Since, by the (CBS)-inequality we have%

2572: \begin{equation}

2573: \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert \leq \left( \frac{1}{n}%

2574: \sum_{i=1}^{n}\left\Vert x_{i}\right\Vert ^{2}\right) ^{\frac{1}{2}},

2575: \label{rev4.9}

2576: \end{equation}%

2577: hence, by (\ref{rev4.8}) and (\ref{rev4.5}) we have \cite{5ab}:%

2578: \begin{equation}

2579: (0\leq )\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert -\left\Vert

2580: \sum_{i=1}^{n}x_{i}\right\Vert \leq \frac{1}{2}n\cdot \frac{r^{2}}{%

2581: \left\Vert v\right\Vert }  \label{rev4.9.1}

2582: \end{equation}%

2583: provided that (\ref{rev4.7.1}) holds true.

2584: 

2585: Utilising Corollary \ref{revc3.1}, we may state that, if 

2586: \begin{equation}

2587: \left\Vert x_{i}-\frac{\gamma +\Gamma }{2}\cdot e\right\Vert \leq \frac{1}{2}%

2588: \left\vert \Gamma -\gamma \right\vert ,\qquad i\in \left\{ 1,\dots

2589: ,n\right\} ,  \label{rev4.10}

2590: \end{equation}%

2591: or, equivalently,%

2592: \begin{equation}

2593: \func{Re}\left\langle \Gamma e-x_{i},x_{i}-\gamma e\right\rangle \geq

2594: 0,\qquad i\in \left\{ 1,\dots ,n\right\} ,  \label{rev4.11}

2595: \end{equation}%

2596: where $e\in H,$ $\left\Vert e\right\Vert =1,\gamma ,\Gamma \in \mathbb{K}$, $%

2597: \Gamma \neq -\gamma $ and $x_{i}\in H,$ $i\in \left\{ 1,\dots ,n\right\} ,$

2598: then%

2599: \begin{align}

2600: (0& \leq )\left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2601: ^{2}\right) ^{\frac{1}{2}}-\left\Vert \frac{1}{n}\sum_{i=1}^{n}x_{i}\right%

2602: \Vert  \label{rev4.12} \\

2603: & \leq \left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2604: ^{2}\right) ^{\frac{1}{2}}-\left\vert \left\langle \frac{1}{n}%

2605: \sum_{i=1}^{n}x_{i},e\right\rangle \right\vert  \notag \\

2606: & \leq \left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2607: ^{2}\right) ^{\frac{1}{2}}-\left\vert \func{Re}\left[ \frac{\bar{\Gamma}+%

2608: \bar{\gamma}}{\left\vert \Gamma +\gamma \right\vert }\left\langle \frac{1}{n}%

2609: \sum_{i=1}^{n}x_{i},e\right\rangle \right] \right\vert  \notag \\

2610: & \leq \left( \frac{1}{n}\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert

2611: ^{2}\right) ^{\frac{1}{2}}-\func{Re}\left[ \frac{\bar{\Gamma}+\bar{\gamma}}{%

2612: \left\vert \Gamma +\gamma \right\vert }\left\langle \frac{1}{n}%

2613: \sum_{i=1}^{n}x_{i},e\right\rangle \right]  \notag \\

2614: & \leq \frac{1}{4}\cdot \frac{\left\vert \Gamma -\gamma \right\vert ^{2}}{%

2615: \left\vert \Gamma +\gamma \right\vert }.  \notag

2616: \end{align}%

2617: Now, making use of (\ref{rev4.9}) and (\ref{rev4.12}) we can establish the

2618: following additive reverse of the generalised triangle inequality \cite{5ab}%

2619: \begin{equation}

2620: (0\leq )\sum_{i=1}^{n}\left\Vert x_{i}\right\Vert -\left\Vert

2621: \sum_{i=1}^{n}x_{i}\right\Vert \leq \frac{1}{4}n\cdot \frac{\left\vert

2622: \Gamma -\gamma \right\vert ^{2}}{\left\vert \Gamma +\gamma \right\vert },

2623: \label{rev4.13}

2624: \end{equation}%

2625: provided either (\ref{rev4.10}) or (\ref{rev4.11}) hold true.

2626: 

2627: \subsection{Applications for Fourier Coefficients}

2628: 

2629: Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) $ be a Hilbert

2630: space over the real or complex number field $\mathbb{K}$ and $\left\{

2631: e_{i}\right\} _{i\in I}$ an \textit{orthonormal basis} for $H.$ Then (see

2632: for instance \cite[p. 54 -- 61]{DExx}):

2633: 

2634: \begin{enumerate}

2635: \item[(i)] Every element $x\in H$ can be expanded in a \textit{Fourier series%

2636: }, i.e.,%

2637: \begin{equation*}

2638: x=\sum_{i\in I}\left\langle x,e_{i}\right\rangle e_{i},

2639: \end{equation*}%

2640: where $\left\langle x,e_{i}\right\rangle ,$ $i\in I$ are the \textit{Fourier

2641: coefficients} of $x;$

2642: 

2643: \item[(ii)] (Parseval identity)%

2644: \begin{equation*}

2645: \left\Vert x\right\Vert ^{2}=\sum_{i\in I}\left\langle x,e_{i}\right\rangle

2646: e_{i},\quad x\in H;

2647: \end{equation*}

2648: 

2649: \item[(iii)] (Extended Parseval's identity)%

2650: \begin{equation*}

2651: \left\langle x,y\right\rangle =\sum_{i\in I}\left\langle

2652: x,e_{i}\right\rangle \left\langle e_{i},y\right\rangle ,\qquad x,y\in H;

2653: \end{equation*}

2654: 

2655: \item[(iv)] (Elements are uniquely determined by their Fourier coefficients)%

2656: \begin{equation*}

2657: \left\langle x,e_{i}\right\rangle =\left\langle y,e_{i}\right\rangle \text{

2658: \ \ for every }i\in I\text{ \ implies that \ }x=y.

2659: \end{equation*}

2660: \end{enumerate}

2661: 

2662: We must remark that all the results from the second and third sections may

2663: be stated for $K=\mathbb{K}$ where $\mathbb{K}$ is the Hilbert space of

2664: complex (real) numbers endowed with the usual norm and inner product.

2665: 

2666: Therefore we can state the following reverses of the Schwarz inequality \cite%

2667: {5ab}:

2668: 

2669: \begin{proposition}

2670: \label{revp5.1}Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) 

2671: $ be a Hilbert space over $\mathbb{K}$ and $\left\{ e_{i}\right\} _{i\in I}$

2672: an orthonormal base for $H.$ If $x,y\in H,$ $y\neq 0,$ $a\in \mathbb{K}$ $%

2673: \left( \mathbb{C},\mathbb{R}\right) $ with $r>0$ such that%

2674: \begin{equation}

2675: \left\vert \frac{\left\langle x,e_{i}\right\rangle }{\left\langle

2676: y,e_{i}\right\rangle }-a\right\vert \leq r\qquad \text{for each \ }i\in I,

2677: \label{rev5.1}

2678: \end{equation}%

2679: then we have the following reverse of the Schwarz inequality:%

2680: \begin{align}

2681: (0& \leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

2682: \left\langle x,y\right\rangle \right\vert  \label{rev5.2} \\

2683: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert \func{Re}%

2684: \left[ \left\langle x,y\right\rangle \cdot \frac{\bar{a}}{\left\vert

2685: a\right\vert }\right] \right\vert  \notag \\

2686: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}\left[

2687: \left\langle x,y\right\rangle \cdot \frac{\bar{a}}{\left\vert a\right\vert }%

2688: \right]  \notag \\

2689: & \leq \frac{1}{2}\cdot \frac{r^{2}}{\left\vert a\right\vert }\left\Vert

2690: y\right\Vert ^{2}.  \notag

2691: \end{align}%

2692: The constant $\frac{1}{2}$ is best possible in (\ref{rev5.2}).

2693: \end{proposition}

2694: 

2695: The proof is similar to the one in Theorem \ref{revt3.1}, where instead of $%

2696: x_{i}$ we take $\left\langle x,e_{i}\right\rangle $, instead of $\alpha _{i}$

2697: we take $\left\langle e_{i},y\right\rangle ,$ $\left\Vert \cdot \right\Vert

2698: =\left\vert \cdot \right\vert ,$ $p_{i}=1$ and use the Parseval identities

2699: mentioned above in (ii) and (iii). We omit the details.

2700: 

2701: The following result may be stated as well \cite{5ab}.

2702: 

2703: \begin{proposition}

2704: \label{revp5.2}Let $\left( H;\left\langle \cdot ,\cdot \right\rangle \right) 

2705: $ be a Hilbert space over $\mathbb{K}$ and $\left\{ e_{i}\right\} _{i\in I}$

2706: an orthonormal base for $H.$ If $x,y\in H,$ $y\neq 0,$ $e,\gamma ,\Gamma \in 

2707: \mathbb{K}$ with $\left\vert e\right\vert =1,$ $\Gamma \neq -\gamma $ and%

2708: \begin{equation}

2709: \left\vert \frac{\left\langle x,e_{i}\right\rangle }{\left\langle

2710: y,e_{i}\right\rangle }-\frac{\gamma +\Gamma }{2}\cdot e\right\vert \leq 

2711: \frac{1}{2}\left\vert \Gamma -\gamma \right\vert  \label{rev5.3}

2712: \end{equation}%

2713: or equivalently,%

2714: \begin{equation}

2715: \func{Re}\left[ \left( \Gamma e-\frac{\left\langle x,e_{i}\right\rangle }{%

2716: \left\langle y,e_{i}\right\rangle }\right) \left( \frac{\left\langle

2717: e_{i},x\right\rangle }{\left\langle e_{i},y\right\rangle }-\bar{\gamma}\bar{e%

2718: }\right) \right] \geq 0  \label{rev5.4}

2719: \end{equation}%

2720: for each $i\in I,$ then%

2721: \begin{align}

2722: (0& \leq )\left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert

2723: \left\langle x,y\right\rangle \right\vert  \label{rev5.5} \\

2724: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\left\vert \func{Re}%

2725: \left[ \frac{\bar{\Gamma}+\bar{\gamma}}{\left\vert \Gamma +\gamma

2726: \right\vert }\left\langle x,y\right\rangle \cdot \bar{e}\right] \right\vert 

2727: \notag \\

2728: & \leq \left\Vert x\right\Vert \left\Vert y\right\Vert -\func{Re}\left[ 

2729: \frac{\bar{\Gamma}+\bar{\gamma}}{\left\vert \Gamma +\gamma \right\vert }%

2730: \left\langle x,y\right\rangle \cdot \bar{e}\right]  \notag \\

2731: & \leq \frac{1}{4}\cdot \frac{\left\vert \Gamma -\gamma \right\vert ^{2}}{%

2732: \left\vert \Gamma +\gamma \right\vert }\left\Vert y\right\Vert ^{2}.  \notag

2733: \end{align}%

2734: The constant $\frac{1}{4}$ is best possible.

2735: \end{proposition}

2736: 

2737: \begin{remark}

2738: If $\Gamma =M\geq m=\gamma >0,$ then one may state simpler inequalities from

2739: (\ref{rev5.5}). We omit the details.

2740: \end{remark}

2741: 

2742: \begin{thebibliography}{99}

2743: \bibitem{BExx} R. BELLMAN, Almost orthogonal series,\textit{\ Bull. Amer.

2744: Math. Soc}., \textbf{50} (1944), 517-519.

2745: 

2746: \bibitem{BOxx} R.P. BOAS, A general moment problem,\textit{\ Amer. J. Math}%

2747: ., \textbf{63} (1941), 361-370.

2748: 

2749: \bibitem{BOMxx} E. BOMBIERI, A note on the large sieve,\textit{\ Acta Arith}%

2750: ., \textbf{18} (1971), 401-404.

2751: 

2752: \bibitem{DExx} F. DEUTSCH, \textit{Best Approximation in Inner Product

2753: Spaces, }CMS Books in Mathematics, Springer-Verlag, New York, Berlin,

2754: Heidelberg, 2001.

2755: 

2756: \bibitem{DMxx} J.B. DIAZ and F.T. METCALF, A complementary triangle

2757: inequality in Hilbert and Banach spaces, \textit{Proc. Amer. Math. Soc., }%

2758: \textbf{17}(1) (1966), 88-99.

2759: 

2760: \bibitem{DRA4xx} S.S. DRAGOMIR, A counterpart of Bessel's inequality in

2761: inner product spaces and some Gr\"{u}ss type related results, \textit{RGMIA

2762: Res. Rep. Coll., }\textbf{6} (2003), Supplement, Article 10. [ONLINE: 

2763: \texttt{http://rgmia.vu.edu.au/v6(E).html}].

2764: 

2765: \bibitem{SSDxx} S.S. DRAGOMIR, \textit{Advances in Inequalities of the

2766: Schwarz, Gr\"{u}ss and Bessel Type in Inner Product Spaces, }RGMIA

2767: Monographs, Victoria University, 2004. [ONLINE\texttt{\

2768: http://rgmia.vu.edu.au/monographs/advancees.htm}].

2769: 

2770: \bibitem{SSDR1} S.S. DRAGOMIR, New reverses of Schwarz, triangle and Bessel

2771: inequalities in inner product spaces, \textit{Australian J. Math. Anal. \&

2772: Appl., }\textbf{1}(1) (2004), Art. 1. [ONLINE \texttt{http://ajmaa.org/}].

2773: 

2774: \bibitem{SSDR0} S.S. DRAGOMIR, Reverses of Schwarz, triangle and Bessel

2775: inequalities in inner product spaces, \textit{J. Ineq. Pure \& Appl. Math., }%

2776: \textbf{5}(3) (2004), Art. 74. [ONLINE \texttt{%

2777: http://jipam.vu.edu.au/article.php?sid=432}].

2778: 

2779: \bibitem{DRA2xx} S.S. DRAGOMIR, Reverses of the triangle inequality in inner

2780: product spaces, \textit{RGMIA Res. Rep. Coll., }\textbf{7} (2004),

2781: Supplement, Article 7. [ONLINE: \texttt{http://rgmia.vu.edu.au/v7(E).html}].

2782: 

2783: \bibitem{DRA3xx} S.S. DRAGOMIR, Reversing the CBS-inequality for sequences

2784: of vectors in Hilbert spaces with applications (I), \textit{RGMIA\ Res. Rep.

2785: Coll}., \textbf{8}(2005), Supplement, Article 2. \texttt{[ONLINE

2786: http://rgmia.vu.edu.au/v8(E).html]}.

2787: 

2788: \bibitem{5ab} S.S. DRAGOMIR, Reversing the CBS-inequality for sequences of

2789: vectors in Hilbert spaces with applications (II), \textit{RGMIA\ Res. Rep.

2790: Coll}., \textbf{8}(2005), Supplement, Article 3. \texttt{[ONLINE

2791: http://rgmia.vu.edu.au/v8(E).html]}.

2792: 

2793: \bibitem{DRA5xx} S.S. DRAGOMIR, Some Bombieri type inequalities in inner

2794: product spaces,\textit{\ J. Indones. Math. Soc}., \textbf{10}(2) (2004),

2795: 91-97.

2796: 

2797: \bibitem{SEVER1} S.S. DRAGOMIR, Reverses of the triangle inequality in inner

2798: product spaces, \textit{RGMIA\ Res. Rep. Coll}., \textbf{7 }(2004),

2799: Supplement, Article 7. [ONLINE\texttt{\ http://rgmia.vu.edu.au/v7(E).html}].

2800: 

2801: \bibitem{DRA1xx} S.S. DRAGOMIR, On the Boas-Bellman inequality in inner

2802: product spaces,\textit{\ Bull. Austral. Math. Soc}., \textbf{69}(2) (2004),

2803: 217-225.

2804: 

2805: \bibitem{SILV1xx} S.S. DRAGOMIR, Upper bounds for the distance to

2806: finite-dimensional subspaces in inner product spaces, \textit{RGMIA Res.

2807: Rep. Coll., }\textbf{7}(1) (2005), Article 2. [ONLINE: \texttt{%

2808: http://rgmia.vu.edu.au/v7n1.html}].

2809: 

2810: \bibitem{MPFxx} D.S. MITRINOVI\'{C}, J.E. PE\v{C}ARI\'{C} and\ A.M. FINK, 

2811: \textit{Classical and New Inequalities in Analysis, }Kluwer Academic,

2812: Dordrecht, 1993.

2813: \end{thebibliography}

2814: