1: \documentclass{elsart}

2: \usepackage{graphicx}

3: \usepackage {subfigure,epsfig}

4: 

5: \usepackage {amsmath} \usepackage{amssymb} \usepackage{cite} %\usepackage{amsthm}

6: \numberwithin{equation}{section} %\numberwithin{table}{section}

7: %\mathindent=0pt

8: %\theoremstyle{plain} \newtheorem{theorem}{Theorem}

9: 

10: %\numberwithin{theorem}{section}

11: 

12: \begin{document}

13: \begin{frontmatter}

14: 

15: \title{Power and non-power expansions of the solutions for the fourth-order analogue to the second Painlev\'{e} equation}

16: \author{Maria V. Demina,}

17: \author{Nikolai A. Kudryashov\corauthref{cor}}

18: \corauth[cor]{Corresponding author.}

19: \address{Department of Applied Mathematics, Moscow  Engineering

20:  and Physics Institute (State university), 31 Kashirskoe Shosse,

21: 115409 Moscow, Russian Federation} \ead{kudryashov@mephi.ru}

22: 

23: 

24: 

25: 

26: \begin{abstract}

27: Fourth - order analogue to the second Painlev\'{e} equation is

28: studied. This equation has its origin in the modified Korteveg - de

29: Vries equation of the fifth order when we look for its self -

30: similar solution. All power and non - power expansions of the solutions for the fouth

31: - order analogue to the second Painlev\'{e} equation near points

32: $z=0$ and $z=\infty$ are found by means of the power geometry

33: method. The exponential additions to solutions of the equation

34: studied are determined. Comparison of the expansions found with

35: those of the six Painlev\'{e} equations confirm the conjecture that the

36: fourth - order analogue to the second Painlev\'{e} equation defines

37: new transcendental functions.

38: 

39: \end{abstract}

40: 

41: 

42: \end{frontmatter}

43: 

44: 

45: \section{Introduction.}

46: More than one century ago Painlev\'{e} and his collaborators

47: analyzed a certain class of the second-order nonlinear ordinary

48: differential equations (ODE). In fact Painlev\'{e} examined two

49: problems. The first one was to find all the second-order canonical

50: equations with the solutions without movable critical points. The

51: second problem was to pick out those equations which have solutions

52: defining new special functions. The former was posed by Picard in

53: 1876. While the latter was mentioned by A. Pouancare and L. Fuchs in

54: 1884.

55: 

56: As a result of investigations Painlev\'{e} and his school discovered

57: six second-order nonlinear ODEs which solutions could not be

58: expressed in terms of known elementary or special functions.

59: Nowadays they go under the name of the Painlev\'{e} equations and

60: their solutions are called the Painlev\'{e} transcendents.

61: 

62: For a long period of time the Painlev\'{e} equations were regarded

63: as nothing more than a curious moment in the theory of differential

64: equations and were beyond the center of  scientific circles

65: attention. However since the sixtieth of the twentieth century a

66: number of works appeared where it was shown that the Painlev\'{e}

67: transcendents arose in the models describing physical phenomena as

68: frequently as many other special functions \cite{Ablowitz01,

69: Barouch03, Brezin04, De_Boer05, Ablowitz06, Hall07, Chandrasecar08,

70: Kudryashov01, Kudryashov02}. This fact caused a significant interest

71: to the studying of their properties and set a problem to find other

72: nonlinear ODEs defining new transcendents.

73: 

74: It is necessary to mention that exact solutions of many partial

75: nonlinear differential equations (such as the Korteweg-de Vries

76: equation, the modified Korteweg-de Vries equation, the nonlinear

77: Schr\"{o}dinger equation, the sine-Gordon equation and so on)

78: solvable by the inverse scattering transform can be expressed via

79: the Painlev\'{e} transcendents and it is one of their most important

80: applications \cite{Ablowitz01, Ablowitz06, Kudryashov01,

81: Kudryashov03, Hone01}. In this connection it is quiet natural to

82: suppose that the partial solutions of exactly solvable equations of

83: the order higher than the order of previously mentioned equations

84: also can be expressed in terms of the transcendents defined as the

85: solutions of nonlinear ODEs. This idea was developed in the work

86: \cite{Kudryashov03}, where an hierarchy of the first Painlev\'{e}

87: equation was introduced. Nowadays the higher analogues of the

88: Painlev\'{e} equations are intensively studied \cite{Kudryashov03,

89: Kudryashov04,Kudryashov05, Kudryashov06, Kudryashov07, Kudryashov08,

90: Kudryashov09, Kudryashov10, Kudryashov11, Kudryashov12,

91: Kudryashov13, Kudryashov14, Kudryashov15, Kudryashov16,

92: Airault01,Clarkson01, Clarkson02, Cosgrove01, Joshi01, Hone01,

93: Hone02, Gordoa01, Gordoa02, Flashka01, Kawai01, Mugan01, Mugan02,

94: Mugan03, Mugan04, Nijhoff01, Pickering01, Shimomura01}.

95: 

96: Taking after \cite{Kudryashov03, Airault01, Hone01, Flashka01} we

97: will show that a solution of the modified Korteveg - de Vries

98: equation of the fifth order

99: \begin{equation}

100: \label{1.1}u_t+u_{xxxxx}-10u^{2}u_{xxx}-40u_{x}u_{xx}-10u_{x}^{3}+30u^{4}u_{x}=0

101: \end{equation}

102: is defined by the solution of the fourth-order analogue of the

103: second Painlev\'{e} equation. Let us look for the solution of the

104: equation \eqref{1.1} in the form

105: \begin{equation}

106: \label{1.2}u(x,t)=\frac{1}{(5t)^{1/5}} \,w(z),\,\,\,\,\,z=\frac

107: x{(5t)^{1/5}}.

108: \end{equation}

109: After substitution of \eqref{1.2} into \eqref{1.1} and integrating

110: once by $z$ we get the equation

111: \begin{equation}

112: \label{1.3}f(z,w)\stackrel{def}{=}w_{zzzz} - 10w^2w_{zz} - 10ww^2_z

113: + 6 w^5  - zw - \alpha =0

114: \end{equation}

115: This equation has a number of properties similar to that of

116: Painlev\'{e} equations. More exactly it possesses B\"{a}cklund

117: transformations \cite{Kudryashov05, Clarkson01, Hone01}, a Lax pair

118: \cite{Kudryashov05}, rational and special solutions at certain

119: values of the parameter $\alpha$ \cite{Kudryashov05, Kudryashov11,

120: Clarkson01, Hone01}. The Caushy problems for the equation

121: \eqref{1.3} can be solved by the isomonodromic transfom method. It

122: is likely that the equation \eqref{1.3} defines new transcendental

123: functions as Painlev\'{e} equations do. However the exact proof of

124: this statement yet is an open question. In this connection an

125: important step of the investigation of the equation \eqref{1.3}

126: properties is the finding of all its solution asymptotics.

127: 

128: The aim of this work is to calculate all power and non-power

129: asymptotics, power and exponential expansions for the solutions of

130: the equation \eqref{1.3} with a help of the power geometry method

131: \cite{Bruno01, Bruno02, Bruno03}.

132: 

133: The problem definition is

134: 

135: 1) to find all power asymptotics of its solutions. If a solution of

136: the equation supposing that $z\rightarrow 0$ or $z\rightarrow

137: \infty$ can be presented in the form

138: \begin{equation}

139: \label{1.4} w(z)= c_r\,z^{r}+ o(|z|^{r+\varepsilon}),

140: \end{equation}

141: where the coefficient $c_r=const\in \mathbb{C}$, $c_r \neq 0$, the

142: exponents $r,\varepsilon \in \mathbb{R}$, $\varepsilon \omega < 0$

143: and

144: \begin{equation*}

145: \omega=\left\{

146: \begin{gathered}

147: -1,\,z\rightarrow 0;\\

148: \hfill 1,z\rightarrow\infty,

149: \end{gathered}

150: \right.

151: \end{equation*}

152: then the power asymptotic of the solution \eqref{1.4} is

153: \begin{equation}

154: \label{1.5} w(z)= c_r\,z^{r}.

155: \end{equation}

156: 

157: 2) to calculate all power expansions of its solutions given by

158: \begin{equation}

159: \label{1.6} w(z)= c_r\,z^{r}+ \sum_{s}\,c_s\,z^{s},

160: \end{equation}

161: where $s>r$ if $z\rightarrow\,0$ and $s<r$ if

162: $z\rightarrow\,\infty$.

163: 

164: 3) to find for the studied equation all non-power asymptotics, i.e.

165: functions $u(z)$ that are connected with the solutions of the

166: equation \eqref{1.3} in the way

167: \begin{equation}

168: \label{1.7} w(z)= u(z)\left[1+o(|z|^{\omega \varepsilon}

169: |u(z)|^{\omega_1\varepsilon})\right]

170: \end{equation}

171: for some $\varepsilon<0$ where

172: \begin{equation*}

173: \omega_1=\left\{

174: \begin{gathered}

175: \,\,0,\,u(z)\rightarrow const\neq0;\\

176: -1,\hfill u(z)\rightarrow0;\\

177: \,\,1,\hfill u(z)\rightarrow\infty.

178: \end{gathered}

179: \right.

180: \end{equation*}

181: 

182: 4) to find all exponential additions to power expansions of the

183: solutions which correspond to the exponentially close solutions. In

184: other words it is required to find functions

185: \begin{equation}

186: \label{1.8} v(z)= \exp \left[\sum \gamma_sz^s\right],\,\,\,\omega

187: s<const,

188: \end{equation}

189: which, added to the solution $w(z)+v(z)$, correspond to the power

190: expansions.

191: 

192: This paper outline is as follows. The general properties of the

193: fourth - order analogue to the second Painlev\'{e} equation are

194: discussed in section 2. Power expansions corresponding to the apexes

195: and to the edges are given in sections 3, 4, 5, 6 and 7. In section

196: 8 we consider the general properties of the equation studied in the

197: special case (at $\alpha=0$). Corresponding power and non-power

198: expansions of solutions are presented in sections 9, 10 and 11. The

199: three-level exponential additions are determined in sections 12, 13,

200: 14, 15,16, 17 and 18. The results of the work are gathered in

201: section 19.

202: 

203: 

204: 

205: \section{General properties of the equation \eqref{1.3} at $\alpha \neq 0$.}

206: The following points correspond to the monomials of the studied

207: equation: $M_1=(-4,1),\,\,\, M_2=(-2,3),\,\,\,M_3=(-2,3),\,\,\,

208: M_4=(0,5),\,\,\, M_5=(1,1),\,\,\, M_6=(0,0)$. The carrier $S(f)$ of

209: the equation contains five points $Q_1=M_1,\,\,\, Q_2=M_4,\,\,\,

210: Q_3=M_5 ,\,\,\, Q_4=M_6$ è $Q_5=M_2=M_3$. Their convex hull is the

211: quadrangle with four apexes $\Gamma_j^{(0)}=Q_j\,(j=1,2,3,4)$ and

212: four edges

213: $\Gamma_1^{(1)}=[Q_1,Q_2],\,\Gamma_2^{(1)}=[Q_2,Q_3],\,\Gamma_3^{(1)}=[Q_3,Q_4],\,\Gamma_4^{(1)}=[Q_1,Q_4]$

214: (fig. 1).

215: \begin{figure}[h]

216:  \centerline{\epsfig{file=Fig1.eps,width=120mm}}

217:  \caption{}\label{fig:z_post}

218: \end{figure}

219: 

220: The external normal vectors $N_j\,(j=1,2,3,4)$ to edges

221: $\Gamma_j^{(1)}\,(j=1,2,3,4)$ are $N_1=(-1,1),\, N_2=(4,1),\,

222: N_3=(1,-1),\, N_4=(-1,-4)$. They form the normal cones $U_j^{(1)}$

223: of edges $\Gamma_j^{(1)}$

224: \begin{equation}

225: \label{2.1}U_j^{(1)} =\{P=\lambda N_j,\,\,\, \lambda>0\},\,\,\,

226: j=1,2,3,4.

227: \end{equation}

228: The normal cones $U_j^{(0)}$ of the apexes $\Gamma_j^{(0)}=Q_j\,\,

229: (j=1,\,2,\,3,\,4)$ are the angles between the edges that adjoin to

230: the apex (fig. 2).

231: \begin{figure}[h]

232:  \centerline{\epsfig{file=Fig2.eps,width=120mm}}

233:  \caption{}\label{fig:z_post}

234: \end{figure}

235: 

236: The carrier of the equation \eqref{1.1} lies in the lattice

237: $\mathbb{Z}$ with the basis $B_1=Q_1=(-4,1),\,B_2=Q_3=(1,1)$.

238: Expressing the points $Q_2,\,Q_5$ through the measuring vectors we

239: get $Q_2=4B_2+B_1$, $Q_5=B_1+2B_2$. Examining the reduced equations

240: that correspond to the bounds $\Gamma_j^{(d)}\,\, (d=0,1;\,\,

241: j=1,\,2,\,3,\,4)$ we will study the solutions of the equation

242: \eqref{1.1}. Note that the reduced equations which correspond to the

243: apexes $\Gamma_2^{(0)},\,\Gamma_3^{(0)},\,\Gamma_4^{(0)}$

244: \begin{equation}

245: \begin{gathered}

246: \label{2.4}\hat{f}_2^{(0)}\stackrel{def}{=}6w^5=0,\\

247: \hat{f}_3^{(0)}\stackrel{def}{=} -zw=0,\\

248: \hat{f}_4^{(0)}\stackrel{def}{=}-\alpha=0

249: \end{gathered}

250: \end{equation}

251: are the algebraic ones and that is why they do not have non-trivial

252: power or non-power solutions.

253: 

254: \section{Power expansions, corresponding to the apex $\Gamma_1^{(0)}$ at $\alpha \neq 0$.}

255: The apex $\Gamma_1^{(0)}$ defines the following reduced equation

256: \begin{equation}

257: \label{3.1}\hat{f}_1^{(0)}\stackrel{def}{=} w_{zzzz}=0.

258: \end{equation}

259: Let us find the reduced solutions

260: \begin{equation}

261: \label{3.2}w=c_r z^r,\,\,\, c_r\neq0

262: \end{equation}

263: for $\omega (1,r) \in U_1^{(0)}$. Since $p_1 < 0$ in the cone

264: $U_1^{(0)}$ then $\omega=-1, \,\,\, z\rightarrow 0$ and the

265: expansions will be the ascending power series of $z$. Substituting

266: \eqref{3.2} into $\hat{f}_1^{(0)}$ and canceling the result by

267: $z^{r-4}$ we get the characteristic equation

268: \begin{equation}

269: \label{3.3}\chi(r)\stackrel{def}{=} r(r-1)(r-2)(r-3)=0

270: \end{equation}

271: with the four roots $r_1=0,\,r_2=1,\,r_3=2,\,$ and $r_4=3$. Further

272: we shall examine them separately.

273: 

274: Vector $R=(1,0)$ that corresponds to the root $r_1=0$ being

275: multiplied by $\omega$ belongs to the cone $U_1^{(0)}$. Thus we

276: obtain the family $F_1^{(0)}\,1$ of power asymptotics $w=c_0$ with

277: the arbitrary constant $c_0 \neq 0$. The first variation of the

278: equation \eqref{3.1}

279: \begin{equation}

280: \label{3.4}\frac{\delta\hat{f}_1^{(0)}}{\delta w} = \frac{d^4}{dz^4}

281: \end{equation}

282: gives the operator

283: \begin{equation}

284: \label{3.5}L(z)=\frac{d^4}{dz^4} \neq 0

285: \end{equation}

286: with the characteristic polynomial

287: \begin{equation}

288: \label{3.6}\nu(k) =z^{4-k} L(z) z^k =k(k-1)(k-2)(k-3).

289: \end{equation}

290: The equation $ \nu(k)=0 $ has four roots $k_1=0,\,\,\, k_2=1,\,\,\,

291: k_3=2,\,\,\,$è $k_4=3$. As $\omega=-1$ and $r=0$ then the cone of

292: the problem is $\mathcal{K}=\{k>0\}$. It contains

293: $k_2=1,\,k_3=2,\,k_3=3$ which are the critical numbers. Let us

294: investigate the equation obtained from the equation \eqref{1.3}

295: after changing of variables $w(z)=c_{r_1}z^{r_1}+y(z)$

296: \begin{equation}

297: \label{3.7}g(z,y) \stackrel{def}{=} f(z,c_{r_1}z^{r_1}+y).

298: \end{equation}

299: Its carrier $ S(g)$ contains the points belonging to $ S(f)$ and

300: seven other points: $ (-2,2),\, (-2,1),\, (0,4),\, (0,3),\, (0,2),\,

301: (0,1),\, (1,0)$. Displacing all these points on the vector $(-4,1)$

302: we get $(2,1)$, $ (2,0)$, $(4,3)$, $(4,2)$, $(4,1)$, $(4,0)$,

303: $(5,1)$, $(4,4)$, $(0,0)$, $(4,-1)$, $(5,0)$. The set $ S^{'}_{+}$

304: of the finite sums of the vectors based on these points intersects

305: with the line $ q_{2}=-1 $ at the points

306: \begin{equation}

307: \label{3.8}\mathbb{K}=\{k=4+n,\,\,\,n\in N\cup{\{0\}}\}.

308: \end{equation}

309: The critical numbers do not lie in this set that is why the indexes

310: $s$ in the expression \eqref{1.6} belong to the set

311: $\mathbb{K}(1,2,3)$.

312: \begin{equation}

313: \label{3.9}\mathbb{K}(1,2,3)=\{k=n,\,\,\,n\in N\ \}.

314: \end{equation}

315: Thus the expansion of the solution corresponding to the reduced

316: equation \eqref{3.1} at $r=0$ is the following

317: \begin{equation}

318: \label{3.10}w(z)=c_0+\sum_{k\in\mathbb{K}(1,2,3)}c_kz^k\equiv c_0 +

319: c_1 z + c_2z^2 + c_3z^3 + \sum^{\infty}_{k=4} c_k z^k

320: \end{equation}

321: where all the coefficients are constants, $c_0\neq

322: 0,\,c_1,\,c_2,\,c_3$ are arbitrary ones and $c_k \,\, (k\geq 4)$ are

323: uniquely defined. Denote this family as $G_1^{(0)}\,1$. Taking into

324: account eight members, the expansion \eqref{3.10} can be presented

325: in the form

326: \begin{equation*}\begin{gathered}

327: w(z)=c_{{0}}+c_{{1}}\,z+c_{{2}}\,{z}^{2}+c_{{3}}\,{z}^{3}+\left(

328: \frac1{24} \,\alpha-\frac14\,{c_{{0}}}^{5}+{\frac

329: {5}{12}}\,c_{{0}}{c_{{1}}}^{2}+\frac56 \,{c_{{0}}}^{2}c_{{2}}

330: \right) {z}^{4}+

331: \\+\left(\frac1{12}\,{c_{{1}}}^{3}+\frac12\,{c_{{0}}}^{2}c_{{3}}-\frac14\,{c_{{0}}}^{4}

332: c_{{1}}+{\frac{1}{ 120}}\,c_{{0}}+\frac23\,c_{{0}}c_{{1}}c_{{2}}

333: \right) {z}^{5}+

334: \\+\left( \frac29\,c_{{0}}{c_{{2}}}^{2}+\frac13

335: \,{c_{{0}}}^{2}c_{{4}}+\frac12\,c_{{0}}c_{{1}}c_{{3}}+{\frac

336: {1}{360}}\,c_ {{1}}+{\frac

337: {7}{36}}\,c_{{2}}{c_{{1}}}^{2}-\frac1{12}\,{c_{{0}}}^{4}c_{{2}}

338: -\frac16\,{c_{{0}}}^{3}{c_{{1}}}^{2} \right) {z}^{6}+

339: \\+\left( {\frac

340: {5}{21}}\,{c_{{0}}}^{2}c_{{5}}+{\frac {1}{840}}\,c_{{2}

341: }-\frac1{28}\,{c_{{0}}}^{4}c_{{3}}+{\frac{8}{21}}\,c_{{0}}c_{{1}}c_{{4}}+\frac13\,c_{{0}}c_{{2}}c_{{3}}+

342: \right. \\

343: \left. + \frac17\,{c_{{2}}}^{2}c_{{1}}-\frac17\,{c_{{0}}}^{3}c

344: _{{1}}c_{{2}} +{\frac

345: {13}{84}}\,{c_{{1}}}^{2}c_{{3}}-\frac1{14}\,{c_{{0}}}^{

346: 2}{c_{{1}}}^{3} \right){z}^{7}+...

347: \end{gathered}\end{equation*}

348: 

349: The cone of the problem is $\mathcal{K}=\{k>1\}$ in the case

350: $r_2=1$. Consequently there are two critical numbers $k_2=2,\,

351: k_3=3$. Likewise the previous case we find the solution expansion

352: \begin{equation}

353: \label{3.11}w(z)=c_1z + c_2z^2 + c_3z^3 + \sum^{\infty}_{k=4} c_kz^k

354: \end{equation}

355: that is generated by the power asymptotic $\mathcal{F}_1^{(0)}2:

356: \,\,\, w= c_1\,z$. Here $c_1\neq 0,\, c_2$ and $c_3$ are arbitrary

357: constants.

358: 

359: For the root $r_2=2$ the cone of the problem is

360: $\mathcal{K}=\{k>2\}$. It contains $k_3=3$ that is the unique

361: critical number. The power expansion corresponding to the asymptotic

362: $\mathcal{F}_1^{(0)}3: \,\,\, w= c_2\,z^{2}$ takes the form

363: \begin{equation}

364: \label{3.12}w(z)=c_2z^2 +c_3z^3 + \sum^{\infty}_{k=4} c_kz^k.

365: \end{equation}

366: Again $c_2\neq 0,\,c_3$ are arbitrary constants. Denote this family

367: as $\emph{G}_1^{(0)}3$.

368: 

369: For the root $r_3=3$ the cone of the problem is

370: $\mathcal{K}=\{k>3\}$. There are no critical numbers in this case.

371: The expansion of the solution $\emph{G}_1^{(0)}4$ corresponding to

372: the power asymptotic $\mathcal{F}_1^{(1)}4: \,\,\, w= c_3\,z^{3}$

373: can be written as

374: \begin{equation}

375: \label{3.13}w(z)=c_3z^3 + \sum^{\infty}_{k=4} ñ_kz^k.

376: \end{equation}

377: 

378: Obtained expansions converge for small $|z|$. The existence and the

379: analyticity of the expansions \eqref{3.10}, \eqref{3.11},

380: \eqref{3.12}, \eqref{3.13} follow from the Cauchy theorem. The apex

381: does not define non-power asymptotics and exponential additions.

382: 

383: \section{Power expansions, corresponding to the edge $\Gamma_1^{(1)}$ at $\alpha \neq 0$.}

384: 

385: The edge $\Gamma_1^{(1)}$ is characterized by the reduced equation

386: \begin{equation}

387: \label{4.1}\hat{f}^{(1)}_1 (z,w) \stackrel{def}{=}w_{zzzz}

388: -10\,w^2\,w_{zz} -10\,w\,w_{z}^2+6\,w^5=0

389: \end{equation}

390: and the normal cone $U^{(1)}_1=\{-\lambda(1,-1),\,\lambda>0\}$.

391: Therefore $\omega=-1$, i.e. $z\rightarrow 0$ and $r=-1$.

392: Consequently the solution of the equation \eqref{4.1} should be

393: looked for in the form $w=c_{-1} z^{-1}$. From the determining

394: equation

395: \begin{equation}

396: \label{4.2}c_{-1}(3\,c_{-1}^4 -15\,c_{-1}^2 +12) =0

397: \end{equation}

398: we find the values of coefficient $c_{-1}$ (not equal to zero):

399: $c_{-1}^{(1)} =1,\, c_{-1}^{(2)}=-1,\, c_{-1}^{(3)}=2,\,

400: c_{-1}^{(4)}=-2$. Hence we have four families of power asymptotics

401: \begin{equation}

402: \label{4.3}\mathcal{F}_1^{(1)}1: \,\,\,

403: w=z^{-1};\,\,\quad\,\mathcal{F}_1^{(1)}2: \,\,\, w=-z^{-1};

404: \end{equation}

405: \begin{equation}

406: \label{4.4}\mathcal{F}_1^{(1)}3: \,\,\,

407: w=2z^{-1};\,\,\quad\,\mathcal{F}_1^{(1)}4: \,\,\, w=-2z^{-1}.

408: \end{equation}

409: Let us compute the corresponding critical numbers. The first

410: variation

411: \begin{equation}

412: \label{4.5}\frac{\delta f_2^{(1)}}{\delta w}=\frac{d^4}{dz^4}

413: -20w_{zz}w -10w^2\frac{d^2}{dz^2} -10w_{z}^2 -10\,ww_z \frac d{dz} +

414: 30\,w^4

415: \end{equation}

416: applied to the the solutions \eqref{4.4} yields the operator

417: \begin{equation}

418: \label{4.6}\mathcal{L}^{(1,2)}(z)

419: =\frac{d^4}{dz^4}-\frac{10}{z^2}\frac{d^2}{dz^2}+\frac{20}{z^3}

420: \frac{d}{dz} - \frac{20}{z^4}.

421: \end{equation}

422: Its characteristic polynomial is

423: \begin{equation}

424: \label{4.7}\nu(k ) ={k}^{4}-6\,{k}^{3}+{k}^{2}+24\,k-20.

425: \end{equation}

426: The equation $\nu(k)=0$ has the roots $k_1=-2,\,k_2=1,\,

427: k_3=2,\,k_4=5$.

428: 

429: With reference to the solutions \eqref{4.5} variation \eqref{4.6}

430: gives the operator

431: \begin{equation}

432: \label{4.8}\mathcal{L}^{(3,4)}(z) =\frac{d^4}{dz^4} - \frac{40}{z^2}

433: \frac{d^2}{dz^2} +\frac{80}{z^3} \frac{d}{dz} - \frac{280}{z^4},

434: \end{equation}

435: Its characteristic polynomial

436: \begin{equation}

437: \label{4.9}\nu(k)={k}^{4}-6\,{k}^{3}-29\,{k}^{2}+114\,k+280,

438: \end{equation}

439: has the roots $k_1=-4,\, k_2=-2,\, k_3=5,\, k_4=7$. The cone of the

440: problem looks like

441: \begin{equation}

442: \label{4.10} \mathcal{K}=\{k>-2\}.

443: \end{equation}

444: Thus for the power asymptotics \eqref{4.4} there are three critical

445: numbers (three roots of the characteristic polynomial belong to the

446: cone of the problem) and for the power asymptotics \eqref{4.5} there

447: are only two critical numbers. The shifted carrier of the power

448: asymptotics \eqref{4.4}, \eqref{4.5} is the vector $(-1,-1)=-B_{2}$.

449: It belongs to the lattice that consists of the points $Q=

450: m(-4,1)+l(1,1)$ where m and l  are the whole numbers. These points

451: intersect with the line $q_{2}=-1$ if $m+l=-1$, i.e. $l=-m-1$. As

452: the cone of the problem is \eqref{4.11} then

453: \begin{equation}

454: \label{4.11}\textbf{K}=\{k=-1+5m,\,\,\,m\in \mathbb{N}\}.

455: \end{equation}

456: Now we will look for the solution expansions generated by the

457: families \eqref{4.4}. The sets $\textbf{K}(1)$, $\textbf{K}(1,2)$

458: and $\textbf{K}(1,2,5)$ can be written as

459: \begin{equation} \begin{gathered}

460: \label{4.12}\textbf{K}(1)=\{k=-1+5m+2l,\,\,l,m\in

461: \mathbb{N}\cup\{0\},\,\,l+m\neq\,0\}=\\

462: =\{1,3,4,5,6,7,8,...\};

463: \end{gathered}\end{equation}

464: \begin{equation}\begin{gathered}

465: \label{4.13}\textbf{K}(1,2)=\{k=-1+5m+2l+3k,\,l,m,k\in

466: \mathbb{N}\cup\{0\},\,\,m+l+k\neq\,0\}=\\

467: =\{1,2,3,4,5,6,7,8,...\};

468: \end{gathered} \end{equation}

469: \begin{equation}\begin{gathered}

470: \label{4.14}\textbf{K}(1,2,5)=\{k=-1+5m+2l+3k+6n,\,n,m,k,l \in

471: \mathbb{N}\cup\{0\} ,\,\,m+n+k+l\neq\,0\}=\\

472: =\{n  ,\,\,n \in \mathbb{N}\}.

473: \end{gathered}\end{equation}

474: In this case the solution expansions are

475: \begin{equation}\begin{gathered}

476: \label{4.15}w(z)=\frac{c^{(1,2)}_{-1}}{z}+

477: \sum^{}_{k\,\in\,\textbf{K}(1,2,5)} c_{k}^{(1,2)}\,z^{k}.

478: \end{gathered}\end{equation}

479: Denote these families as $G_{1}^{(1)}1$ and $G_{1}^{(1)}2$.

480: Evidently the critical number $1$ does not belong to the set

481: $\textbf{K}$ that is why the compatibility condition for $c_1$ holds

482: automatically and in consequence $c_1$ is an arbitrary constant. The

483: critical number $2$ also does not belong to the sets $\textbf{K}$,

484: $\textbf{K}(1)$ and thus $c_2$ is an arbitrary constant. However the

485: critical number $5$ lies in the sets $\textbf{K}(1)$ and

486: $\textbf{K}(1,2)$. As a result it is necessary to verify that the

487: compatibility condition for $c_5$ is true. It turns out that it is

488: so then $c_5$ is an arbitrary constant. The three-parametric power

489: expansion that corresponds to  the power asymptotics \eqref{4.4} (at

490: $c_{-1}=1$) is the following

491: \begin{equation*}\begin{gathered}

492: \label{}

493: w(z)=\frac{1}{z}+c_{{1}}z+c_{{2}}{z}^{2}+\frac52\,{c_{{1}}}^{2}{z}^{3}+

494: \left( \frac53 \,c_{{1}}c_{{2}}-\frac1{36}\,\alpha-\frac1{36}

495: \right) {z}^{4}+c_{{5}}{z}^{5}+

496: \\ +\left( \frac58\,c_{{2}}c_{{3}}+{\frac

497: {37}{48}}\,{c_{{1}}}^{2}c_{{2}}-{ \frac

498: {7}{288}}\,c_{{1}}\alpha-{\frac {13}{720}}\,c_{{1}} \right) {z}^

499: {6}+

500: \\+ \left( {\frac

501: { 17}{27}}\,c_{{1}}{c_{{2}}}^{2}-{\frac

502: {17}{1620}}\,c_{{2}}-\frac1{18}\,{c_{{1}}}^{4}-{\frac

503: {1}{81}}\,c_{{2}}\alpha-{\frac {

504: 5}{54}}\,{c_{{1}}}^{2}c_{{3}}+\frac59\,c_{{1}}c_{{5}}+{\frac

505: {11}{54}}\,{c _{{3}}}^{2} \right) {z}^{7}+...

506: \end{gathered}

507: \end{equation*}

508: Here $c_i\equiv c_i^{(1)},\,i=1,2,5$.

509: 

510: The carrier of the power expansions generated by the families

511: \eqref{4.5} is defined by the sets

512: \begin{equation}\begin{gathered}

513: \label{4.16}\textbf{K}(5)=\{k=-1+5m+6l,\,l,m\in

514: \mathbb{N}\cup\{0\},\,\,m+l\neq\,0\}=\\

515: =\{4,5,9,10,11,14,15,16,17,19,...\};

516: \end{gathered}\end{equation}

517: \begin{equation}\begin{gathered}

518: \label{4.17}\textbf{K}(5,7)=\{k=-1+5m+6l+8k,\,m,l,k\in

519: \mathbb{N}\cup\{0\},\,\,m+l+k\neq\,0\}=\\

520: =\{4,5,7,9,10,11,12,13,14,...\}.

521: \end{gathered}\end{equation}

522: The solution expansions in this case are

523: \begin{equation}\begin{gathered}

524: \label{4.18}w(z)=\frac{c^{(3,4)}_{-1}}{z}+ \sum^{}_{k\,\, \in \,\,

525: \textbf{K}(5,7)} c_{k}^{(3,4)}\,z^{k}.

526: \end{gathered}\end{equation}

527: Denote these families as $G_{1}^{(1)}3$, $G_{1}^{(1)}4$. Obviously the

528: critical numbers $5$ and $7$ do not lie in the set $\textbf{K}$ and

529: besides the critical number $7$ does not belong to the set

530: $\textbf{K}(5)$. For the values 5, 7 the compatibility condition

531: holds automatically and as a consequence the coefficients $c_5$,

532: $c_7$ are arbitrary ones. The two-parametric power expansion that

533: corresponds to the reduced solutions \eqref{4.5} (at $c_{-1}=2$) is

534: \begin{equation*}\begin{gathered}

535: \label{}w(z)=\frac{2}{{z}}+2\,{z}^{-1}+ \left( {\frac

536: {1}{72}}+{\frac {1}{144}}\,\alpha \right) {

537: z}^{4}+c_{{5}}{z}^{5}+c_{{7}}{z}^{7}+

538: \\+ \left( {\frac {17}{370656}}+{

539: \frac {59}{1482624}}\,\alpha+{\frac {25}{2965248}}\,{\alpha}^{2}

540:  \right) {z}^{9}+c_{{5}} \left( {\frac {19}{3780}}+{\frac {1}{432}}\,

541: \alpha \right) {z}^{10}+

542: \\+{\frac {11}{78}}\,{c_{{5}}}^{2}{z}^{11}+c_{{7}

543: } \left( {\frac {127}{35280}}+{\frac {1}{576}}\,\alpha \right)

544: {z}^{12}+...,

545: \end{gathered}\end{equation*}

546: where $c_i\equiv c_i^{(3)},\,i=5,7$. Obtained expansions converge

547: for small $|z|$. The exponential additions for the expansions

548: \eqref{4.15}, \eqref{4.18} do not exist. The reduced equation

549: \eqref{4.1} does not have non-power solutions.

550: 

551: 

552: 

553: \section{Power expansions, corresponding to the edge $\Gamma_2^{(1)}$ at $\alpha \neq 0$.}

554: 

555: The edge $\Gamma_2^{(1)}$ is characterized by the reduced equation

556: \begin{equation}

557: \label{5.1}\hat{f}_2^{(1)} (z,w) \stackrel{def}{=} 6\,w^5 - zw=0

558: \end{equation}

559: and the normal cone

560: $U^{(1)}_2=\{\lambda(4,1)=4\lambda(1,1/4),\,\lambda>0\}$. It means

561: that $r=1/4$, $\omega=1$, $z\rightarrow  \, \infty$ and the solution

562: of the reduced equation is $y=c_{1/4}z^{1/4}$. Substitution this

563: expression into \eqref{5.1} and cancelation the result by $z^{5/4}$

564: yields the determining equation $ c_{1/4}(6c_{1/4}^4-1)=0$. Thus we

565: have four families of power asymptotics

566: \begin{equation}

567: \label{5.2}\mathcal{F}_2^{(1)}1:\,\,\,\, w=

568: c_{1/4}^{(1)}\,z^{1/4},\,\quad\,c_{1/4}^{(1)}=\left(\frac{1}{6}\right)^{1/4};

569: \end{equation}

570: \begin{equation}

571: \label{5.3}\mathcal{F}_2^{(1)}2:\,\,\,\, w=c_{1/4}^{(2)}\,

572: z^{1/4},\,\quad\,c_{1/4}^{(2)}\,=-\left(\frac {1}{6}\right)^{1/4};

573: \end{equation}

574: \begin{equation}

575: \label{5.4}\mathcal{F}_2^{(1)}3:\,\,\,\,w=c_{1/4}^{(3)}\,z^{1/4},\,\quad\,c_{1/4}^{(3)}\,=i\left(\frac

576: {1}{6}\right)^{1/4};

577: \end{equation}

578: \begin{equation}

579: \label{5.5}\mathcal{F}_2^{(1)}4:\,\,\,\, w=

580: c_{1/4}^{(4)}\,z^{1/4},\,\quad\,c_{1/4}^{(4)}=-i\left(\frac{1}{6}\right)^{1/4}.

581: \end{equation}

582: Since the reduced equation \eqref{5.1} is algebraic and the roots of

583: the determining equation are simple then asymptotics do not have

584: proper (and consequently  critical) numbers and $\nu(k)\equiv

585: const\neq0$. The shifted carrier of the power asymptotics

586: \eqref{5.2} -- \eqref{5.5} gives a vector $B=\left(1/4,-1\right)$.

587: Points belonging to the lattice generated by vectors $B_{2}$, $B$

588: are the following $Q=(q_1,q_2) =m(1,\,1) +l(1/4,\,-1)

589: =\left(m+l/4,\,\,m-l \right)$ where $m$, $l$ are whole numbers. At

590: the line $q_2=-1$ we have $q_2=-1$ and $q_1=-1+5/4l$. Taking into

591: consideration that the cone of the problem is

592: $\mathcal{K}=\{k<1/4\}$ we find the set $\mathbf{K}$

593: \begin{equation}

594: \label{5.6}\mathbf{K}=\{k=-1-5l/4,\,\,\, l\in \mathbb{N}\cup\{0\}

595: \},

596: \end{equation}

597: The expansions of the solutions can be written as

598: \begin{equation}

599: \label{5.7}G_2^{(1)} n:\,\,\,\, w(z)=\varphi^{(n)}(z)=c^{(n)}_{1/4}

600: z^{1/4} + \sum^{\infty}_{l=0} c^{(n)}_{-1-5l/4}\, z^{-1-5l/4}.

601: \end{equation}

602: In this expression coefficients $c^{(n)}_{-1-5l/4}$ can be

603: sequentially computed. The calculation of the coefficients $c_{-1}$

604: yields $c_{-1}=\alpha/4$. Taking into account four terms, the

605: expansions are

606: \begin{equation*}

607: \begin{gathered}

608: \label{}w(z)=\varphi^{(n)}(z)=c_{{1/4}}^{(n)}\,{z^{1/4}}+\frac14\,{\frac

609: {\alpha}{z}}-{\frac {5}{16}}\,{ \frac {(c_{{1/4}}^{(n)})^{3} \left(

610: 1+3\,{\alpha}^{2} \right) }{{z}^{9/4}}}+

611: \\+{\frac {5}{128}}\,{\frac

612: {(c_{{1/4}}^{(n)})^{2}\alpha\, \left( 29+24\,{ \alpha}^{2}

613: \right)}{{z}^{7/2}}}+...

614: \end{gathered}

615: \end{equation*}

616: It is likely that the obtained expansions diverge. Non-power

617: asymptotics do not correspond to the edge $\Gamma^{(1)}_2$ but it

618: generates exponential additions which will be computed later.

619: 

620: \section{Power expansions, corresponding to the edge $\Gamma_3^{(1)}$ at $\alpha \neq 0$.}

621: 

622: The edge $\Gamma_3^{(1)}$ is characterized by the reduced equation

623: \begin{equation}

624: \label{6.1}\hat{f}_3^{(1)} (z,w) \stackrel{def}{=} - zw - \alpha=0

625: \end{equation}

626: and the normal cone $U^{(1)}_3=\{\lambda(1,-1),\,\lambda>0\}$. In

627: this case $r=-1$, $\omega=1$, $z\longrightarrow \, \infty$ and power

628: asymptotic can be presented in the form

629: \begin{equation}

630: \label{6.2}\mathcal{F}_3^{(1)}1:\,\,\,\, w= \frac

631: {c_{-1}}{z},\,\quad\,c_{-1}=-\alpha.

632: \end{equation}

633: As the equation \eqref{6.1} is algebraic, then its solutions do not

634: have critical numbers and

635: \begin{equation}

636: \label{6.3}\nu(k)=z^{-1}\frac{\delta f_3^{(1)}}{\delta y}=-1.

637: \end{equation}

638: The cone of the problem is $\mathcal{K}=\{k<-1\}$. The shifted

639: carrier of the power asymptotic \eqref{6.2} gives the vector

640: $(-1,-1)=-B_{2}$, which belongs to the lattice generated by the

641: carrier of the studied equation. That is why

642: \begin{equation}

643: \label{6.4}\mathbf{K}=\{k=-1-5m,\,\,\, m\in \mathbb{N} \}.

644: \end{equation}

645: So we have determined the power expansion corresponding to the

646: asymptotic \eqref{6.2}

647: \begin{equation}

648: \label{6.5}G_3^{(1)}:\,\,\,\, w(z)=\psi(z)=\frac {c_{-1}}{z} +

649: \sum^{\infty}_{m=1} c_{-1-5m}\, z^{-1-5m}.

650: \end{equation}

651: In this expression all coefficients can be sequentially found. Again

652: taking into account four terms, it can be rewritten as

653: \begin{equation*}

654: \begin{gathered}

655: \label{}w(z)=\psi(z)=-{\frac {\alpha}{z}}-6\,{\frac

656: {c}{{z}^{6}}}-12\,{\frac {c \left( 15\,

657: {\alpha}^{4}-295\,{\alpha}^{2}+1512 \right) }{{z}^{11}}}- \\

658: \\ -216\,{\frac { c \left(35\,{\alpha}^{8}-2090\,{\alpha}^{6}+49755\,{\alpha}^{4}-

659: 528180\,{\alpha}^{2}+2018016 \right) }{{z}^{16}}}+...,

660: \end{gathered}

661: \end{equation*}

662: where $c=\alpha(\alpha-1)(\alpha-2)(\alpha+1)(\alpha+2)$ and all

663: other not written out coefficients are proportional to this factor.

664: Unless $c=0$ the expansion \eqref{6.5} seems to be divergent one for

665: all $|z|^{-1}\neq0$.

666: 

667: The reduced equation \eqref{6.1} does not have non-power solutions.

668: The edge $\Gamma^{(1)}_3$ defines exponential additions which will

669: be found below.

670: 

671: 

672: \section{Power expansions, corresponding to the edge $\Gamma_4^{(1)}$ at $\alpha \neq 0$.}

673: 

674: The edge $\Gamma_4^{(1)}$ defines the following reduced equation

675: \begin{equation}

676: \label{7.1}\hat{f}_4^{(1)} (z,w)\stackrel{def}{=}w_{zzzz} - \alpha=0

677: \end{equation}

678: and the normal cone $U^{(1)}_4 =\{-\lambda(1,4),\,\,\lambda>0\}$.

679: Thus $\omega=-1$, i.e. $z\rightarrow 0$ è $r=4$ and we have the

680: unique family of power asymptotics

681: \begin{equation}

682: \label{7.2}\mathcal{F}_4^{(1)}1:

683: \,\,\,w=c_4z^4,\,\,\,c_4=\frac{\alpha}{24}.

684: \end{equation}

685: Let us find the critical numbers. The first variation of the

686: equation \eqref{7.1} is

687: \begin{equation}

688: \label{7.3}\frac{\delta \hat{f}^{(1)}_4}{\delta w}

689: =\frac{d^4}{dz^4}.

690: \end{equation}

691: The proper numbers are $k_1=0,\,\,\, k_2=1,\,\,\, k_3=2,\,\,\,

692: k_4=3$. None of them belongs to the cone of the problem $

693: \mathcal{K}=\{k>4\}$. Consequently in this case there are no

694: critical numbers. The shifted carrier of \eqref{7.2} is equal to the

695: vector $B_1$ and the lattice generated by the vectors $B_1$, $B_2$

696: corresponds to the power asymptotic \eqref{7.2}. Because of this the

697: set $\textbf{K}$ is

698: \begin{equation}

699: \label{7.4}\textbf{K}=\{k=4+5m,\,\,\,m\in \mathbb{N}\}.

700: \end{equation}

701: The power expansion can be written as

702: \begin{equation}

703: \label{7.5}w(z)=z^4\left(\frac{\alpha}{24}+\sum^{\infty}_{m=1}

704: c_{4+5m}\,z^{5m}\right).

705: \end{equation}

706: The coefficient $c_{4+5m}$ can be uniquely computed. The first three

707: terms of the found expansion $G_4^{(1)}1$ are

708: \begin{equation*}

709: \label{}w(z)=\alpha\left(\frac1{24}\, {z}^{4}+{\frac

710: {1}{72576}}\,{z}^{9}+( {\frac {1}{1743565824}}+{\frac

711: {5}{5930496}}\,{\alpha}^{ 2} ) {z}^{14}+...\right).

712: \end{equation*}

713: The expansion \eqref{7.5} can be regarded as the special case of the

714: expansion \eqref{3.10} at $c_0=c_1=c_2=c_3=0$. It converge for small

715: $|z|$. The existence, the uniqueness and the analyticity of such

716: expansion follow from the above mentioned Caushy theorem.

717: 

718: The edge $\Gamma^{(1)}_4$ does not define exponential additions and

719: non-power asymptotics.

720: 

721: \section{General properties of the equation \eqref{1.3} at $\alpha = 0$.}

722: 

723: If $\alpha=0$ the fourth-order analogue to the second Painlev\'{e}

724: equation is

725: \begin{equation}

726: \label{8.1}\tilde{f}(z,w)\stackrel{def}{=}w_{zzzz} - 10w^2w_{zz} -

727: 10ww^2_z + 6 w^5  - zw =0.

728: \end{equation}

729: The carrier $S(\tilde{f})$ of this equation consists of the four

730: points: $Q_1=(-4,1),\,\,Q_2=(0,5),\,\,Q_3=(1,1)$ è $Q_4=(-2,3)$.

731: Their convex hull is the triangle with the apexes

732: $\Gamma^{(0)}_j=Q_j$ $(j=1,2,3)$ and edges

733: $\Gamma^{(1)}_1=\{Q_1,Q_2\},\,\,\Gamma^{(1)}_2=\{Q_2,Q_3\},\,\,\Gamma^{(1)}_1=\{Q_3,Q_1\}$

734: (fig. 3).

735: \begin{figure}[h]

736:  \centerline{\epsfig{file=Fig3.eps,width=100mm}}

737:  \caption{}\label{fig:z_post}

738: \end{figure}

739: 

740: The external normals $N_j$ to the edges $\Gamma^{(1)}_j$ are the

741: vectors $N_1=(-1,1),\,\,N_2=(4,1),\,\,N_3=(0,-1)$. Normal cones

742: $U_j^{(1)}$ of the edges $\Gamma_j^{(1)}$ can be written as

743: \begin{equation}

744: \label{8.2}U_j^{(1)} =\{P=\lambda N_j,\,\,\, \lambda>0\},\,\,\,

745: j=1,2,3.

746: \end{equation}

747: Normal cones $U_j^{(0)}$ of the apexes $\Gamma_j^{(0)}$ and the

748: normal cones \eqref{8.2} are presented at the fig. 4. The shifted

749: carrier of the equation \eqref{8.1} lies in the lattice with the

750: basis $B_3=(-3,2)$ and $B_4=(-1,4)$.

751: \begin{figure}[h]

752:  \centerline{\epsfig{file=Fig4.eps,width=100mm}}

753:  \caption{}\label{fig:z_post}

754: \end{figure}

755: 

756: Further we will study the reduced equations corresponding to the

757: apexes $\Gamma_j^{(0)}$ and to the edges $\Gamma_j^{(1)}$

758: $(j=1,2,3)$. Again the reduced equations

759: \begin{equation}

760: \label{8.3}\hat{\tilde{f}}_2^{(0)}(z,w)\stackrel{def}{=}6w^5=0,\,\,\hat{\tilde{f}}_3^{(0)}(z,w)\stackrel{def}{=}-zw=0,

761: \end{equation}

762: which are defined by the apexes $\Gamma_2^{(0)}$, $\Gamma_3^{(0)}$

763: accordingly are the algebraic ones. Because of this they do not have

764: non-trivial power or non-power solutions.

765: 

766: 

767: \section{Power expansions, corresponding to the apex $\Gamma_1^{(0)}$ and to the edge $\Gamma^{(1)}_1$ at $\alpha =0$.}

768: 

769: The apex $\Gamma_1^{(0)}$ is characterized by the following reduced

770: equation

771: \begin{equation}

772: \label{9.1}\hat{\tilde{f}}_1^{(0)}(z,w)\stackrel{def}{=} w_{zzzz}=0.

773: \end{equation}

774: This case is similar to the one discussed in the section 3.

775: Consequently there are four families of power expansions:

776: four-parametric family $\tilde{G}_1^{(0)}1$, tree-parametric family

777: $\tilde{G}_1^{(0)}2$, two-parametric family $\tilde{G}_1^{(0)}3$ and

778: one-parametric family $\tilde{G}_1^{(0)}4$.

779: 

780: Examining by analogy with the section 4 the reduced equation which

781: corresponds to the edge $\Gamma^{(1)}_1$

782: \begin{equation}

783: \label{9.2}\hat{\tilde{f}}^{(1)}_1 (z,w) \stackrel{def}{=}w_{zzzz}

784: -10\,w^2\,w_{zz} -10\,w\,w_{z}^2+6\,w^5=0,

785: \end{equation}

786: we get two tree-parametric families $\tilde{G}_1^{(1)}1$,

787: $\tilde{G}_1^{(1)}2$ and two two-parametric families

788: $\tilde{G}_1^{(1)}3$, $\tilde{G}_1^{(1)}4$ of power expansions.

789: 

790: The wave in these designations means that in the similar expressions

791: for the equation \eqref{1.3} it is provided that $\alpha=0$.

792: 

793: 

794: \section{Power expansions, corresponding to the edge $\Gamma_2^{(1)}$ at $\alpha =0$.}

795: 

796: The edge $\Gamma^{(1)}_2$ defines the reduced equation which can be

797: written as

798: \begin{equation}

799: \label{10.1}\hat{\tilde{f}}_2^{(1)} (z,w) \stackrel{def}{=} 6\,w^5 -

800: zw=0.

801: \end{equation}

802: Following the analysis of the section 5 we get that

803: $r=1/4,\,\omega=1$, $z\rightarrow\infty$ and that there are four

804: families of power asymptotics: $\tilde{\mathcal{F}}_2^{(1)}1$

805: \eqref{5.2}, $\tilde{\mathcal{F}}_2^{(1)}2$ \eqref{5.3},

806: $\tilde{\mathcal{F}}_2^{(1)}3$ \eqref{5.4},

807: $\tilde{\mathcal{F}}_2^{(1)}4$ \eqref{5.5}. The shifted carrier of

808: the reduced solutions is the vector $(1/4,-1)$. Together with the

809: vectors $B_3$, $B_4$ it generates the lattice with the basis

810: $(1/4,-1)$ and $(-3,2)$. The points of this lattice are

811: $Q=(q_1,q_2)=m(1/4,-1)+l(-3,2)=(m/4-3l,-m+2l)$. At the line $q_2=-1$

812: we have $q_1=1/4+l/2-3l=1/4-5l/2$. As the cone of the problem is

813: $\mathcal{K}=\{k<1/4\}$, then the set

814: \begin{equation}

815: \label{10.2}\textbf{K}=\{k=1/4-5l/2,\,\,\,l\in \mathbb{N}\}

816: \end{equation}

817: differs from the similar in the section 5. Hence the power

818: expansions can be presented in the form

819: \begin{equation}

820: \label{10.3}\tilde{G}_2^{(1)} n:\,\,\,\,

821: w(z)=\varphi^{(n)}(z)=c^{(n)}_{1/4} z^{1/4} + \sum^{\infty}_{l=1}

822: c^{(n)}_{1/4-5l/2}\, z^{1/4-5l/2}.

823: \end{equation}

824: All coefficients in the expression \eqref{10.3} can be sequently

825: computed. Taking into account three terms these expansions can be

826: rewritten as

827: \begin{equation*}

828: \begin{gathered}

829: \label{}w(z)=\varphi^{(n)}(z)=c_{{1/4}}^{(n)}z^{1/4}-{\frac

830: {5}{16}}\,{\frac {(c_{{1/4}}^{(n)})^{3}}{{z}^{9 /4}}}-{\frac

831: {297}{512}}\,\frac{c_{{1/4}}^{(n)}}{{z}^{{19/4}}}+...

832: \end{gathered}

833: \end{equation*}

834: It is likely that obtained expressions are divergent ones.

835: 

836: \section{Non-power expansions, corresponding to the edge $\Gamma_3^{(1)}$ in the case $\alpha =0$.}

837: 

838: The reduced equation which corresponds to the edge $\Gamma^{(1)}_3$

839: \begin{equation}

840: \label{11.1}\hat{\tilde{f}}_2^{(3)} (z,w) \stackrel{def}{=}

841: w_{zzzz}-zw=0

842: \end{equation}

843: does not possess solutions in the form $w=c_{r}z^{r},\,c_r\neq0$ (it

844: has a trivial solution $w\equiv 0$ ). The edge $\Gamma^{(1)}_3$

845: defines non-power asymptotics of the equation \eqref{8.1}. It is

846: horizontal, consequently in order to find the solutions of the

847: reduced equation \eqref{11.1} it is necessary to make the

848: logarithmic transformation

849: \begin{equation}

850: \label{11.2}y=\frac{d \ln w}{dz}.

851: \end{equation}

852: Hence derivatives of $w(z)$ are

853: \begin{equation*}

854: \begin{gathered}

855: \label{}w'=yw,\,\,\,w''=(y'+y^2)w,\,\,\,w'''=(y''+3yy'+y^3)w,\\

856: w''''=(y'''+4yy''+3(y')^2+6y^2y'+y^4)w.

857: \end{gathered}

858: \end{equation*}

859: After application of this transformation and after cancelation of

860: the result by $w$  the equation \eqref{11.1} can be rewritten as

861: \begin{equation}

862: \label{11.3}h(z,y)\stackrel{def}{=}y'''+4yy''+3(y')^2+6y^2y'+y^4-z=0.

863: \end{equation}

864: The carrier $S(h)$ of this equation consists of five points:

865: $M_1=(-3,1),\,M_2=(0,4),\,M_3=(1,0),\,M_4=(-2,2)$ è $M_5=(-1,3)$.

866: Their convex hull $\Gamma(h)$ is the triangle with the apexes

867: $\tilde{\Gamma}_j^{(0)}=M_j,\,\,\,j=1,2,3$ and the edges

868: $\tilde{\Gamma}_1^{(1)}=\{M_1,M_2\}\,\,\,\tilde{\Gamma}_2^{(1)}=\{M_2,M_3\}\,\,\,\tilde{\Gamma}_3^{(1)}=\{M_3,M_1\}$

869: (fig. 5).

870: \begin{figure}[h]

871:  \centerline{\epsfig{file=Fig5.eps,width=100mm}}

872:  \caption{}\label{fig:z_post}

873: \end{figure}

874: 

875: The external normals $\tilde{N}_j$ to the edges

876: $\tilde{\Gamma}_j^{(1)}$ are

877: $\tilde{N}_1=(-1,1),\,\tilde{N}_2=(4,1),\,\tilde{N}_3=(-1,-4)$. The

878: normal cones of the apexes and of the edges are presented at the

879: fig. 6. The shifted carrier of the equation \eqref{11.3} lies in the

880: lattice with the basis $B_5=(1,1),\,B_6=(1,-4)$. In this case the

881: cone of the problem is $p_1+p_2>0$. It intersects with the normal

882: cones $\tilde{U}_2^{(0)},\,\tilde{U}_3^{(0)},\,\tilde{U}_2^{(1)}$.

883: That is why later it is sufficient to study the reduced equations

884: which correspond to the bounds

885: $\tilde{\Gamma}_2^{(0)},\,\tilde{\Gamma}_3^{(0)},\,\tilde{\Gamma}_2^{(1)}$

886: only.

887: \begin{figure}[h]

888:  \centerline{\epsfig{file=Fig6.eps,width=100mm}}

889:  \caption{}\label{fig:z_post}

890: \end{figure}

891: The first two of them are characterized by the algebraic reduced

892: equations $y^4=0,\,-z=0$, accordingly. They do not give suitable

893: solutions. The edge $\tilde{\Gamma}_2^{(1)}$ defines algebraic

894: reduced equation

895: \begin{equation}

896: \label{11.4}y^4-z=0,

897: \end{equation}

898: which has four suitable solutions $y^{(n)}(z)=c_{1/4}^{(n)} z^{1/4}$

899: where $c_{1/4}^{(n=\{1,2\})}=\pm1,\,c_{1/4}^{(n=\{3,4\})}=\pm i$ and

900: $r=1/4,\,\omega=1$. These solutions do not have critical numbers.

901: The shifted carrier of the equation \eqref{11.3} is the vector

902: $(1/4,-1)=1/4B_6$. Thus we have a lattice with the basis

903: $(1/4,-1),\,(1,1)$. Taking into account the fact that the cone of

904: the problem is $ \mathcal{K}=\{k<1/4\}$, we find the set

905: $\textbf{K}$

906: \begin{equation}

907: \label{11.5}\textbf{K}=\{k=1/4-5l/4,\,\,\,l\in \mathbb{N}\}.

908: \end{equation}

909: The equation \eqref{11.3} has solutions in the form of the series

910: \begin{equation}

911: \label{11.6}y^{(n)}(z)=c^{(n)}_{1/4} z^{1/4} + \sum^{\infty}_{l=1}

912: c^{(n)}_{1/4-5l/4}\, z^{1/4-5l/4},\,\,\,n=1,2,3,4,

913: \end{equation}

914: where the coefficients $c^{(n)}_{1/4-5l/4}$ are uniquely defined.

915: The coefficient $c^{(n)}_{-1}$ does not depend on $n$ and

916: $c^{(n)}_{-1}=-3/8$. Introducing the new variable $\eta=\ln w$ we

917: get

918: \begin{equation}

919: \label{11.7}\eta =\int ydz=

920: \frac45c^{(n)}_{1/4}z^{5/4}+C_0-\frac38\ln z+\sum^{\infty}

921: _{l=2}\frac{4c^{(n)}_{1/4-5l/4}}{5(1-l)}z^{5(1-l)/4}.

922: \end{equation}

923: In this expression $C_0$ is a constant of integration. Returning to

924: the variable $w(z)$ yields

925: \begin{equation}

926: \begin{gathered}

927: \label{11.8}w(z)=\frac{C_1}{z^{3/8}}\exp\left[\frac45c^{(n)}_{1/4}z^{5/4}+\sum^{\infty}

928: _{l=2}\frac{4c^{(n)}_{1/4-5l/4}}{5(1-l)}z^{5(1-l)/4}\right].

929: \end{gathered}

930: \end{equation}

931: Taking into account two terms of the series the obtained non-power

932: asymptotics of the equation \eqref{8.1} can be written as

933: \begin{equation*}

934: \begin{gathered}

935: \label{}w(z)=\frac{C_1}{z^{3/8}}\exp\left[\frac45c^{(n)}_{1/4}z^{5/4}+{\frac

936: {9}{32}}\,{\frac {1}{c^{(n)}_{1/4}{z}^{5/4} }}+{\frac

937: {45}{256}}\,\frac {1}{(c^{(n)}_{1/4})^{2}{z}^{5/2}}+... \right].

938: \end{gathered}

939: \end{equation*}

940: Providing that $w\rightarrow 0$ at $z\rightarrow\infty$, we should

941: take the number $n$ in \eqref{11.8} so that

942: \begin{equation}

943: \label{11.9}Re(c^{(n)}_{1/4}z^{5/4})<0.

944: \end{equation}

945: Consequently we have found four one-parametric non-power asymptotics

946: for the solutions of the equation \eqref{8.1}:

947: $\tilde{G}_3^{(1)}1,\,\tilde{G}_3^{(1)}2,\,\tilde{G}_3^{(1)}3,\,\tilde{G}_3^{(1)}4$.

948: 

949: 

950: \section{Exponential additions to the solutions.}

951: 

952: Let us find exponential additions to the obtained solutions. Since

953: the equations $\partial\hat{f}^{(d)}_{j}/\partial w^{\Pi(f)}=0$ do

954: not have solutions for $\forall (d=0,1; j=1,2,3,4)$ and the

955: condition $\Pi(\hat{f}^{(d)}_{j}) < \Pi(f)=4$ holds for the edges

956: $\Gamma^{(1)}_2$ and $\Gamma^{(1)}_3$ only (in the case

957: $\alpha\neq0$) then these very edges define exponential additions

958: which are to be computed. At $\alpha=0$ the condition

959: $\Pi(\hat{\tilde{f}}^{(d)}_{j}) < \Pi(\tilde{f})=4$ is true for the

960: edge $\Gamma^{(1)}_2$, consequently solutions corresponding to this

961: edge possess exponential additions which can be calculated as it

962: will be done for the edge $\Gamma^{(1)}_2$ of the equation

963: \eqref{1.3}.

964: 

965: 

966: \section{Exponential additions of the first level, corresponding to the edge $\Gamma^{(1)}_2$ at $\alpha\neq0$.}

967: 

968: To begin with we will look for the exponential additions of the

969: first level $u^{(n)}(z)$ to the expansions \eqref{5.7}, i.e. we will

970: look for the functions

971: \begin{equation}

972: \label{13.1} w(z)=\varphi^{(n)}(z) + u^{(n)}(z),\,\,\, n=1,2,3,4.

973: \end{equation}

974: The reduced equation for the addition $u^{(n)}(z)$ is a linear

975: equation

976: \begin{equation}

977: \label{13.2}M_{n}^{(1)}(z) u^{(n)}(z)=0,

978: \end{equation}

979: where $M_{n}^{(1)}(z)$ is the first variation at the solutions

980: $w(z)=\varphi^{(n)}(z)$. As long as

981: \begin{equation}

982: \label{13.3}\frac{\delta f}{\delta w} =\frac{d^4}{dz^4}

983: -10w^{2}\frac{d^2}{dz^2} -20ww_z \frac{d}{dz}- 20 w_{zz}w

984: -10w_z^{2}+30w^4-z,

985: \end{equation}

986: then

987: \begin{equation}

988: \label{13.4}M_{n}^{(1)}(z) =\frac{d^4}{dz^4}

989:  -10(\varphi^{(n)})^{2}\frac{d^2}{dz^2}

990: -20\varphi^{(n)}\varphi^{(n)}_z \frac

991: d{dz}-20\varphi^{(n)}_{zz}\varphi^{(n)} -10(\varphi^{(n)}_z)^{2}+

992: 30(\varphi^{(n)})^4 -z.

993: \end{equation}

994: And equation \eqref{13.2} can be rewritten as

995: \begin{equation}

996: \begin{gathered} \label{13.5}\frac{d^4u^{(n)}}{dz^4}  -10

997: (\varphi{(n)})^{2} \frac{d^2u^{(n)}}{dz^2}-20

998: \varphi^{(n)}\varphi^{(n)} _z \frac{du^{(n)}}{dz}-\\

999: -[10(2\varphi^{(n)}_{zz}\varphi^{(n)} +(\varphi^{(n)}_z)^{2}

1000: -3(\varphi^{(n)})^4) +z]\,u^{(n)}=0,\,\,\,

1001:  n=1,2,3,4.

1002: \end{gathered}

1003: \end{equation}

1004: Let us make the change of the variables

1005: \begin{equation}

1006: \label{13.6}\zeta^{(n)}(z)=\frac{d \ln u^{(n)}(z)}{dz}.

1007: \end{equation}

1008: Computing the derivatives of the function $u^{(n)}(z)$ yields

1009: \begin{equation*} \label{}\frac{du^{(n)}}{dz} =\zeta^{(n)} u^{(n)},

1010: \,\,\quad\, \frac{d^2u^{(n)}}{dz^2 }=\zeta _z^{(n)} u^{(n)} + (\zeta

1011: ^{(n)})^2 u^{(n)},

1012: \end{equation*}

1013: \begin{equation*}

1014: \label{}\frac{d^3u^{(n)}}{dz^3} =\zeta^{(n)}_{zz} u^{(n)}+ 3 \zeta

1015: ^{(n)} \zeta_z^{(n)} u^{(n)} + (\zeta^{(n)})^3 u^{(n)},

1016: \end{equation*}

1017: \begin{equation*}

1018: \label{}\frac{d^4u^{(n)}}{dz^4} =\zeta^{(n)}_{zzz} u^{(n)}+ 4 \zeta

1019: ^{(n)}\zeta^{(n)}_{zz} u^{(n)} + 3 (\zeta_z^{(n)})^2  u^{(n)} + 6

1020: (\zeta^{(n)})^2 \zeta _z^{(n)} u^{(n)} + (\zeta ^{(n)})^4 u^{(n)}.

1021: \end{equation*}

1022: Substituting these expressions into the equation \eqref{13.5} and

1023: setting to zero the factor at $u^{(n)}(z)$ we get the equation

1024: \begin{equation}

1025: \begin{gathered}

1026: \label{13.7} \zeta^{(n)}_{zzz} + 4\zeta^{(n)}\zeta^{(n)}_{zz}

1027: +3(\zeta^{(n)}_z)^2 +6 (\zeta^{(n)})^2

1028: \zeta^{(n)}_z +  (\zeta^{(n)})^4 -20\varphi_z^{(n)}\varphi_{zz}^{(n)} +\\

1029:   -10(\varphi^{(n)})^2( \zeta^{(n)}_z + (\zeta^{(n)})^2)

1030: -20\varphi^{(n)}\varphi_z^{(n)} \zeta^{(n)}

1031: -10(\varphi_z^{(n)})^2+30 (\varphi^{(n)})^4 -z=0.

1032: \end{gathered}

1033: \end{equation}

1034: It is essential to find power expansions of its solutions. The

1035: carrier of the equation \eqref{13.7} consists of the points

1036: \begin{equation}

1037: \begin{gathered}

1038: \label{13.8}Q_{1,1} =(-3,1),\,\,\, Q_{1,2}=(-2,2),\,\,\, Q_{1,3}=(-1,3), \\

1039: Q_{1,4}=(0,4), \,\,\, Q_{1,5}=(1,0), \,\,\,Q_{2,\,k}=\left(\frac14-\frac{5}4k,0\right),\,\,\, \\

1040: Q_{3,\,k}=\left(-\frac12-\frac{5}4k,1\right),\,\,

1041: Q_{4,\,k}=\left(\frac12-\frac{5}4k,2\right),\,\,

1042: Q_{5,\,k}=\left(-\frac32-\frac{5}4k,0\right),\\

1043: Q_{6,\,k}=\left(1-\frac{5}4k,0\right),\,\, \,\,\ k \in

1044: \mathbb{N}\cup\{0\},

1045: \end{gathered}

1046: \end{equation}

1047: where the doubled index in the points numeration is introduced for

1048: the notational convenience. The convex closing of the points

1049: \eqref{13.8} is the half-string presented at the fig. 7.

1050: \begin{figure}[h]

1051:  \centerline{\epsfig{file=Fig7.eps,width=110mm}}

1052:  \caption{}\label{fig:z_post}

1053: \end{figure}

1054: The periphery of the half-string contains two apexes

1055: $Q_{1,\,4},Q_{1,\,5}$ and three edges $\Gamma^{(1)}_j\,\,(j=1,2,3)$

1056: with the normal vectors $N_1=(4,1),\,\, N_2=(0,-1),\,\, N_3=(0,1)$.

1057: Suitable expansions can be obtained examining the reduced equation

1058: corresponding to the edge $\Gamma^{(1)}_1$

1059: \begin{equation}

1060: \label{13.9}h_1^{(1)}(z,\zeta^{(n)}) \stackrel{def}{=}

1061: (\zeta^{(n)})^4 -10(c_{1/4}^{(n)})^2 (\zeta^{(n)})^2z^{\frac12} +

1062: (30(c_{1/4}^{(n)})^4-1)z=0.

1063: \end{equation}

1064: It has sixteen solutions

1065: \begin{equation}\begin{gathered}

1066: \label{13.10}\zeta^{(n,\,l)}(z) =g_{1/4}^{(n,\,l)}z^{1/4},\,\,\,

1067: n,l=1,2,3,4\,\,\,,

1068: \end{gathered}

1069: \end{equation}

1070: where

1071: \begin{equation*}\begin{gathered}

1072: \label{}g_{1/4}^{(n;\,\,\,l=\{1,2,3,4\})}=\pm\left(\,\,\,5(c_{1/4}^{(n)})^2\pm\sqrt{1-5(c_{1/4}^{(n)})^4}\,\,\,\right)^{1/2}.

1073: \end{gathered}\end{equation*}

1074: Taking into account the fact that $(c_{1/4}^{(n)})^4=1/6$, we get

1075: \begin{equation*}\begin{gathered}

1076: \label{}g_{1/4}^{(n;\,\,\,l=\{1,2\})}=\pm6^{1/4},n=1,2\,\,\,;\,g_{1/4}^{(n;\,\,\,l=\{3,4\})}=\pm(8/3)^{1/4},n=1,2;\\

1077: g_{1/4}^{(n;\,\,\,l=\{1,2\})}=\pm6^{1/4}i,n=3,4\,\,\,;\,g_{1/4}^{(n;\,\,\,l=\{3,4\})}=\pm(8/3)^{1/4}i,n=3,4.

1078: \end{gathered}\end{equation*}

1079: 

1080: The reduced equation is an algebraic one that is why it does not

1081: have critical numbers. Now we will look for the power expansions of

1082: the equation \eqref{13.7} solutions which possess the power

1083: asymptotic \eqref{13.10}. The shifted carrier of the equation

1084: \eqref{13.7} lies in the lattice generated by the vectors

1085: $B_7=\left(5/4,0\right),\,\, B_8=(1,1)$. The shifted carrier of the

1086: reduced solutions \eqref{13.10} gives the vector $B_9=\left(-1/4,1

1087: \right)$. Taking into consideration the correlation

1088: $B_8-B_9=\left(5/4,0\right)= B_7$ we get that the vector $B_9$

1089: belongs to the lattice with the basis $B_7,\,B_8$. The points of

1090: this lattice are the following

1091: \begin{equation*}

1092: Q=(q_1,q_2)=k(1,1)

1093: +m\left(\frac54,0\right)=\left(k+\frac{5m}4,k\right).

1094: \end{equation*}

1095: At the line $q_2=-1$ we have $k=-1$ and hence $q_1=-1+5m/4$. Since

1096: the cone of the problem is $\mathcal{K}=\left\{k<\frac14\right\}$,

1097: then the set $\mathbf{K}$ can be presented in the form

1098: \begin{equation}

1099: \label{13.11}\mathbf{K}=\left\{\frac{1-5k}4,k\in\mathbb{N}\right\}.

1100: \end{equation}

1101: Power expansions of the equation \eqref{13.7} solutions are

1102: \begin{equation}\begin{gathered}

1103: \label{13.12}\zeta^{(n,\,l)}(z)=g^{(n,\,l)}_{1/4} z^{1/4}

1104: +\sum_{k\,\in\,\,\mathbb{N}}g^{(n,\,l)}_{(1-5k)/4}\,z^{(1-5k)/4},\,\,\,

1105:  n,l=1,2,3,4.

1106: \end{gathered}\end{equation}

1107: Using the values of the coefficients $c_{1/4}^{(n)}$, $c_{-1}^{(n)}$

1108: and $g^{(n,\,l)}_{1/4}$ we can compute the coefficient

1109: $g^{(n,\,l)}_{-1}$.

1110: \begin{equation*}

1111: \begin{gathered}

1112: \label{}

1113: g_{-1}^{(n,\,l)}=-\frac38,n=1,2,3,4,\,l=1,2;\,\,\,g_{-1}^{(n,\,l)}=-\frac38+\frac54\alpha,(n,\,l)=\{(1,3),(2,4),(3,3),(4,4)\};\,\,\,\\

1114: g_{-1}^{(n,\,l)}=-\frac38-\frac54\alpha,(n,\,l)=\{(1,4),(2,3),(3,4),(4,3)\}.

1115: \end{gathered}

1116: \end{equation*}

1117: In view of the transformation \eqref{13.6} the additions

1118: $u^{(n,\,l)}(z)$ can be found. They are

1119: \begin{equation}

1120: \label{13.13}u^{(n,\,l)}(z) = C_1 \exp \int \zeta^{(n,\,l)}(z) dz.

1121: \end{equation}

1122: Where from we get

1123: \begin{equation}

1124: \begin{gathered}

1125: \label{13.14}u^{(n,\,l)}(z)=C_1\,z^{g^{(n,\,l)}_{-1}}\, \exp

1126: \left[\frac45\, g^{(n,\,l)}_{1/4}\,z^{5/4} + \sum^{\infty}_{k=2}

1127: \frac{4}{5(1-k)} g^{(n,\,l)}_{(1-5k)/4}

1128: z^{5(1-k)/4}\right],\\

1129: \,\, n,l=1,2,3,4.

1130: \end{gathered}

1131: \end{equation}

1132: Here $C_1$ and later $C_2$ and $C_3$ are the constants of

1133: integration. The additions $u^{(n,\,l)}(z)$ are exponentially small

1134: at $z\rightarrow \infty$ in those sectors of the complex plane $z$,

1135: where

1136: \begin{equation}

1137: \label{13.15}Re \left[g^{(n,\,l)}_{1/4}\, z^{5/4}\right]<0.

1138: \end{equation}

1139: 

1140: Consequently for each expansion $G^{(1)}_{3}n$ we have found four

1141: one-parametric families of additions $G_2^{(1)}n\, G^{(1)}_1 l$

1142: ($n,l=1,2,3,4$).

1143: 

1144: 

1145: \section{Exponential additions of the second level, corresponding to the edge $\Gamma^{(1)}_2$ at $\alpha\neq0$.}

1146: 

1147: In this section we will compute the exponential additions of the

1148: second level $v^{(n,\,l)}(z)$, i.e. the additions to the solutions

1149: $\zeta^{(n,\,l)}(z)$. The reduced equation to the addition

1150: $v^{(n,\,l)}(z)$ is

1151: \begin{equation}

1152: \label{14.1}M_{n,\,l}^{(2)} (z) v^{(n,\,l)}(z)=0,

1153: \end{equation}

1154: where an operator $M_{n,\,l}^{(2)}(z)$ is the first variation of

1155: \eqref{13.7} at the solutions $\zeta^{(n,\,l)}(z)$. The equation

1156: \eqref{14.1} can be rewritten as

1157: \begin{equation}

1158: \begin{gathered}

1159: \label{14.2}\frac{d^3v}{dz^3} + 4\zeta \frac{d^2v}{dz^2} +

1160: 2(3\zeta_z + 3\zeta^{2}-5\varphi^2)\frac{dv}{dz} +  4(\zeta_{zz}

1161: +3\zeta\zeta_z + \zeta^3 - 5\varphi^2\zeta - 5\varphi\varphi_z)v=0.

1162: \end{gathered}

1163: \end{equation}

1164: Here and later $\varphi\stackrel{def}{=}\varphi^{(n)}$,

1165: $v\stackrel{def}{=} v^{(n,\,l)}$ è $\zeta\stackrel{def}{=}

1166: \zeta^{(n,\,l)}$. Making the transformation of the variables

1167: \begin{equation}

1168: \label{14.3}\frac{d \ln v}{dz}=\xi,

1169: \end{equation}

1170: we get

1171: \begin{equation*}

1172: \label{}\frac{dv}{dz}=\xi v,\,\,\quad\, \frac{d^2v}{dz^2} =\xi_z

1173: v+\xi^2 v,\,\,\quad\, \frac{d^3 v}{dz^3}=\xi_{zz} v + 3\xi\xi_{z}

1174: v+\zeta^3 v.

1175: \end{equation*}

1176: And the equation \eqref{14.2} transferees to

1177: \begin{equation}\begin{gathered}

1178: \label{14.4}\xi_{zz} + 3\, \xi\,\xi_{z} + \xi^3

1179: +4\zeta(\xi_z+\xi^2)+6(\zeta_z+\zeta^2)\xi -10\,\varphi^2\xi+

1180: 4\zeta_{zz} + 12 \zeta\zeta_z  +

1181: \\+4\,\zeta^3  -20\, \varphi^2\,\zeta

1182: -20\,\varphi\,\varphi_z=0.

1183: \end{gathered}\end{equation}

1184: The carrier of this equation is defined by the following set of

1185: points

1186: \begin{equation}

1187: \begin{gathered}

1188: \label{14.5}M_{1,\,1}=(-2,1),\,\,\, M_{1,\,2}=(-1,2),\,\,\,

1189: M_{1,\,3}=(0,3),\,\,\,

1190: M_{2,\,k}=\left(-\frac34-\frac54k,\,1\right),\,\,\,\\

1191: M_{3,\,k}=\left(\frac14-\frac54k,2\right),\,\,\,M_{4,\,k}=\left(\frac12-\frac54k,1\right),\,\,\,M_{5,\,k}=\left(-\frac74-\frac54k,\,0\right),\,\,\,\\

1192: M_{6,\,k}=\left(-\frac12-\frac54k,\,0\right),\,\,\,M_{7,\,k}=\left(\frac34-\frac54k,\,0\right),\,\,

1193: \, k \in \mathbb{N}\cup\{0\}.

1194: \end{gathered}

1195: \end{equation}

1196: Their convex hull is the half-string similar to the one presented at

1197: the fig. 7. Its edges  $\Gamma_3^{(1)}$, $\Gamma_2^{(1)}$ are the

1198: rays starting out of the points $(0,\,3)$, $(3/4,\,0)$, accordingly,

1199: and the edge $\Gamma_1^{(1)}$ connects the points $(3/4,\,0)$ è

1200: $(0,\,3)$. Examining the edge $\Gamma_1^{(1)}$ which contains the

1201: points $(0,\,3)$, $(1/4,\,2)$, $(1/2,\,1)$ and $(3/4,\,0)$ we can

1202: find suitable solutions. The reduced equation corresponding to this

1203: edge is

1204: \begin{equation}

1205: \begin{gathered}

1206: \label{14.6}\xi^3 + 4g^{(n,\,l)}_{1/4}z^{1/4}\xi^2 +

1207: 2[3(g^{(n,\,l)}_{1/4})^2-5(c_{1/4}^{(n)})^2]z^{1/2}\xi +\\

1208: +4g^{(n,\,l)}_{1/4}[(g^{(n,\,l)}_{1/4})^2 -

1209: 5(c_{1/4}^{(n)})^2]z^{3/4} =0.

1210: \end{gathered}

1211: \end{equation}

1212: The solutions of the equation \eqref{14.6} can be presented in the

1213: form

1214: \begin{equation}

1215: \begin{gathered}

1216: \label{14.7}\xi^{(n,\,l,\,m)}(z)=r_{1/4}^{(n,\,l,\,m)}\,

1217: z^{1/4},\,\,\,\, n,\,l=1,2,3,4;\,\,\,\, m=1,2,3,

1218: \end{gathered}

1219: \end{equation}

1220: where $r\stackrel{def}{=}r_{1/4}^{(n,l,m)}$ are the roots of the

1221: cubic equation

1222: \begin{equation}

1223: \begin{gathered}

1224: \label{14.8} r^3 +4\,g_{1/4}^{(n,\,l)}r^2+

1225: 2\left(3\,(g_{1/4}^{(n,\,l)})^2 -5\,(c_{1/4}^{(n)})^2\right)\,r+\\

1226: +4\,(g_{1/4}^{(n,\,l)})^{3} -20\,g_{1/4}^{(n,\,l)}\,(c_{1/4}^{(n)})^2=0.

1227: \end{gathered}

1228: \end{equation}

1229: Solving this equation we get

1230: \begin{equation*}

1231: \begin{gathered}

1232: \label{}r_{1/4}^{(n,\,l,\,1)}=-2\,g_{1/4}^{(n,\,l)},\,\quad\,

1233: r_{1/4}^{(n,\,l,\,2)}=-g_{1/4}^{(n,\,l)}+\left({10\,(c_{1/4}^{(n)})^2-(g_{1/4}^{(n,\,l)})^2}\right)^{1/2}\\

1234: r_{1/4}^{(n,\,l,\,3)}=-g_{1/4}^{(n,\,l)}-\left({10\,(c_{1/4}^{(n)})^2-(g_{1/4}^{(n,\,l)})^2}\right)^{1/2}.

1235: \end{gathered}

1236: \end{equation*}

1237: 

1238: The vectors $(1,\,1)$, $(5/4,\,0)$ are the basis of the lattice

1239: which corresponds to the shifted carrier of the equation

1240: \eqref{14.4}. The shifted carrier $(-1/4,1)$ of the reduced

1241: solutions \eqref{14.7} belongs to this lattice. Consequently the set

1242: $\mathbf{K}$ coincides with \eqref{13.11}. Power expansions of the

1243: functions $\xi^{(n,\,l,\,m)}(z)$ are the following

1244: \begin{equation}

1245: \begin{gathered}

1246: \label{14.9} \xi^{(n,\,l,\,m)}(z)=r_{1/4}^{(n,\,l,\,m)} z^{1/4}

1247: +\sum_{k\,\in\,\,\mathbb{N}}r^{(n,\,l,\,m)}_{(1-5k)/4}\,z^{(1-5k)/4},\\

1248: \,\quad\,n,l=1,2,3,4;\,\,\quad\, m=1,2,3.

1249: \end{gathered}

1250: \end{equation}

1251: Calculation of the coefficients $r_{-1}^{(n,\,l,\,m)}$ yields

1252: $r_{-1}^{(n,\,l,\,m)}=1/4$ at $n=1,2,3,4;\,\,\,l=1,2;\,\,\,m=1$. In

1253: other cases $r_{-1}^{(n,\,l,\,m)}$ depend on the parameter $\alpha$.

1254: 

1255: So we have found the exponential additions $v^{(n,\,l,\,m)}(z)$ to

1256: the solutions $\zeta^{(n,\,l)}(z)$

1257: \begin{equation}

1258: \begin{gathered}

1259: \label{14.10}v^{(n,\,l,\,m)}(z)=C_2\,z^{r_{-1}^{(n,\,l,\,m)}}\, \exp

1260: \left[\frac45\, r_{1/4}^{(n,\,l,\,m)}\,z^{5/4} + \sum^{\infty}_{k=2}

1261: \frac{4}{5(1-k)} r^{(n,\,l,\,m)}_{(1-5k)/4}

1262: z^{5(1-k)/4}\right]\\

1263: \,\, n=1,2,3,4;\,\quad\,l=1,2,3,4;\,\quad\,m=1,2,3.

1264: \end{gathered}

1265: \end{equation}

1266: They are exponentially small provided that

1267: \begin{equation}

1268: \label{14.11}Re

1269: \left[r_{1/4}^{(n,\,l,\,m)}\,z^{5/4}\right]<0,\,\,z\rightarrow\infty.

1270: \end{equation}

1271: 

1272: Thus, taking into account two-level additions $G_2^{(1)}n\,

1273: G^{(1)}_1 l\,G^{(1)}_1 m$, we have $48$ expansions of the equation

1274: \eqref{1.3} solutions ($n,l=1,2,3,4;\,\,\,m=1,2,3$).

1275: 

1276: 

1277: 

1278: \section{Exponential additions of the third level, corresponding to the edge $\Gamma^{(1)}_2$ at $\alpha\neq0$.}

1279: 

1280: In this section we will look for exponential additions of the third

1281: level $y^{(n,\,l,\,m)}(z)$, i.e. we will look for additions to the

1282: solutions $\xi^{(n,\,l,\,m)}(z)$. The reduced equation to the

1283: addition $y^{(n,\,l,\,m)}(z)$ is the following

1284: \begin{equation}

1285: \label{15.1}M_{n,\,l,\,m}^{(3)} (z) y^{(n,\,l,\,m)}=0.

1286: \end{equation}

1287: Operator $M_{n,\,l,\,m}^{(3)}(z)$ can be found as the first

1288: variation of \eqref{14.4} at the solutions $\xi^{(n,\,l,\,m)}(z)$

1289: and then the equation \eqref{15.1} for the function

1290: $y\stackrel{def}{=}y^{(n,\,l,\,m)}$ can be rewritten as

1291: \begin{equation}

1292: \begin{gathered}

1293: \label{15.2}y_{zz} + (3\xi + 4\zeta)y_z +

1294: (3\xi_z+3\xi^2+8\xi\zeta+6\zeta_z+6\zeta^2-10\varphi^2)y=0.

1295: \end{gathered}

1296: \end{equation}

1297: Introducing the new variable $\eta$ by the rule

1298: \begin{equation}

1299: \label{15.3}\frac{d \ln y}{dz}=\eta

1300: \end{equation}

1301: we get that

1302: \begin{equation*}

1303: \label{}\frac{dy}{dz}=\eta y,\,\,\quad\, \frac{d^2y}{dz^2} =\eta_z

1304: y+\eta^2 y.

1305: \end{equation*}

1306: Hence equation \eqref{15.2} transfers to the following

1307: \begin{equation}\begin{gathered}

1308: \label{15.4}\eta_{z} +\eta^2+ (3\xi +

1309: 4\zeta)\eta+3\xi_z+3\xi^2+8\xi\zeta+6\zeta_z+6\zeta^2-10\varphi^2=0.

1310: \end{gathered}\end{equation}

1311: The carrier of this equation is composed of the points

1312: \begin{equation}

1313: \begin{gathered}

1314: \label{15.5}R_{1,\,1}=(-1,\,1),\,\,\, R_{1,\,2}=(0,\,2),\,\,\, R_{2,\,k}=(\frac14-\frac54k,\,1),\,\,\, R_{3,\,k}=\left(-\frac34-\frac54k,\,0\right),\\

1315: R_{4,\,k}=\left(\frac12-\frac54k,\,0\right),\,\, \, k \in

1316: \mathbb{N}\cup\{0\}.

1317: \end{gathered}

1318: \end{equation}

1319: Closing the convex hull based on these points we obtain the

1320: half-string similar to the one presented at fig. 7. Its edges

1321: $\Gamma_3^{(1)}$ and $\Gamma_2^{(1)}$ are the rays starting out of

1322: the points $(0,\,2)$ and $(1/2,\,0)$, accordingly. The edge

1323: $\Gamma_1^{(1)}$ is limited by the points $(0,\,2)$ and $(1/2,\,0)$.

1324: Now we will examine the reduced equation

1325: \begin{equation}

1326: \begin{gathered}

1327: \label{15.6}\eta^2+

1328: (3r_{1/4}^{(n,\,l,\,m)}+4g_{1/4}^{(n,\,l)})z^{1/4}\eta+

1329: \left(3(r_{1/4}^{(n,\,l,\,m)})^2+8r_{1/4}^{(n,\,l,\,m)}g_{1/4}^{(n,\,l)}+\right.\\

1330: \left. 6(g_{1/4}^{(n,\,l)})^2-10(c_{1/4}^{(n)})^2 \right)z^{1/2}=0,

1331: \end{gathered}

1332: \end{equation}

1333: which corresponds to the edge $\Gamma_1^{(1)}$ (this edge contains

1334: three points $(0,\,2)$, $(1/4,\,1)$ and $(1/2,\,0)$). The equation

1335: \eqref{15.6} possesses the following solutions

1336: \begin{equation}

1337: \begin{gathered}

1338: \label{15.7}\eta^{(n,\,l,\,m,\,p)}(z)=q^{(n,\,l,\,m,\,p)}_{1/4}\,

1339: z^{1/4};\,\,\, n,l=1,2,3,4;\,\,\,\, m=1,2,3; \,\,\,\,p=1,2\,

1340: \end{gathered}

1341: \end{equation}

1342: where $q\stackrel{def}{=}q^{(n,\,l,\,m,\,p)}_{1/4}$ are the

1343: solutions of the quadratic equation

1344: \begin{equation}

1345: \begin{gathered}

1346: \label{15.8}q^2+ (3r_{1/4}^{(n,\,l,\,m)}+4g_{1/4}^{(n,\,l)})q+

1347: 3(r_{1/4}^{(n,\,l,\,m)})^2+8r_{1/4}^{(n,\,l,\,m)}g_{1/4}^{(n,\,l)}+\\

1348: 6(g_{1/4}^{(n,\,l)})^2-10(c_{1/4}^{(n)})^2 =0.

1349: \end{gathered}

1350: \end{equation}

1351: At fixed $n,\,l,\,m$ it has two solutions

1352: \begin{equation*}

1353: \begin{gathered}

1354: \label{}

1355: q_{1/4}^{(n,\,l,\,m,\,p=\{1,\,2\})}=-\frac32\,r_{1/4}^{(n,\,l,\,m)}-2\,g_{1/4}^{(n,\,l)}\pm\\

1356: \pm\,\frac12\left({40\,(c_{1/4}^{(n)})^2-3\,(r_{1/4}^{(n,\,l,\,m)})^{2}-

1357: -8\,{r_{1/4}^{(n,\,l,\,m)}}\,g_{1/4}^{(n,\,l)}-8\,

1358: (g_{1/4}^{(n,\,l)})^{2}}\right)^{(1/2)}.

1359: \end{gathered}

1360: \end{equation*}

1361: The basis of the lattice defined by the shifted carrier of the

1362: equation \eqref{15.4} is composed of the vectors $(1,\,1)$,

1363: $(5/4,\,0)$. The set $\mathbf{K}$ coincides with \eqref{13.11}.

1364: Power expansions for $\eta^{(n,\,l,\,m,\,p)}(z)$ can be presented as

1365: \begin{equation}

1366: \begin{gathered}

1367: \label{15.9}\eta^{(n,\,l,\,m,\,p)}(z)=q^{(n,\,l,\,m,\,p)}_{1/4}

1368: z^{1/4}

1369: +\sum_{k\,\in\,\,\mathbb{N}}q^{(n,\,l,\,m,\,p)}_{(1-5k)/4}\,z^{(1-5k)/4},\\

1370: n,l=1,2,3,4;\,\, m=1,2,3;\,\, p=1,2.

1371: \end{gathered}

1372: \end{equation}

1373: It can be shown that the coefficients $q^{(n,\,l,\,m,\,p)}_{-1}$

1374: depend on $\alpha$. The exponential additions

1375: $y^{(n,\,l,\,m,\,p)}(z)$ to the solutions $\xi^{(n,\,l,\,m)}(z)$ can

1376: be written as

1377: \begin{equation}

1378: \begin{gathered}

1379: \label{15.10}y^{(n,\,l,\,m,\,p)}(z)=C_3\,z^{q_{-1}^{(n,\,l,\,m,\,p)}}

1380: \exp \left[\frac45\, q^{(n,\,l,\,m,\,p)}_{1/4}\,z^{5/4} +

1381: \sum^{\infty}_{k=2} \frac{4}{5(1-k)} q^{(n,\,l,\,m,\,p)}_{(1-5k)/4}

1382: z^{5(1-k)/4}\right],\\

1383: n,l=1,2,3,4;\,\quad\, m=1,2,3;\,\quad\, p=1,2.

1384: \end{gathered}

1385: \end{equation}

1386: They are exponentially small when

1387: \begin{equation}

1388: \label{15.11}Re

1389: \left[q_{1/4}^{(n,\,l,\,m,\,p)}\,z^{5/4}\right]<0,\,\,\,z\rightarrow\infty.

1390: \end{equation}

1391: 

1392: Thus for the solution expansions of the studied equation near the

1393: point $z=\infty$ three-level exponential additions have been found.

1394: Taking into account exponential additions, the solutions $w(z)$ at

1395: $z\rightarrow\infty$ can be written as

1396: \begin{equation*}

1397: \begin{gathered}

1398: \label{}w(z)=\varphi^{(n)}(z)+\exp\left[\int dz \left(

1399: \zeta^{(n,\,l)}(z)+\exp \left[\int

1400: dz\left(\xi^{(n,\,l,\,m)}(z)+\right. \right.\right.\right.\\

1401: \left. \left. \left. \left. \exp\left(\int dz\,\eta^{(n,\,l,\,m,\,p)}(z)\right)\right)\right]\right)\right]

1402: \end{gathered}

1403: \end{equation*}

1404: or as

1405: \begin{equation}

1406: \begin{gathered}

1407: \label{15.12}w(z)=\varphi^{(n)}(z)+\exp\left[u^{(n,\,l)}(z)+\int dz

1408: \,\exp \left[v^{(n,\,l,\,m)}(z)+\right.\right.\\

1409: \left. \left. +\int dz\,\exp\left(y^{(n,\,l,\,m,\,p)}(z)\right)\right]\right],\\

1410: n,l=1,2,3,4;\,\,\,m=1,2,3;\,\,\,p=1,2.

1411: \end{gathered}

1412: \end{equation}

1413: 

1414: Denote these solutions (in view of three-level additions) as

1415: $G_2^{(1)}n\, G^{(1)}_1 l\,G^{(1)}_1 m\,G^{(1)}_1 p$ where

1416: $n,l=1,2,3,4;\,\,\,m=1,2,3;\,\,\,p=1,2$.

1417: 

1418: 

1419: \section{Exponential additions of the first level, corresponding to the edge $\Gamma^{(1)}_3$ at $\alpha\neq0$.}

1420: 

1421: Let us find the exponential addition of the first level $u(z)$ to

1422: the expansion \eqref{6.5}, i.e. we will look for the solutions in

1423: the form

1424: \begin{equation}

1425: \label{16.1} w(z)=\psi(z) + u(z).

1426: \end{equation}

1427: Everything written in section 13 up to the formula \eqref{13.7} is

1428: true in the case of the edge $\Gamma^{(1)}_3$, taking into account

1429: the only fact that instead of four expansions $\varphi^{(n)}(z)$ we

1430: have only one $\psi(z)$. Thus we obtain the equation

1431: \begin{equation}

1432: \begin{gathered}

1433: \label{16.2} \zeta_{zzz} + 4\zeta\zeta_{zz} +3(\zeta_z)^2 +6 \zeta^2

1434: \zeta_z +  \zeta^4 -10\psi^2( \zeta_z + \zeta^2)-\\

1435: -20\psi\psi_z\zeta-20\psi_z\psi_{zz}-10(\psi_z)^2+30 \psi^4 -z=0.

1436: \end{gathered}

1437: \end{equation}

1438: Its carrier consists of the points

1439: \begin{equation}

1440: \begin{gathered}

1441: \label{16.3}Q_{1,\,1} =(-3,1),\,\,\, Q_{1,\,2}=(-2,2),\,\,\, Q_{1,\,3}=(-1,3),\,\,\, Q_{1,\,4}=(0,4),\\

1442: Q_{1,\,5}=(1,0),

1443: \,\,\,Q_{2,\,k}=\left(-3-5k,1\right),\,\,\,Q_{3,\,k}=\left(-2-5k,2\right),\,\,\\

1444: Q_{4,\,k}=\left(-4-5k,0\right),\,\,\,k \in \mathbb{N}\cup\{0\}.

1445: \end{gathered}

1446: \end{equation}

1447: Convex closing of these points yields the half-string presented at

1448: fig. 8.

1449: \begin{figure}[h]

1450:  \centerline{\epsfig{file=Fig8.eps,width=110mm}}

1451:  \caption{}\label{fig:z_post}

1452: \end{figure}

1453: Its periphery is composed of two apexes $Q_{1,\,4}$, $Q_{1,\,5}$ and

1454: of three edges $\Gamma^{(1)}_j\,\,(j=1,2,3)$ with the normal vectors

1455: $N_1=(4,1),\,\, N_2=(0,-1),\,\, N_3=(0,1)$. The sufficient solutions

1456: are given by the edge $\Gamma^{(1)}_1$. It is characterized by the

1457: following reduced equation

1458: \begin{equation}

1459: \label{16.4}h_1^{(1)}(z,\zeta) \stackrel{def}{=} \zeta^4 -z=0.

1460: \end{equation}

1461: Equation \eqref{16.4} has four solutions

1462: \begin{equation}\begin{gathered}

1463: \label{16.5}\zeta^{(l)}(z)

1464: =g_{1/4}^{(l)}z^{1/4},\,\,\,l=1,2,3,4;\,\,\,g_{1/4}^{(l=\{1,\,2,\,3,\,4\})}=1,-1,i,-i.

1465: \end{gathered}

1466: \end{equation}

1467: The shifted carrier of the equation \eqref{16.2} lies in the lattice

1468: with the basis $(-1,4),\,(-2,3)$. Together with the shifted carrier

1469: of the reduced solutions \eqref{16.5} they generate the new lattice

1470: with the basis $(-1/4,1),\,(-2,3)$. The cone of the problem is

1471: $\mathcal{K}=\left\{k<\frac14\right\}$, then we get

1472: \begin{equation}

1473: \label{16.6}\mathbf{K}=\left\{\frac{1-5k}4,k\in\mathbb{N}\right\}.

1474: \end{equation}

1475: Power expansions for the reduced solutions \eqref{16.5} can be

1476: written as

1477: \begin{equation}\begin{gathered}

1478: \label{16.7}\zeta^{(l)}(z)=g^{(l)}_{1/4} z^{1/4}

1479: +\sum_{k\,\in\,\,\mathbb{N}}g^{(l)}_{(1-5k)/4}\,z^{(1-5k)/4},\,\,\,

1480:  l=1,2,3,4.

1481: \end{gathered}\end{equation}

1482: In this expression $g^{(l)}_{-1}=-3/8$. Using the formula

1483: \eqref{13.13} we get exponential additions

1484: \begin{equation}

1485: \begin{gathered}

1486: \label{16.8}u^{(l)}(z)=\frac{\tilde{C}_1}{z^{3/8}}\, \exp

1487: \left[\frac45\, g^{(l)}_{1/4}\,z^{5/4} + \sum^{\infty}_{k=2}

1488: \frac{4}{5(1-k)} g^{(l)}_{(1-5k)/4} z^{5(1-k)/4}\right], \,\,

1489: l=1,2,3,4.

1490: \end{gathered}

1491: \end{equation}

1492: Denote them as $G_3^{(1)}\, G^{(1)}_1 l$. Here and later

1493: $\tilde{C}_1,\,\tilde{C}_2,\,\tilde{C}_3$ are the arbitrary

1494: constants. Taking into consideration first two members of the series

1495: in \eqref{16.8} we can rewrite this expression in the form

1496: \begin{equation*}

1497: \begin{gathered}

1498: \label{}u^{(l)}(z)=\frac{\tilde{C}_1}{z^{3/8}}\, \exp

1499: \left[\frac45\,g^{(l)}_{1/4}{z}^{5/4}-\,\frac

1500: {64\,\alpha^2-9}{32g^{(l)}_{1/4}{z}^{5 /4}}-\,\frac

1501: {5(64\,\alpha^2-9)}{256(g^{(l)}_{1/4})^2 z^{5/2}}+...\right], \,\,

1502: l=1,2,3,4.

1503: \end{gathered}

1504: \end{equation*}

1505: The applicability condition \eqref{13.15} holds for the additions

1506: \eqref{16.8} provided that $g^{(n,\,l)}_{1/4}$ replace

1507: $g^{(l)}_{1/4}$.

1508: 

1509: 

1510: \section{Exponential additions of the second level, corresponding to the edge $\Gamma^{(1)}_3$ at $\alpha\neq0$.}

1511: 

1512: Now we will look for exponential additions of the second level

1513: $v^{(l)}(z)$, i.e. we will look for the additions to the solutions

1514: $\zeta^{(l)}(z)$. All transformations of the section 14, which

1515: converted the equation \eqref{14.1} to the equation \eqref{14.4} do

1516: not change under the condition $v\stackrel{def}{=} v^{(l)}$ è

1517: $\zeta\stackrel{def}{=} \zeta^{(l)}$. As a result we get the

1518: equation

1519: \begin{equation}

1520: \begin{gathered}

1521: \label{17.1}\xi_{zz} + 3\, \xi\,\xi_{z} + \xi^3

1522: +4\zeta(\xi_z+\xi^2)+6(\zeta_z+\zeta^2)\xi -10\,\psi^2\xi+

1523: 4\zeta_{zz} + 12 \zeta\zeta_z  +

1524: \\+4\,\zeta^3  -20\, \psi^2\,\zeta

1525: -20\,\psi\,\psi_z=0.

1526: \end{gathered}

1527: \end{equation}

1528: The carrier of this equation differs from the carrier of the

1529: equation \eqref{14.4}:

1530: \begin{equation}

1531: \begin{gathered}

1532: \label{17.2}M_{1,\,1}=(-2,1),\,\,\, M_{1,\,2}=(-1,2),\,\,\,

1533: M_{1,\,3}=(0,3),\,\,\,

1534: M_{2,\,k_1}=\left(-\frac34-\frac54k_1,\,1\right),\,\,\,\\

1535: M_{3,\,k_1}=\left(\frac14-\frac54k_1,2\right),\,\,\,M_{4,\,k_1}=\left(\frac12-\frac54k_1,1\right),\,\,\,M_{5,\,k_1}=\left(-2-5k_1,\,1\right),\,\,\,\\

1536: M_{6,\,k_1}=\left(-\frac74-\frac54k_1,\,0\right),\,\,\,M_{7,\,k_1}=\left(-\frac12-\frac54k_1,\,0\right),\,\,

1537: \,M_{8,\,k_1}=\left(\frac34-\frac54k_1,\,0\right),\,\, \,\\

1538: M_{9,\,k_1}=\left(-3-5k_1,\,0\right),\,\, \,

1539: M_{10,\,k_1,\,k_{2}}=\left(-\frac74-5k_1-\frac54k_{2},\,0\right),\,\,

1540: \,k_{i} \in \mathbb{N}\cup\{0\},\, i=1,2.

1541: \end{gathered}

1542: \end{equation}

1543: The convex hull obtained after locking of these points is similar to

1544: the one described in the section 14. The reduced equation

1545: corresponding to the edge $\Gamma_1^{(1)}$ is the following

1546: \begin{equation}

1547: \begin{gathered}

1548: \label{17.3}\xi^3 + 4g^{(l)}_{1/4}z^{1/4}\xi^2 + 6(g^{(l)}_{1/4})^2

1549: z^{1/2}\xi + 4(g^{(l)}_{1/4})^3 z^{3/4} =0.

1550: \end{gathered}

1551: \end{equation}

1552: The equation \eqref{17.3} has $12$ solutions

1553: \begin{equation}

1554: \begin{gathered}

1555: \label{17.4}\xi^{(l,\,m)}(z)=r_{1/4}^{(l,\,m)}\, z^{1/4},\,\,\,\,

1556: l=1,2,3,4;\,\,\,\, m=1,2,3,

1557: \end{gathered}

1558: \end{equation}

1559: where calculation of the coefficients $r_{1/4}^{(l,\,m)}$ yields

1560: \begin{equation*}

1561: \begin{gathered}

1562: \label{}r_{1/4}^{(l,\,1)}\,=-2g_{1/4}^{(l)};\,

1563: r_{1/4}^{(l,\,2)}\,=(-1+i)\,g_{1/4}^{(l)};\,r_{1/4}^{(l,\,3)}\,=-(1+i)\,g_{1/4}^{(l)},\,\,\,\,

1564: l=1,2,3,4.

1565: \end{gathered}

1566: \end{equation*}

1567: The lattice which contains the shifted carriers of the equation

1568: \eqref{17.1} and the reduced solutions \eqref{17.4} coincides with

1569: the lattice of the section 13. That is why power expansions for the

1570: reduced solutions \eqref{17.4} are

1571: \begin{equation}

1572: \begin{gathered}

1573: \label{17.5} \xi^{(l,\,m)}=r_{1/4}^{(l,\,m)} z^{1/4}

1574: +\sum_{k\,\in\,\,\mathbb{N}}r^{(l,\,m)}_{(1-5k)/4}\,z^{(1-5k)/4},\quad\,l=1,2,3,4;\,\quad\,

1575: m=1,2,3.

1576: \end{gathered}

1577: \end{equation}

1578: Here $r_{-1}^{(l,\,m)}=1/4$. Returning to the variables

1579: $v^{(l,\,m)}(z)$ we get the exponential additions $G_3^{(1)}\,

1580: G^{(1)}_1 l\,G^{(1)}_1 m$ and their applicability condition

1581: \begin{equation}

1582: \begin{gathered}

1583: \label{17.6}v^{(l,\,m)}(z)=\tilde{C}_2\,z^{1/4}\, \exp

1584: \left[\frac45\, r_{1/4}^{(l,\,m)}\,z^{5/4} + \sum^{\infty}_{k=2}

1585: \frac{4}{5(1-k)}

1586: r^{(l,\,m)}_{(1-5k)/4}z^{5(1-k)/4}\right];\\

1587: Re\left[r_{1/4}^{(l,\,m)}\,z^{5/4}\right]<0,\,\,z\rightarrow\infty;

1588: \,\quad\,l=1,2,3,4;\,\quad\,m=1,2,3.

1589: \end{gathered}

1590: \end{equation}

1591: 

1592: \section{Exponential additions of the third level, corresponding to the edge $\Gamma^{(1)}_3$ at $\alpha\neq0$.}

1593: 

1594: Exponential additions of the third level $y^{(l,\,m)}(z)$ (additions

1595: to the solutions $\xi^{(l,\,m)}(z)$) can be found as they have been

1596: found for the edge $\Gamma^{(1)}_2$ in the section 15. Let us in

1597: brief follow the procedure of calculations. After essential

1598: transformations we get the equation

1599: \begin{equation}

1600: \begin{gathered}

1601: \label{18.1}\eta_{z} +\eta^2+ (3\xi +

1602: 4\zeta)\eta+3\xi_z+3\xi^2+8\xi\zeta+6\zeta_z+6\zeta^2-10\psi^2=0.

1603: \end{gathered}

1604: \end{equation}

1605: Its carrier consists of the following points

1606: \begin{equation}

1607: \begin{gathered}

1608: \label{18.2}R_{1,\,1}=(-1,\,1),\,\,\, R_{1,\,2}=(0,\,2),\,\,\, R_{2,\,k}=(-\frac34-\frac54k,\,0),\,\,\, R_{3,\,k}=\left(\frac12-\frac54k,\,0\right),\\

1609: R_{4,\,k}=\left(-2-5k,\,0\right),\,\, \,

1610: R_{5,\,k}=\left(\frac14-\frac54k,\,1\right),\,\, \,k \in

1611: \mathbb{N}\cup\{0\}.

1612: \end{gathered}

1613: \end{equation}

1614: Closing of their convex hull yields the half-string similar to the

1615: one of the section 15. The edge $\Gamma_1^{(1)}$ is characterized by

1616: the reduced equation

1617: \begin{equation}

1618: \begin{gathered}

1619: \label{18.3}\eta^2+ (3r_{1/4}^{(l,\,m)}+4g_{1/4}^{(l)})z^{1/4}\eta+

1620: \left(3(r_{1/4}^{(l,\,m)})^2+8r_{1/4}^{(l,\,m)}g_{1/4}^{(l)}+

1621: 6(g_{1/4}^{(l)})^2 \right)z^{1/2}=0

1622: \end{gathered}

1623: \end{equation}

1624: which possesses $24$ solutions

1625: \begin{equation}

1626: \begin{gathered}

1627: \label{18.4}\eta^{(l,\,m,\,p)}(z)=q^{(l,\,m,\,p)}_{1/4}\,

1628: z^{1/4};\,\,\, l=1,2,3,4;\,\,\,\, m=1,2,3; \,\,\,\,p=1,2.

1629: \end{gathered}

1630: \end{equation}

1631: 

1632: The coefficients $q^{(l,\,m,\,p)}_{1/4}$ are equal to

1633: \begin{equation*}

1634: \begin{gathered}

1635: \label{}

1636: q_{1/4}^{(l,\,m,\,p=\{1,\,2\})}=-\frac32\,r_{1/4}^{(l,\,m)}-2\,g_{1/4}^{(l)}

1637: \pm\,\frac12i\sqrt{{(3\,(r_{1/4}^{(l,\,m)})^{2}+8\,{r_{1/4}^{(l,\,m)}}\,g_{1/4}^{(l)}+8\,

1638: (g_{1/4}^{(l)})^{2}}},\\

1639:  l=1,2,3,4;\,\,\,\,m=1,2,3.

1640: \end{gathered}

1641: \end{equation*}

1642: The set $\mathbf{K}$ in this case is \eqref{13.11}. Consequently

1643: power expansions for the solutions \eqref{18.4} can be written as

1644: \begin{equation}

1645: \begin{gathered}

1646: \label{18.5}\eta^{(l,\,m,\,p)}(z)=q^{(l,\,m,\,p)}_{1/4} z^{1/4}

1647: +\sum_{k\,\in\,\,\mathbb{N}}q^{(l,\,m,\,p)}_{(1-5k)/4}\,z^{(1-5k)/4},\,\,\,\\

1648: l=1,2,3,4;\,\quad\, m=1,2,3;\,\quad\, p=1,2.

1649: \end{gathered}

1650: \end{equation}

1651: The coefficients $q^{(l,\,m,\,p)}_{-1}$ do not depend on $\alpha$

1652: and $q^{(l,\,m,\,p)}_{-1}=1/4$.

1653: 

1654: So the exponential additions $G_3^{(1)}\, G^{(1)}_1 l\,G^{(1)}_1

1655: m\,G^{(1)}_1 p$ are the following

1656: \begin{equation}

1657: \begin{gathered}

1658: \label{18.6}y^{(l,\,m,\,p)}(z)=\tilde{C}_3\,z^{1/4} \exp

1659: \left[\frac45\, q^{(l,\,m,\,p)}_{1/4}\,z^{5/4} + \sum^{\infty}_{k=2}

1660: \frac{4}{5(1-k)} q^{(l,\,m,\,p)}_{(1-5k)/4}

1661: z^{5(1-k)/4}\right],\\

1662: l=1,2,3,4;\,\quad\, m=1,2,3;\,\quad\, p=1,2.

1663: \end{gathered}

1664: \end{equation}

1665: 

1666: The applicability condition for the additions \eqref{18.6} coincides

1667: with \eqref{15.11} provided that $q_{1/4}^{(n,\,l,\,m,\,p)}$ replace

1668: $q_{1/4}^{(l,\,m,\,p)}$. Taking into account obtained additions the

1669: solution $w(z)$ can be written as it has been done in formula

1670: \eqref{15.12}.

1671: 

1672: 

1673: \section{Conclusion.}

1674: 

1675: In this work the fourth-order analogue to the second Painlev\'{e}

1676: equation was studied with a help of the power geometry method

1677: \cite{Bruno01, Bruno02, Bruno03}. We found all power and non-power

1678: asymptotics of its solutions, power expansions generated by these

1679: power asymptotics and exponential additions which correspond to

1680: certain expansions. The results depend on the value of parameter

1681: $\alpha$.

1682: 

1683: 

1684: First of all let us briefly review the case $\alpha \neq0$. Near the

1685: point $z=0$ we obtained one-parametric family $G_1^{(0)}4$, three

1686: two-parametric families $G_1^{(0)}3$, $G_1^{(1)}3$, $G_1^{(1)}4$,

1687: three three-parametric families $G_1^{(0)}2$, $G_1^{(1)}1$,

1688: $G_1^{(1)}2$, one four-parametric family $G_1^{(0)}1$ and the family

1689: $G_4^{(1)}1$ (the families $G_1^{(0)}2$, $G_1^{(0)}3$, $G_1^{(0)}4$,

1690: $G_4^{(1)}1$ are the special cases of $G_1^{(0)}1$). These

1691: expansions converge for small $|z|$. Their existence and analyticity

1692: follow from the Cauchy theorem.

1693: 

1694: Besides near the point $z=\infty$ we found the families:

1695: $G_2^{(1)}n\,\,(n=1,2,3,4)$ and $G_3^{(1)}1$. For each of these

1696: expansions three-level exponential additions we computed:

1697: $G_2^{(1)}n\, G^{(1)}_1 l\,G^{(1)}_1 m\,G^{(1)}_1

1698: p\,\,\,(n,l=1,2,3,4;\,\,\,m=1,2,3;\,\,\,p=1,2)$ and $G_3^{(1)}\,

1699: G^{(1)}_1 l\,G^{(1)}_1 m\,G^{(1)}_1 p\,\,\,

1700: (l=1,2,3,4;\,\,\,m=1,2,3;\,\,\,p=1,2)$.

1701: 

1702: In addition it is important to mention that some partial solutions

1703: of the studied equation can be found using the expansion

1704: $G_3^{(1)}1$. It is obvious that at the certain values of the

1705: parameter $\alpha$ (more exactly $\alpha=-2,\,\,-1,\,\,1,\,\,2$) the

1706: series in the formula \eqref{6.5} truncates and we get rational

1707: solutions

1708: \begin{equation}

1709: \label{19.1}w(z)=\frac{-\alpha}{z},\,\,\,\,  \alpha=-2,-1,\,1,\,2.

1710: \end{equation}

1711: Application of these solutions as "seed solutions" in the

1712: B\"{a}cklund transformations for the studied equation yields other

1713: rational solutions at whole values of the parameter $\alpha$.

1714: Partial solutions \eqref{19.1} can be also obtained from the

1715: expansions $G_1^{(1)}1$, $G_1^{(1)}2$, $G_1^{(1)}3$ and

1716: $G_1^{(1)}4$.

1717: 

1718: In the case $\alpha=0$ near the point $z=0$ we found one-parametric

1719: family $\tilde{G}_1^{(0)}4$, three two-parametric families

1720: $\tilde{G}_1^{(0)}3$, $\tilde{G}_1^{(1)}3$, $\tilde{G}_1^{(1)}4$,

1721: three three-parametrical families $\tilde{G}_1^{(0)}2$,

1722: $\tilde{G}_1^{(1)}1$, $\tilde{G}_1^{(1)}2$ and one four-parametric

1723: family $\tilde{G}_1^{(0)}1$ of power expansions (the families

1724: $\tilde{G}_1^{(0)}2$, $\tilde{G}_1^{(0)}3$, $\tilde{G}_1^{(0)}4$ are

1725: the special cases of $G_1^{(0)}1$). All of them converge for small

1726: $|z|$.

1727: 

1728: Near the point $z=\infty$ we computed four families of non-power

1729: asymptotics $\tilde{G}_3^{(1)}n,\,\,(n=1,2,3,4)$ and four families

1730: of power expansions $\tilde{G}_2^{(1)}n\,\,(n=1,2,3,4)$.

1731: 

1732: The expansions $G_1^{(1)}n\,\,(n=1,2,3,4)$ were found earlier

1733: \cite{Kudryashov01,Kudryashov10}, while all other expansions are the

1734: new ones.

1735: 

1736: To sum up we would like to emphasize that the obtained power

1737: expansions differ from the power expansions of the Painlev\'{e}

1738: equations $P_1\div P_6$ solutions  \cite{Bruno04, Bruno05, Bruno06,

1739: Bruno07, Gromak01},. This fact can be interpreted as the additional

1740: proof of the hypothesis that the fourth-order equation \eqref{1.3}

1741: determines new transcendental functions as the equations $P_1\div

1742: P_6$ do.

1743: 

1744: 

1745: 

1746: \begin{thebibliography}{99}

1747: 

1748: 

1749: \bibitem{Ablowitz01} \textit{Ablowitz M.J., Clarcson P.A.} Solitons, Nonlinear Evolution Equations and Inverse Scattering.

1750: Cambridge University Press; 1991.

1751: 

1752: \bibitem{Barouch03} \textit{Barouch E., McCay B.M., Wu T.T.} Phys Rev Lett 1973; 31.

1753: 

1754: \bibitem{Brezin04} \textit{Brezin E., Kazakov V.} Phys Lett B 1990;236:144-150.

1755: 

1756: \bibitem{De_Boer05} \textit{De Boer P.C.T., Ludford L.S.S.} Plazm Phys 1975;17:29-43.

1757: 

1758: \bibitem{Ablowitz06} \textit{Ablowitz M.J., Segur H.} Phys Rev Lett 1997;38:1103-1106.

1759: 

1760: \bibitem{Hall07} \textit{Hall P.} IMA J Appl Math 1982;29:173-196.

1761: 

1762: \bibitem{Chandrasecar08} \textit{Chandrasekar S.} Proc Roy Soc London A 1986;408:209-232.

1763: 

1764: \bibitem{Kudryashov01} \textit{Kudryashov N.A.} Analytical theory of nonlinear differential

1765: equations,  Institute of Computer Investigations, Moscow-Igevsk,

1766: 2004, 360 p. (in Russian).

1767: 

1768: \bibitem{Kudryashov02} \textit{Kudryashov N.A.} Phys

1769: Lett A 1997;233:387-400.

1770: 

1771: \bibitem{Kudryashov03} \textit{Kudryashov N.A.} Phys Lett A 1997;224 N 6.

1772: P. 353--360.

1773: 

1774: \bibitem{Kudryashov04} \textit{Kudryashov N.A.} J Phys A: Math Gen 1998;31:N 6.

1775: P. L.129--L.137.

1776: 

1777: \bibitem{Kudryashov05} \textit{Kudryashov N.A., Soukharev M.B.} Phys Lett A 1998;237:206-216.

1778: 

1779: \bibitem{Kudryashov06} \textit{Kudryashov N.A., Pickering A.} J Phys A: Math Gen 1998;31:999 - 1014.

1780: 

1781: \bibitem{Kudryashov07} \textit{Kudryashov N.A.} Phys Lett A: 1999;252:173-179.

1782: 

1783: \bibitem{Kudryashov08} \textit{Kudryashov N.A.} J Phys A: Math Gen 1999;32:999--1013.

1784: 

1785: \bibitem{Kudryashov09} \textit{Kudryashov N.A.} Phys Lett A 2000;273:194 -- 202,353--360.

1786: 

1787: \bibitem{Kudryashov10} \textit{Kudryashov N.A.} Theoretical and mathematical physics 2000;122:72 - 86.

1788: 

1789: \bibitem{Kudryashov11} \textit{Kudryashov N.A., Pickering A.} CRM Proceedings and Lecture Notes 2000;25:245-253.

1790: 

1791: \bibitem{Kudryashov12} \textit{Kudryashov N.A.} J Phys A: Math Gen 2002;35:93 -- 99.

1792: 

1793: \bibitem{Kudryashov13} \textit{Kudryashov N.A.} J Phys A: Math Gen 2002;35:4617--4632.

1794: 

1795: \bibitem{Kudryashov14} \textit{Kudryashov N.A., Soukharev M.B.} ANZIAM Industrial and Applied Mathematics 2002.

1796: 

1797: \bibitem{Kudryashov15} \textit{Kudryashov N.A.} J Math Phys 2003;44:6160--6178.

1798: 

1799: \bibitem{Kudryashov16} \textit{Kudryashov N.A., Efimova O.Yu.} Chaos, Solitons \& Fractals; 2006 (in press).

1800: 

1801: \bibitem{Airault01} \textit{Airault H.} Studies in applied Mathematics 1979;61:31 -- 53.

1802: 

1803: \bibitem{Clarkson01} \textit{Clarkson P.A., Joschi N., Pickering A.} Inverse problems 1999;15:175 -- 187.

1804: 

1805: \bibitem{Clarkson02} \textit{Clarkson P.A., Hone A.N.W., Joschi N.} Journal of Nonlinear Mathematical

1806: physics 2003;10.

1807: 

1808: \bibitem{Cosgrove01} \textit{Cosgrove C.M.} Study Appl Math 2000;104:1 -- 65.

1809: 

1810: \bibitem{Joshi01} \textit{Creswell G., Joshi N.} J Phys A: Math Gen 1999;32:655 - 669.

1811: 

1812: \bibitem{Hone01} \textit{Hone Andrew N.W.} Physica D 1998;118:1 -- 16.

1813: 

1814: \bibitem{Hone02} \textit{Hone Andrew N.W.} J Phys A 2001;34:2235 -- 2245.

1815: 

1816: \bibitem{Gordoa01} \textit{Gordoa P.R., Pickering A.} Journal of mathematical physics 1999;11:5749 -- 5766.

1817: 

1818: \bibitem{Gordoa02} \textit{Gordoa P.R.}Phys Lett A 2001;287:365 - 370.

1819: 

1820: \bibitem{Flashka01} \textit{Flaschka H., Newell A.C.} Communications in Mathematical Physics 1980;76:65 -- 116.

1821: 

1822: \bibitem{Kawai01} \textit{Kawai T., Koike T., Nishikawa Y., Takei Y.} 2004 preprint/RS/RIMS 1471,Kioto.

1823: 

1824: \bibitem{Mugan01} \textit{Mugan U., Jrad F.} J Phys A: Math Gen 1999;32:7933 - 7952.

1825: 

1826: \bibitem{Mugan02} \textit{Mugan U., Jrad F.} Journal of Nonlinear Mathematical Physics A 2002;9(3):282-310.

1827: 

1828: \bibitem{Mugan03} \textit{Mugan U., Jrad F.} Zaitschrift fur Naturforshing 2004;9 (3):282-310.

1829: 

1830: \bibitem{Mugan04} \textit{Mugan U., Jrad F.} Zaitschrift fur Naturforshung 2005;9(3):282-310.

1831: 

1832: \bibitem{Nijhoff01} \textit{Nijhoff F.W., Walker A.J.} Glasgow Math J 2001;43A:199 -- 123.

1833: 

1834: \bibitem{Pickering01} \textit{Pickering A.} Phys Lett A 2002;301:275 - 280.

1835: 

1836: \bibitem{Shimomura01} \textit{Shimomura S.} Proc Japan Acad 80 Ser A 2004:105 -- 109.

1837: 

1838: \bibitem{Bruno01} \textit{Bruno A.D.} Power geometry in algebraic and differential equations,

1839: Moscow, Nauka, Fizmatlit, 1998, 288 p (in Russian).

1840: 

1841: \bibitem{Bruno02} \textit{Bruno A.D. } ISAAC; 2001;51-71.

1842: 

1843: \bibitem{Bruno03} \textit{Bruno A.D.} Uspehi of mathematical nauk 2004;59:31-80.

1844: 

1845: \bibitem{Bruno04} \textit{Bruno A.D., Petrovich V.Yu.} KIAM preprint 2004;No.9,Moscow(in Russian).

1846: 

1847: \bibitem{Bruno05} \textit{Bruno A.D., Zavgorodnya Yu.B.} KIAM preprint 2003;No.48,Moscow(in Russian).

1848: 

1849: \bibitem{Bruno06} \textit{Bruno A.D., Karulina E.S.} Doklady RAN 2004;395,No.4:439 -- 444(in Russian).

1850: 

1851: \bibitem{Bruno07} \textit{Bruno A.D., Goruchkina I.B.} Doklady RAN 2004;395,No.6:733-737(in Russian).

1852: 

1853: \bibitem{Gromak01} \textit{Gromak V.I., Laine I., Shimomura S.} Painleve Differential Equations in

1854: the Complex Plane, Walter de Gruyter, Berlin, New York, 2002.

1855: 

1856: \end{thebibliography}

1857: 

1858: \end{document}

1859: