1: 

2: \documentstyle[11pt,epsf]{article}

3: \def\xx{\vrule height 0.3em depth 0.2em width 0.3em}

4: %\input psfig.sty

5: \def\ra{\rightarrow}

6: \def\longra{\longrightarrow}

7: 

8: \def\fl{\forall}

9: \def\ify{\infty}

10: \def\lgl{\langle}

11: \def\op{\oplus}

12: \def\ot{\otimes}

13: \def\ov{\overline}

14: \def\rgl{\rangle}

15: \def\ts{\times}

16: \def\wt{\widetilde}

17: 

18: \def\a{\alpha}

19: \def\b{\beta}

20: \def\d{\delta}

21: \def\g{\gamma}

22: \def\s{\sigma}

23: \def\t{\theta}

24: \def\ve{\varepsilon}

25: \def\vp{\varphi}

26: 

27: \def\D{\Delta}

28: \def\G{\Gamma}

29: 

30: \font\tenbb=msbm10

31: \font\sevenbb=msbm7

32: \font\fivebb=msbm5

33: \newfam\bbfam

34: \textfont\bbfam=\tenbb \scriptfont\bbfam=\sevenbb

35: \scriptscriptfont\bbfam=\fivebb

36: \def\bb{\fam\bbfam}

37: 

38: \def\Cb{{\bb C}}

39: \def\Nb{{\bb N}}

40: \def\Rb{{\bb R}}

41: 

42: \def\Hc{{\cal H}}

43: 

44: \catcode`\@=11

45: \def\displaylinesno #1{\displ@y\halign{

46: \hbox to\displaywidth{$\@lign\hfil\displaystyle##\hfil$}&

47: \llap{$##$}\crcr#1\crcr}}

48: 

49: \def\ldisplaylinesno #1{\displ@y\halign{

50: \hbox to\displaywidth{$\@lign\hfil\displaystyle##\hfil$}&

51: \kern-\displaywidth\rlap{$##$}

52: \tabskip\displaywidth\crcr#1\crcr}}

53: \catcode`\@=12

54: 

55: \def\semi{\mathop{>\!\!\!\triangleleft}}

56: 

57: \def\build#1_#2^#3{\mathrel{

58: \mathop{\kern 0pt#1}\limits_{#2}^{#3}}}

59: 

60: 

61: \def\ot{\otimes}

62: \def\v{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{v.eps}}\;}

63: \def\vlv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlv.eps}}\;}

64: \def\vuv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuv.eps}}\;}

65: \def\vdv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdv.eps}}\;}

66: \def\vup{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vup.eps}}\;}

67: \def\vdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdp.eps}}\;}

68: \def\vrp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vrp.eps}}\;}

69: \def\w{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{w.eps}}\;}

70: \def\vlvlv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlvlv.eps}}\;}

71: \def\vlvuv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlvuv.eps}}\;}

72: \def\vlvdv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlvdv.eps}}\;}

73: \def\vlvup{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlvup.eps}}\;}

74: \def\vlvdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlvdp.eps}}\;}

75: \def\vlvrp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlvrp.eps}}\;}

76: \def\vlw{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlw.eps}}\;}

77: \def\vuvlv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuvlv.eps}}\;}

78: \def\vuvuv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuvuv.eps}}\;}

79: \def\vuvdv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuvdv.eps}}\;}

80: \def\vuvup{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuvup.eps}}\;}

81: \def\vuvdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuvdp.eps}}\;}

82: \def\vuvrp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuvrp.eps}}\;}

83: \def\vuw{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuw.eps}}\;}

84: \def\vdvlv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdvlv.eps}}\;}

85: \def\vdvuv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdvuv.eps}}\;}

86: \def\vdvdv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdvdv.eps}}\;}

87: \def\vdvup{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdvup.eps}}\;}

88: \def\vdvdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdvdp.eps}}\;}

89: \def\vdvrp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdvrp.eps}}\;}

90: \def\vdw{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdw.eps}}\;}

91: \def\vluvv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vluvv.eps}}\;}

92: \def\vudvv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vudvv.eps}}\;}

93: \def\vldvv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vldvv.eps}}\;}

94: \def\vldvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vldvp.eps}}\;}

95: \def\vlrvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vlrvp.eps}}\;}

96: \def\vluvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vluvp.eps}}\;}

97: \def\vdrvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdrvp.eps}}\;}

98: \def\vduvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vduvp.eps}}\;}

99: \def\vddvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vddvp.eps}}\;}

100: \def\vuuvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuuvp.eps}}\;}

101: \def\vudvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vudvp.eps}}\;}

102: \def\vurvp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vurvp.eps}}\;}

103: \def\p{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{p.eps}}\;}

104: \def\pv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pv.eps}}\;}

105: \def\pdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pdp.eps}}\;}

106: \def\pdpdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pdpdp.eps}}\;}

107: \def\pdpv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pdpv.eps}}\;}

108: \def\pddpp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pddpp.eps}}\;}

109: \def\pdupp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pdupp.eps}}\;}

110: \def\pw{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pw.eps}}\;}

111: \def\pvmp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pvmp.eps}}\;}

112: \def\pvlp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pvlp.eps}}\;}

113: \def\pvrp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pvrp.eps}}\;}

114: \def\plrvv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{plrvv.eps}}\;}

115: \def\pvuv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{pvuv.eps}}\;}

116: \def\wlv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wlv.eps}}\;}

117: \def\wulv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wulv.eps}}\;}

118: \def\wurv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wurv.eps}}\;}

119: \def\wdrv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wdrv.eps}}\;}

120: \def\wulp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wulp.eps}}\;}

121: \def\wurp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wurp.eps}}\;}

122: \def\wdlp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wdlp.eps}}\;}

123: \def\wdrp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wdrp.eps}}\;}

124: \def\wmrp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wmrp.eps}}\;}

125: \def\wmlp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{wmlp.eps}}\;}

126: \def\ws{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{ws.eps}}\;}

127: \def\vupv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vupv.eps}}\;}

128: \def\vdpv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdpv.eps}}\;}

129: \def\vrpv{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vrpv.eps}}\;}

130: \def\vupdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vupdp.eps}}\;}

131: \def\vdpdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdpdp.eps}}\;}

132: \def\vrpdp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vrpdp.eps}}\;}

133: \def\vuupp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vuupp.eps}}\;}

134: \def\vrrpp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vrrpp.eps}}\;}

135: \def\vddpp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vddpp.eps}}\;}

136: \def\vurpp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vurpp.eps}}\;}

137: \def\vdrpp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vdrpp.eps}}\;}

138: \def\vudpp{\;\raisebox{-1.5mm}{\epsfysize=6mm\epsfbox{vudpp.eps}}\;}

139: 

140: 

141: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

142: 

143: 

144: 

145: 

146: \begin{document}

147: \begin{center}

148: {\large {\bf  Renormalization in quantum field theory and the

149: Riemann-Hilbert problem II: the $\beta$-function, diffeomorphisms

150: and the renormalization group} \footnote{IHES/M/00/22,

151: hep-th/0003188; connes@ihes.fr, kreimer@ihes.fr}\\[1cm]} {\bf

152: Alain Connes and Dirk Kreimer\\[5mm]} {\small Institut des Hautes

153: \'Etudes Scientifiques, March 2000\\[2cm]}

154: %{\small Institut des Hautes Etudes Scientifiques}\\ {\small

155: %connes@ihes.fr, kreimer@ihes.fr}}

156: \end{center}

157: 

158: % \maketitle

159: 

160: 

161: {\footnotesize \noindent {\bf Abstract } } {\em We showed in part

162: I that the Hopf algebra ${\cal H}$ of Feynman graphs in a given

163: QFT is the algebra of coordinates on a complex infinite

164: dimensional Lie group $G$ and that the renormalized theory is

165: obtained from the unrenormalized one by evaluating at $\ve=0$ the

166: holomorphic part $\gamma_+(\ve)$ of the Riemann--Hilbert

167: decomposition $\gamma_-(\ve)^{-1}\gamma_+(\ve)$ of the loop

168: $\gamma(\ve)\in G$ provided by dimensional regularization. We show

169: in this paper that the group $G$ acts naturally on the complex

170: space $X$ of dimensionless coupling constants of the theory. More

171: precisely, the formula $g_0=gZ_1Z_3^{-3/2}$ for the effective

172: coupling constant, when viewed as a formal power series, does

173: define a Hopf algebra homomorphism between the Hopf algebra of

174: coordinates on the group of formal diffeomorphisms to the Hopf

175: algebra ${\cal H}$. This allows first of all to read off directly,

176: without using the group $G$, the bare coupling constant and the

177: renormalized one from the Riemann--Hilbert decomposition of the

178: unrenormalized effective coupling constant viewed as a loop of

179: formal diffeomorphisms. This shows that renormalization is

180: intimately related with the theory of non-linear complex bundles

181: on the Riemann sphere of the dimensional regularization parameter

182: $\ve$. It also allows to lift both the renormalization group and

183: the $\beta$-function as the asymptotic scaling in the group $G$.

184: This exploits the full power of the Riemann--Hilbert decomposition

185: together with the invariance of $\gamma_-(\ve)$ under a change of

186: unit of mass. This not only gives a conceptual proof of the

187: existence of the renormalization group but also delivers a

188: scattering formula in the group $G$ for the full higher pole

189: structure of minimal subtracted counterterms in terms of the

190: residue.}

191: \newpage

192: 

193: 

194: 

195: 

196: 

197: 

198: \section{Introduction}\label{secI}

199: We showed in part~I of this paper \cite{I}

200: that perturbative

201: renormalization is a special case of a general mathematical

202: procedure of extraction of finite values based on the

203: Riemann-Hilbert problem. More specifically we associated to any

204: given renormalizable quantum field theory an (infinite

205: dimensional) complex Lie group $G$. We then showed that passing

206: from the unrenormalized theory to the renormalized one was exactly

207: the replacement of the loop $d \ra \g (d) \in G$ of elements of

208: $G$ obtained from dimensional regularization (for $d \ne D =$

209: dimension of space-time) by the value $\g_+ (D)$ of its Birkhoff

210: decomposition, $\g (d) = \g_- (d)^{-1} \, \g_+ (d)$.

211: 

212: The original loop $d \ra \g (d)$ not only depends upon the

213: parameters of the theory but also on the additional ``unit of

214: mass'' $\mu$ required by dimensional analysis. We shall show in

215: this paper that the mathematical concepts developped in part~I

216: provide very powerful tools to lift the usual concepts of the

217: $\b$-function and renormalization group from the space of coupling

218: constants of the theory to the complex Lie group $G$.

219: 

220: We first observe, taking $\vp_6^3$ as an illustrative example to

221: fix ideas and notations, that even though the loop $\g (d)$ does

222: depend on the additional parameter $\mu$, $$ \mu \ra \g_{\mu} (d)

223: \, , \leqno (1) $$ the negative part $\g_{\mu^-}$ in the Birkhoff

224: decomposition, $$ \g_{\mu} (d) = \g_{\mu^-} (d)^{-1} \, \g_{\mu^+}

225: (d) \leqno (2) $$ is actually independent of $\mu$, $$

226: \frac{\partial}{\partial \mu} \, \g_{\mu^-} (d) = 0 \, . \leqno

227: (3) $$ This is a restatement of a well known fact and follows

228: immediately from dimensional analysis. Moreover, by construction,

229: the Lie group $G$ turns out to be graded, with grading, $$ \t_t

230: \in {\rm Aut} \, G \ , \quad t \in \Rb \, , \leqno (4) $$

231: inherited from the grading of the Hopf algebra $\Hc$ of Feynman

232: graphs given by the loop number, $$ L (\G) = \hbox{loop number of}

233: \ \G \leqno (5) $$ for any 1PI graph $\G$.

234: 

235: The straightforward equality, $$ \g_{e^t \mu} (d) = \t_{t \ve}

236: (\g_{\mu} (d)) \qquad \fl \, t \in \Rb \, , \ \ve = D-d \leqno (6)

237: $$ shows that the loops $\g_{\mu}$ associated to the

238: unrenormalized theory satisfy the striking property that the

239: negative part of their Birkhoff decomposition is unaltered by the

240: operation, $$ \g (\ve) \ra \t_{t\ve} (\g (\ve)) \, . \leqno (7) $$

241: In other words, if we replace $\g (\ve) $ by $\t_{t\ve} (\g

242: (\ve))$ we do not change the negative part of its Birkhoff

243: decomposition. We settled now for the variable, $$ \ve = D - d \in

244: \Cb \backslash \{ 0 \} \, . \leqno (8) $$ Our first result

245: (section~\ref{secII}) is a complete characterization of the loops $\g (\ve)

246: \in G$ fulfilling the above striking invariance. This

247: characterization only involves the negative part $\g_- (\ve)$ of

248: their Birkhoff decomposition which by hypothesis fulfills, $$ \g_-

249: (\ve) \, \t_{t \ve} (\g_- (\ve)^{-1}) \ \hbox{is convergent for} \

250: \ve \ra 0 \, . \leqno (9) $$ It is easy to see that the limit of

251: (9) for $\ve \ra 0$ defines a one parameter subgroup, $$ F_t \in G

252: \, , \ t \in \Rb \leqno (10) $$ and that the generator $\b =

253: \left( \frac{\partial}{\partial t} \, F_t \right)_{t=0}$ of this

254: one parameter group is related to the {\it residue} of $\g$ $$

255: \build{\rm Res}_{\ve = 0}^{} \g = - \left( \frac{\partial}{\partial

256: u} \, \g_- \left( \frac{1}{u} \right) \right)_{u=0} \leqno (11) $$

257: by the simple equation, $$ \b = Y \, {\rm Res} \, \g \, , \leqno

258: (12) $$ where $Y = \left( \frac{\partial}{\partial t} \, \t_t

259: \right)_{t=0}$ is the grading.

260: 

261: This is straightforward but our result is the following formula

262: (14) which gives $\g_- (\ve)$ in closed form as a function of

263: $\b$. We shall for convenience introduce an additional generator

264: in the Lie algebra of $G$ (i.e.~primitive elements of $\Hc^*$)

265: such that, $$ [Z_0 , X] = Y(X) \qquad \fl \, X \in \hbox{Lie} \ G

266: \, . \leqno (13) $$ The scattering formula for $\g_- (\ve)$ is

267: then, $$ \g_- (\ve) = \lim_{t \ra \ify} e^{-t \left(

268: \frac{\b}{\ve} + Z_0 \right)} \, e^{t Z_0} \, . \leqno (14) $$

269: Both factors in the right hand side belong to the semi-direct

270: product, $$ \wt G = G \, \semi_{\t} \, \Rb \leqno (15) $$ of the

271: group $G$ by the grading, but of course the ratio (14) belongs to

272: the group $G$.

273: 

274: This shows (section~\ref{secIII}) that the higher pole structure of the

275: divergences is uniquely determined by the residue and gives a

276: strong form of the t'Hooft relations, which will come as an

277: immediate corollary.

278: 

279: In section~\ref{sec4} we show, specializing to the massless case, that the

280: formula for the bare coupling constant, $$ g_0 = g \, Z_1 \,

281: Z_3^{-3/2} \leqno (16) $$ where both $g \, Z_1 =g + \d g$ and the

282: field strength renormalization constant $Z_3$ are thought of as

283: power series (in $g$) of elements of the Hopf algebra $\Hc$, does

284: define a Hopf algebra homomorphism, $$ \Hc_{CM} \build

285: \longra_{}^{g_0} \Hc \, , \leqno (17) $$ from the Hopf algebra

286: $\Hc_{CM}$ of coordinates on the group of formal diffeomorphisms

287: of $\Cb$ such that, $$ \vp (0) = 0 \, , \ \vp' (0) = {\rm id}

288: \leqno (18) $$ to the Hopf algebra $\Hc$ of the massless theory.

289: We had already constructed in \cite{CK} a Hopf algebra

290: homomorphism from $\Hc_{\rm CM}$ to the Hopf algebra of rooted

291: trees, but the physical significance of this construction was

292: unclear.

293: 

294: The homomorphism (17) is quite different in that for instance the

295: transposed group homomorphism, $$ G \build \longra_{}^{\rho} {\rm

296: Diff} (\Cb) \leqno (19) $$ lands in the subgroup of {\it odd}

297: diffeomorphisms, $$ \vp (-z) = -\vp (z) \qquad \fl \, z \, .

298: \leqno (20) $$ Moreover its physical significance will be

299: transparent. We shall show in particular that the image by $\rho$

300: of $\b = Y \, {\rm Res} \, \g$ is the usual $\b$-function of the

301: coupling constant $g$.

302: 

303: We discovered the homomorphism (17) by lengthy concrete

304: computations. We have chosen to include them in an appendix

305: besides our conceptual proof given in section~4. The main reason

306: for this choice is that the explicit computation allows to

307: validate the concrete ways of handling the coproduct,

308: coassociativity, symmetry factors$\ldots$ that underly the theory.

309: 

310: As a corollary of the construction of $\rho$ we get an {\it

311: action} by (formal) diffeomorphisms of the group $G$ on the space

312: $X$ of (dimensionless) coupling constants of the theory. We can

313: then in particular formulate the Birkhoff decomposition {\it

314: directly} in the group, $$ {\rm Diff} \, (X) \leqno (21) $$ of

315: formal diffeomorphisms of the space of coupling constants.

316: 

317: The unrenormalized theory delivers a loop $$ \d (\ve) \in {\rm

318: Diff} \, (X) \, , \ \ve \ne 0 \leqno (22) $$ whose value at $\ve$

319: is simply the unrenormalized effective coupling constant.

320: 

321: The Birkhoff decomposition, $$ \d (\ve) = \, \d_+

322: (\ve) \, \d_- (\ve)^{-1} \leqno (23) $$ of this loop gives directly then, $$ \d_-

323: (\ve) = \hbox{ bare coupling constant} \leqno (24) $$ and, $$ \d_+

324: (D) = \hbox{renormalized effective coupling constant.} \leqno (25)

325: $$ This result is now, in its statement, no longer depending upon

326: our group $G$ or the Hopf algebra $\Hc$. But of course the proof

327: makes heavy use of the above ingredients.

328: 

329: Now the Birkhoff decomposition of a loop, $$ \d (\ve) \in {\rm

330: Diff} \, (X) \, , \leqno (26) $$ admits a beautiful geometric

331: interpretation. If we let $X$ be a complex manifold and pass from

332: formal diffeomorphisms to actual ones, the data (26) is the initial

333: data to perform, by the clutching operation, the construction of a

334: complex bundle, $$ P = (S^+ \ts X) \, \cup_{\d} (S^- \ts X) \leqno

335: (27) $$ over the sphere $S = P_1 (\Cb) = S^+ \cup S^-$, and with

336: fiber $X$, $$ X \longra P \build \longra_{}^{\pi} S \, . \leqno

337: (28) $$ The meaning of the Birkhoff decomposition (23), $$ \d (\ve) =

338:  \d_+ \, (\ve) \, \d_- (\ve)^{-1}  $$ is then exactly

339: captured by an isomorphism of the bundle $P$ with the trivial

340: bundle, $$ S \ts X \, . \leqno (29) $$

341: 

342: \section{Asymptotic scaling in graded complex Lie groups}\label{secII}

343: We shall first prove the formula (14) of the introduction in the

344: general context of graded Hopf algebras and then apply it to the

345: Birkhoff decomposition of the loop associated in part~I to the

346: unrenormalized theory.

347: 

348: We let $\Hc$ be a connected commutative graded Hopf algebra

349: (connected means that $\Hc^{(0)} = \Cb$) and let $\t_t$, $t \in

350: \Rb$ be the one parameter group of automorphisms of $\Hc$

351: associated with the grading so that for $x \in \Hc$ of degree $n$,

352: $$ \t_t (x) = e^{tn} \, x \qquad \fl \, t \in \Rb \, . \leqno (1)

353: $$ By construction $\t_t$ is a Hopf algebra automorphism, $$ \t_t

354: \in {\rm Aut} (\Hc) \, . \leqno (2) $$ We also let $Y = \left(

355: \frac{\partial}{\partial t} \, \t_t \right)_{t=0}$ be the

356: generator which is a derivation of $\Hc$.

357: 

358: We let $G$ be the group of characters of $\Hc$, $$ \vp : \Hc \ra

359: \Cb \leqno (3) $$ i.e. of homomorphisms from the algebra $\Hc$ to

360: $\Cb$. The product in $G$ is given by, $$ (\vp_1 \, \vp_2) (x) =

361: \lgl \vp_1 \ot \vp_2 \, , \, \D x \rgl \leqno (4) $$  where $\D$ is

362: the coproduct in $\Hc$. The augmentation $\ov e$ of $\Hc$ is the

363: unit of $G$ and the inverse of $\vp \in G$ is given by, $$ \lgl

364: \vp^{-1} , x \rgl = \lgl \vp , S x \rgl \leqno (5) $$ where $S$ is

365: the antipode in $\Hc$.

366: 

367: We let $L$ be the Lie algebra of derivations, $$ \d : \Hc \ra \Cb

368: \, , \leqno (6) $$ i.e. of linear maps on $\Hc$ such that $$ \d

369: (xy) = \d (x) \, \ov e (y) + \ov e (x) \, \d (y) \qquad \fl \, x,y

370: \in \Hc \, . \leqno (7) $$ Even if $\Hc$ is of finite type so that

371: $\Hc^{(n)}$ is finite dimensional for any $n \in \Nb$ there are

372: more elements in $L$ than in the Lie algebra $P$ of primitive

373: elements, $$ \D Z = Z \ot 1 + 1 \ot Z \leqno (8) $$ in the graded

374: dual Hopf algebra $\Hc_{\rm gr}^*$ of $\Hc$. But one passes from

375: $P$ to $L$ by completion relative to the $I$-adic topology, $I$

376: being the augmentation ideal of $\Hc_{\rm gr}^*$.

377: 

378: The linear dual $\Hc^*$ is in general an algebra (with product

379: given by (4)) but not a Hopf algebra since the coproduct is not

380: necessarily well defined. It is however well defined for

381: characters $\vp$ or derivations $\d$ which satisfy respectively

382: $\D \vp = \vp \ot \vp$, and $\D \d = \d \ot 1 + 1 \ot \d$.

383: 

384: For $\d \in L$ the expression, $$ \vp = \exp \, \d \leqno (9) $$

385: makes sense in the algebra $\Hc^*$ since when evaluated on $x \in

386: \Hc$ one has $\lgl x , \d^n \rgl = 0$ for $n$ large enough (since

387: $\lgl x , \d^n \rgl = \lgl \D^{(n-1)} x , \d \ot \cdots \ot \d

388: \rgl$ vanishes for $n > \deg x$). Moreover $\vp$ is a group-like

389: element of $\Hc^*$, i.e. a character of $\Hc$. Thus $\vp \in G$.

390: 

391: The one parameter group $\t_t \in {\rm Aut} \, (\Hc)$ acts by

392: automorphisms on the group $G$, $$ \lgl \t_t (\vp) , x \rgl = \lgl

393: \vp , \t_t (x) \rgl \qquad \fl \, x \in \Hc \leqno (10) $$ and the

394: derivation $Y$ of $\Hc$ acts on $L$ by $$ \lgl Y (\d) , x \rgl =

395: \lgl \d , Y(x) \rgl \, , \leqno (11) $$ and defines a derivation

396: of the Lie algebra $L$ where we recall that the Lie bracket in $L$

397: is given by, $$ \lgl [\d_1 , \d_2] , x \rgl = \lgl \d_1 \ot \d_2 -

398: \d_2 \ot \d_1 \, , \, \D x \rgl \qquad \fl \, x \in \Hc \, .

399: \leqno (12) $$ Let us now consider a map, $$ \ve \in \Cb

400: \backslash \{ 0 \} \ra \vp_{\ve} \in G \leqno (13) $$ such that

401: for any $x \in \Hc$, $\ov e (x) = 0$ one has, $$ \ve \ra \lgl

402: \vp_{\ve} , x \rgl \ \hbox{is a polynomial in} \ \frac{1}{\ve} \

403: \hbox{without constant term.} \leqno (14) $$ Thus $\ve \ra

404: \vp_{\ve}$ extends to a map from $P_1 (\Cb) \backslash \{ 0 \}$ to

405: $G$, such that $$ \vp_{\ify} = 1 \, . \leqno (15) $$ For such a

406: map we define its {\it residue} as the derivative at $\ify$, i.e.~as,

407: $$ {\rm Res} \, \vp = \lim_{\ve \ra \ify} \ve (\vp_{\ve} - 1)

408: \, . \leqno (16) $$ By construction ${\rm Res} \, \vp \in L$ is a

409: derivation $\Hc \ra \Cb$. When evaluated on $x \in \Hc$, ${\rm

410: Res} \, \vp$ is just the residue at $\ve = 0$ of the function $\ve

411: \ra \lgl \vp_{\ve} , x \rgl$.

412: 

413: We shall now assume that for any $t \in \Rb$ the following limit

414: exists for any $x \in \Hc$, $$ \lim_{\ve \ra 0} \ \lgl

415: \vp_{\ve}^{-1} \, \t_{t\ve} (\vp_{\ve}) , x \rgl \, . \leqno (17)

416: $$ Using (10), (4) and (5) we have, $$ \lgl \vp_{\ve}^{-1} \,

417: \t_{t\ve} (\vp_{\ve}) , x \rgl = \lgl \vp_{\ve} \ot \vp_{\ve} , (S

418: \ot \t_{t\ve}) \, \D x \rgl \, , \leqno (18) $$ so that with $\D x

419: = \sum x_{(1)} \, \ot \, x_{(2)}$ we get a sum of terms $\lgl

420: \vp_{\ve} , S \, x_{(1)} \rgl \, \lgl \vp_{\ve} , \t_{t\ve} (

421: x_{(2)}) \rgl$ $= P_1 \left( \frac{1}{\ve} \right) \, e^{kt\ve} \,

422: P_2 \left( \frac{1}{\ve} \right)$. Thus (17) just means that the

423: sum of these terms is holomorphic at $\ve = 0$. It is clear that

424: the value at $\ve = 0$ is then a polynomial in $t$, $$ \lgl F_t ,

425: x \rgl = \lim_{\ve \ra 0} \ \lgl \vp_{\ve}^{-1} \, \t_{t\ve}

426: (\vp_{\ve}) , x \rgl \, . \leqno (19) $$ Let us check that $t \ra

427: F_t \in G$ is a one parameter group, $$ F_{t_1 + t_2} = F_{t_1} \,

428: F_{t_2} \qquad \fl \, t_i \in \Rb \, . \leqno (20) $$ The group

429: $G$ is a topological group for the topology of simple convergence,

430: i.e., $$ \vp_n \ra \vp \quad \hbox{iff} \quad \lgl \vp_n , x \rgl

431: \ra \lgl \vp , x \rgl \qquad \fl \, x \in \Hc \, . \leqno (21) $$

432: Moreover, using (10) one checks that $$ \t_{t_1 \ve}

433: (\vp_{\ve}^{-1} \, \t_{t_2 \ve} (\vp_{\ve})) \ra F_{t_2} \qquad

434: \hbox{when} \ \ve \ra 0 \, . \leqno (22) $$ We then have $F_{t_1 +

435: t_2}$ $=$ ${\displaystyle \lim_{\ve \ra 0}} \ \vp_{\ve}^{-1} \,

436: \t_{(t_1 + t_2) \ve} \, (\vp_{\ve})$ $=$ ${\displaystyle \lim_{\ve

437: \ra 0}} \ \vp_{\ve}^{-1} \, \t_{t_1 \ve} (\vp_{\ve}) \, \t_{t_1

438: \ve} (\vp_{\ve}^{-1} \, \t_{t_2 \ve} (\vp_{\ve}))$ $=$ $F_{t_1} \,

439: F_{t_2}$.

440: 

441: This proves (20) and we let, $$ \b = \left(

442: \frac{\partial}{\partial t} \, F_t \right)_{t=0} \leqno (23) $$

443: which defines an element of $L$ such that, $$ F_t = \exp (t\b)

444: \qquad \fl \, t \in \Rb \, . \leqno (24) $$ As above, we view

445: $\Hc^*$ as an algebra on which $Y$ acts as a derivation by (11).

446: Let us prove,

447: 

448: \medskip

449: 

450: \noindent {\bf Lemma 1.} {\it Let $\ve \ra \vp_{\ve} \in G$

451: satisfy $(17)$ with $\vp_{\ve} = 1 + {\displaystyle

452: \sum_{n=1}^{\ify}} \ \frac{d_n}{\ve^n}$, $d_n \in \Hc^*$. One then

453: has $$  \ Y

454: (d_1) = \b \qquad Y \, d_{n+1} = d_n \, \b \qquad \fl \, n \geq 1  . $$ }

455: 

456: \medskip

457: 

458: \noindent {\it Proof.} Let $x \in \Hc$ and let us show that $$

459: \lgl \b , x \rgl = \lim_{\ve \ra 0} \, \ve \lgl \vp_{\ve} \ot

460: \vp_{\ve} , (S \ot Y) \, \D (x) \rgl \, . \leqno (25) $$

461: 

462: Using (18) we know by hypothesis that, $$ \lgl \vp_{\ve} \ot

463: \vp_{\ve} , (S \ot \t_{t\ve}) \, \D (x) \rgl \ra \lgl F_t , x \rgl

464: \leqno (26) $$ where the convergence holds in the space of

465: holomorphic functions of $t$ in say $\vert t \vert \leq 1$ so that

466: the derivatives of both sides at $t = 0$ are also convergent thus

467: yielding (25).

468: 

469: Now the function $\ve \ra \ve \, \lgl \vp_{\ve} \ot \vp_{\ve} , (S

470: \ot Y) \, \D x \rgl$ is holomorphic for $\ve \in \Cb \backslash \{

471: 0 \}$ and also at $\ve = \ify \in P_1 (\Cb)$ since $\vp_{\ify} =

472: 1$. Moreover by (25) it is also holomorphic at $\ve = 0$ and is

473: thus a constant, which gives, $$ \lgl \vp_{\ve} \ot \vp_{\ve} , (S

474: \ot Y) \, \D (x) \rgl = \frac{1}{\ve} \ \lgl \b , x \rgl \, .

475: \leqno (27) $$ Using the product in $\Hc^*$ this means that $$

476: \vp_{\ve}^{-1} \, Y (\vp_{\ve}) = \frac{1}{\ve} \, \b \, , \leqno

477: (28) $$ and multiplying by $\vp_{\ve}$ on the left, that, $$ Y

478: (\vp_{\ve}) = \frac{1}{\ve} \ \vp_{\ve} \, \b \, . \leqno (29) $$

479: One has $Y (\vp_{\ve}) = {\displaystyle \sum_{n=1}^{\ify}} \

480: \frac{Y (d_n)}{\ve^n}$ and $\frac{1}{\ve} \, \vp_{\ve} \, \b =

481: \frac{1}{\ve} \, \b + {\displaystyle \sum_{n=1}^{\ify}} \

482: \frac{1}{\ve^{n+1}} \, d_n \, \b$. Thus (29) gives the lemma.~\xx

483: 

484: \bigskip

485: 

486: In particular we get $Y (d_1) = \b$ and since $d_1$ is the

487: residue, ${\rm Res} \, \vp$, this gives, $$ \b = Y \, ({\rm Res}

488: \, \vp) \, , \leqno (30) $$ which shows that $\b$ is uniquely

489: determined by the residue of $\vp_{\ve}$.

490: 

491: We shall now write a formula for $\vp_{\ve}$ in terms of $\b$.

492: This is made possible by Lemma~1  which shows that $\b$ uniquely

493: determines $\vp_{\ve}$. What is not transparent from Lemma~1 is

494: that for $\b \in L$ the elements $\vp_{\ve} \in \Hc^*$ are

495: group-like, so that $\vp_{\ve} \in G$. In order to obtain a nice

496: formula we take the semi direct product of $G$ by $\Rb$ acting on

497: $G$ by the grading $\t_t$, $$ \wt G = G \, \semi_{\t} \, \Rb \, ,

498: \leqno (31) $$ and similarly we let $\wt L$ be the Lie algebra $$

499: \wt L = L \op \Cb \, Z_0 \leqno (32) $$ where the Lie bracket is

500: given by $$ [Z_0 , \a ] = Y (\a) \qquad \fl \, \a \in L \leqno

501: (33) $$ and extends the Lie bracket of $L$.

502: 

503: We view $\wt L$ as the Lie algebra of $\wt G$ in a way which will

504: become clear in the proof of the following,

505: 

506: \medskip

507: 

508: \noindent {\bf Theorem 2.} {\it Let $\ve \ra \vp_{\ve} \in G$

509: satisfy $(17)$ as above. Then with $\b = Y ({\rm Res} \, \vp)$ one

510: has, $$ \vp_{\ve} = \lim_{t \ra \ify} e^{-tZ_0} \, e^{t \left(

511: \frac{\b}{\ve} + Z_0 \right)} \, . $$ }

512: 

513: \medskip

514: 

515: The limit holds in the topology of simple convergence in $G$. Both

516: terms $e^{-t Z_0}$ and $e^{t \left( \frac{\b}{\ve} + Z_0 \right)}$

517: belong to $\wt G$ but their product belongs to $G$.

518: 

519: \medskip

520: 

521: \noindent {\it Proof.} We endow  $\Hc^*$ with the topology of

522: simple convergence on $\Hc$ and let $\t_t$ act by automorphisms of

523: the topological algebra $\Hc^*$ by (10). Let us first show, with,

524: $$ \vp_{\ve} = 1 + \sum_{n=1}^{\ify} \ \frac{d_n}{\ve^n} \, ,

525: \quad d_n \in \Hc^* \, , \leqno (34) $$ that the following holds,

526: $$ d_n = \int_{s_1 \geq s_2 \geq \cdots \geq s_n \geq 0} \t_{-s_1}

527: (\b) \, \t_{-s_2} (\b) \ldots \t_{-s_n} (\b) \, \Pi  \, ds_i \, .

528: \leqno (35) $$ For $n = 1$, this just means that, $$ d_1 =

529: \int_0^{\ify} \t_{-s} (\b) \, ds \, , \leqno (36) $$ which follows

530: from (30) and the equality $$ Y^{-1} (x) =  \int_0^{\ify} \t_{-s}

531: (x) \, ds \qquad \fl \, x \in \Hc \, , \ \ov e (x) = 0 \, . \leqno

532: (37) $$ We see from (37) that for $\a , \a' \in \Hc^*$ such that

533: $$ Y (\a) = \a' \, , \ \lgl \a , 1 \rgl = \lgl \a' , 1 \rgl = 0

534: \leqno (38) $$ one has, $$ \a = \int_0^{\ify} \t_{-s} (\a') \, ds

535: \, . \leqno (39) $$ Combining this equality with Lemma~1 and the

536: fact that $\t_s \in {\rm Aut} \, \Hc^*$ is an automorphism, gives

537: an inductive proof of (35). The meaning of this formula should be

538: clear, we pair both sides with $x \in \Hc$, and let, $$ \D^{(n-1)}

539: \, x = \sum x_{(1)} \ot x_{(2)} \ot \cdots \ot x_{(n)} \, . \leqno

540: (40) $$ Then the right hand side of (35) is just, $$ \int_{s_1

541: \geq \cdots \geq s_n \geq 0}  \, \lgl \b \ot \cdots \ot \b \ , \

542: \t_{-s_1} (x_{(1)}) \ot \t_{-s_2} (x_{(2)}) \cdots \ot \t_{-s_n}

543: (x_{(n)}) \rgl  \Pi  \, ds_i \leqno (41) $$ and the convergence of

544: the multiple integral is exponential since, $$ \lgl \b , \t_{-s}

545: (x_{(i)}) \rgl = O \, (e^{-s}) \qquad \hbox{for} \quad s \ra +

546: \ify \, . \leqno (42) $$ We see moreover that if $x$ is

547: homogeneous of degree $\deg (x)$, and if $n

548: > \deg (x)$, at least one of the $x_{(i)}$ has degree 0 so that $\lgl \b ,

549: \t_{-s} (x_{(i)}) \rgl = 0$ and (41) gives 0. This shows that the

550: pairing of $\vp_{\ve}$ with $x \in \Hc$ only involves finitely

551: many non zero terms in the formula, $$ \lgl \vp_{\ve} , x \rgl =

552: \ov e (x) + \sum_{n=1}^{\ify} \frac{1}{\ve^n} \ \lgl d_n , x \rgl

553: \, . \leqno (43) $$ With all convergence problems out of the way

554: we can now proceed to prove the formula of Theorem~2 without care

555: for convergence.

556: 

557: Let us first recall the expansional formula \cite{Ar}, $$

558: e^{(A+B)} = \sum_{n=0}^{\ify} \, \int_{\sum u_j = 1 , \, u_j \geq

559: 0} \, e^{u_0 A} \, B e^{u_1 A} \ldots B e^{u_n A} \, \Pi  \, du_j

560: \leqno (44) $$ (cf.~\cite{Ar} for the exact range of validity of

561: (44)).

562: 

563: We apply this with $A = t Z_0$, $B = t \b$, $t > 0$ and get, $$

564: e^{t(\b + Z_0)} = \sum_{n=0}^{\ify} \ \int_{\sum v_j = t , \, v_j

565: \geq 0} \ e^{v_0 Z_0} \, \b e^{v_1 Z_0} \, \b \ldots \b e^{v_n

566: Z_0} \, \Pi  \, dv_j \, . \leqno (45) $$ Thus, with $s_1 = t -

567: v_0$, $s_1 - s_2 = v_1 , \ldots ,s_{n-1} - s_n = v_{n-1}$, $s_n =

568: v_n$ and replacing $\b$ by $\frac{1}{\ve} \, \b$, we obtain, $$

569: e^{t (\b / \ve + Z_0)} = \sum_{n=0}^{\ify} \ \frac{1}{\ve^n} \

570: \int_{t \geq s_1 \geq s_2 \geq \cdots \geq s_n \geq 0} \, e^{t

571: Z_0} \, \t_{-s_1} (\b) \ldots \t_{-s_n} (\b) \,  \Pi  \,ds_i .

572: \leqno (46) $$ Multiplying by $e^{-t Z_0}$ on the left and using

573: (41) thus gives, $$ \vp_{\ve} = \lim_{t \ra \ify} \ e^{-t Z_0} \,

574: e^{t (\b / \ve + Z_0)} \, . \leqno (47) $$ \hfill ~\xx

575: 

576: \bigskip

577: 

578: It is obvious conversely that this formula defines a family $\ve

579: \ra \vp_{\ve}$ of group-like elements of $\Hc^*$ associated to any

580: preassigned element $\b \in L$.

581: 

582: \medskip

583: 

584: \noindent {\bf Corollary 3.} {\it For any $\b \in L$ there exists

585: a (unique) map $\ve \ra \vp_{\ve} \in G$ satisfying $(17)$ and

586: $(34)$.}

587: 

588: \medskip

589: 

590: 

591: \section{The renormalization group flow}\label{secIII}

592: Let us now apply the above results to the group $G$ associated in

593: part~I to the Hopf algebra $\Hc$ of 1PI Feynman graphs of a

594: quantum field theory. We choose $\vp_6^3$ for simplicity. As

595: explained in part~I the group $G$ is a semi-direct product, $$ G =

596: G_0 \, \semi \, G_c \leqno (1) $$ of an abelian group $G_0$ by the

597: group $G_c$ associated to the Hopf subalgebra $\Hc_c$ constructed

598: on 1PI graphs with two or three external legs and fixed external

599: structure. Passing from $G_c$ to $G$ is a trivial step and we

600: shall thus concentrate on the group $G_c$. The unrenormalized

601: theory delivers, using dimensional regularization with the unit of

602: mass $\mu$, a loop, $$ \ve \ra \g_{\mu} (\ve) \in G_c \, , \leqno

603: (2) $$ and we first need to see the exact $\mu$ dependence of this

604: loop. We consider the grading of $\Hc_c$ and $G_c$ given by the

605: loop number of a graph, $$ L (\G) = I - V + 1 \leqno (3) $$ where

606: $I$ is the number of internal lines and $V$ the number of

607: vertices.

608: 

609: One has, $$ \g_{e^t \mu} (\ve) = \t_{t\ve} (\g_{\mu} (\ve)) \qquad

610: \fl \, t \in \Rb \, . \leqno (4) $$ Let us check this using the

611: formulas of section~3 of part~I. For $N=2$ external legs the

612: dimension $B$ of $\lgl \s , U_{\G} \rgl$ is equal to 0 by (12) of

613: loc.cit. Thus the $\mu$ dependence is given by $$

614: \mu^{\frac{\ve}{2} V_3} \leqno (5) $$ where $V_3$ is the number of

615: 3-point vertices of $\G$. One checks that $\frac{1}{2} \, V_3 = L$

616: as required. Similarly if $N = 3$ the dimension $B$ of $\lgl \s ,

617: U_{\G} \rgl$ is equal to $\left( 1 - \frac{3}{2} \right) \, d +

618: 3$, $d = 6 - \ve$ by (12) of loc.cit.~so that the $\mu$-dependence

619: is, $$ \mu^{\frac{\ve}{2} V_3} \, \mu^{-\ve / 2} \, . \leqno (6)

620: $$ But this time, $V_3 = 2L+1$ and we get $$ \mu^{\ve L}

621: \leqno (7) $$ as required.

622: 

623: We now reformulate a well known result, the fact that

624: counterterms, once appropriately normalized, are independent of

625: $m^2$ and $\mu^2$,

626: 

627: \medskip

628: 

629: \noindent {\bf Lemma 4.} {\it Let $\g_{\mu} = (\g_{\mu^-})^{-1} \,

630: (\g_{\mu^+})$ be the Birkhoff decomposition of $\g_{\mu}$. Then

631: $\g_{\mu^-}$ is independent of $\mu$.}

632: 

633: \medskip

634: 

635: As in part I we perform the Birkhoff decomposition with respect to

636: a small circle $C$ with center $D = 6$ and radius $<1$.

637: 

638: The proof of the lemma follows immediately from \cite{Col}. Indeed

639: the dependence in $m^2$ has in the minimal subtraction scheme the

640: same origin as the dependence in $p^2$ and we have chosen the

641: external structure of graphs (eq. (41) of part I) so that no $m^2$

642: dependence is left \footnote{This can be easily achieved by

643: maintaining non-vanishing  fixed external momenta.

644: $\gamma_{\mu^-}$ is independent on such external structures by

645: construction \cite{I}.}. But then, since $\mu^2$ is a dimensionful

646: parameter, it cannot be involved any longer.~\xx

647: 

648: \bigskip

649: 

650: \noindent {\bf Corollary 5.} {\it Let $\vp_{\ve} =

651: (\g_{\mu^-})^{-1} (\ve)$, then for any $t \in \Rb$ the following

652: limit exists in $G_c$, $$ \lim_{\ve \ra 0} \, \vp_{\ve}^{-1} \,

653: \t_{t\ve} (\vp_{\ve}) \, . $$ }

654: 

655: \medskip

656: 

657: In other words $\ve \ra \vp_{\ve} \in G_c$ fulfills the condition

658: (17) of section~2.

659: 

660: \medskip

661: 

662: \noindent {\it Proof.} The product $\vp_{\ve}^{-1} \, \g_{\mu}

663: (\ve)$ is holomorphic at $\ve = 0$ for any value of $\mu$. Thus by

664: (4), for any $t \in \Rb$, both $\vp_{\ve}^{-1} \, \g_{\mu} (\ve)$

665: and $\vp_{\ve}^{-1} \, \t_{\ve t} (\g_{\mu} (\ve))$ are

666: holomorphic at $\ve = 0$. The same holds for $\t_{-\ve t} \,

667: (\vp_{\ve}^{-1}) \, \g_{\mu} (\ve)$ and hence for the ratio $$

668: \vp_{\ve}^{-1} \, \g_{\mu} (\ve) \, (\t_{-\ve t} (\vp_{\ve}^{-1})

669: \, \g_{\mu} (\ve))^{-1} = \vp_{\ve}^{-1} \, \t_{-\ve t}

670: (\vp_{\ve}) \, . $$ \hfill ~\xx

671: 

672: \bigskip

673: 

674: We let $\g_- (\ve) = \vp_{\ve}^{-1}$ and translate the results of

675: section~2.

676: 

677: \medskip

678: 

679: \noindent {\bf Corollary 6.} {\it Let $F_t = {\displaystyle

680: \lim_{\ve \ra 0}} \, \g_- (\ve) \, \t_{t\ve} (\g_- \,

681: (\ve)^{-1})$. Then $F_t$ is a one parameter subgroup of $G_c$ and

682: $F_t = \exp (t\b)$ where $\b = Y \, {\rm Res} \, \vp_{\ve}$ is the

683: grading operator $Y$ applied to the residue of the loop $\g

684: (\ve)$.}

685: 

686: \medskip

687: 

688: In general, given a loop $\ve \ra \g (\ve) \in G$ it is natural to

689: define its {\it residue} at $\ve = 0$ by first performing the

690: Birkhoff decomposition on a small circle $C$ around $\ve = 0$ and

691: then taking, $$ {\rm Res}_{\ve = 0} \, \g =

692: \frac{\partial}{\partial \, u} \, (\vp_{1/u})_{u=0} \leqno (8) $$

693: where $\vp_{\ve} = \g_- (\ve)^{-1}$ and $\g (\ve) = \g_-

694: (\ve)^{-1} \, \g_+ (\ve)$ is the Birkhoff decomposition.

695: 

696: As shown in section~2, the residue or equivalently $\b = Y \, {\rm

697: Res}$ uniquely determines $\vp_{\ve} = \g_- (\ve)^{-1}$ and we thus

698: get, from Theorem~2,

699: 

700: \medskip

701: 

702: \noindent {\bf Corollary 7.} {\it The negative part $\g_- (\ve)$

703: of the Birkhoff decomposition of $\g_{\mu} (\ve)$ is independent

704: of $\mu$ and given by, $$ \g_- (\ve) = \lim_{t \ra \ify} \, e^{-t

705: \left( \frac{\b}{\ve} + Z_0 \right)} \, e^{t Z_0} \, . $$ }

706: 

707: \medskip

708: 

709: As above we adjoined the primitive element $Z_0$ to implement the

710: grading $Y$ (cf.~section~\ref{secII}). Our choice of the letter $\b$ is of

711: course not innocent and we shall see in section~5 the relation with the

712: $\b$-function.

713: 

714: \section{The action of $G_c$ on the coupling constants}\label{sec4}

715: We shall show in this section that the formula for the bare

716: coupling constant $g_0$ in terms of 1PI graphs, i.e. the

717: generating function, $$ g_0 = (g \, Z_1) \, (Z_3)^{-3/2} \leqno

718: (1) $$ where we consider the right hand side as a formal power

719: series with values in $\Hc_c$ given explicitly by, (with $\ell = L

720: (\G)$ the loop number of the graphs), $$ g_0=\left(

721: x+\sum_{\epsfysize=3mm\epsfbox{fig1a.eps}} x^{2l+1}

722: \frac{\Gamma}{S(\Gamma)} \right) \left(

723: 1-\sum_{\epsfysize=3mm\epsfbox{fig1b.eps}}  x^{2l}

724: \frac{\Gamma}{S(\Gamma)}\right)^{-3/2} \leqno (2) $$

725: 

726: %$$

727: %\hbox{ \epsfbox{Fig-1.eps} } \leqno \matrix{(2) \cr \cr \cr \cr}

728: %$$

729: %$$ g_0 = \left( x + \quad \sum \quad x^{2\ell + 1} \,

730: %\frac{\G}{S(\G)} \right) \left( 1 - \quad \sum \quad x^{2\ell} \,

731: %\frac{\G}{S(\G)} \right)^{-3/2} \leqno (2) $$

732:  does define a

733: Hopf algebra homomorphism, $$ \Phi : \Hc_{\rm CM} \ra \Hc \leqno (3) $$

734: from the Hopf algebra $\Hc_{\rm CM}$ of coordinates on the group

735: of formal diffeomorphisms of $\Cb$ with $$ \vp (0) = 0 \, ,

736: \ \vp' (0) = 1 \, , \leqno (4) $$ to the Hopf algebra $\Hc_c$ of

737: 1PI graphs.

738: 

739: This result is only valid if we perform on $\Hc_c$ the

740: simplification that pertains to the massless case, $m=0$, but

741: because of the $m$-independence of the counterterms all the

742: corollaries will be valid in general. The desired simplification

743: comes because in the case $m=0$ there is no need to indicate by a

744: cross left on an internal line the removal of a self energy

745: subgraph. Indeed and with the notations of part~I we can first of

746: all ignore all the ${\epsfysize=2.5mm\epsfbox{fig1b.eps}}_{(0)}$

747: since $m^2 = 0$, moreover the

748: $\raisebox{-1mm}{\epsfysize=4mm\epsfbox{fig-3.eps}}$ yield a $k^2$

749: term which exactly cancels out with the additional propagator when

750: we remove the subgraph and replace it by $

751: \epsfysize=3mm\epsfbox{fig-4.eps}$. This shows that we can simply

752: ignore all these crosses and write coproducts in the simplest

753: possible way. To get familiar with this coproduct and with the

754: meaning of the Hopf algebra morphism (3) we urge the reader to

755: begin by the concrete computation done in the appendix, which

756: checks its validity up to order six in the coupling constant.

757: 

758: Let us now be more explicit on the meaning of formula (2). We

759: first expand $g_0$ as a power series in $x$ and get a series of

760: the form, $$ g_0 = x + \sum_2^{\ify} \a_{n} \, x^{n} \leqno

761: (5) $$ where the even coefficients $\a_{2n}$ are zero and the

762: coefficients $\a_{2n+1}$ are finite linear

763: combinations of products of graphs, so that, $$ \a_{2n+1} \in \Hc

764: \qquad \fl \, n \geq 1 \, . \leqno (6) $$ We let $\Hc_{\rm CM}$ be the Hopf algebra

765:  of the group of formal diffeomorphisms such that (4) holds. We take the

766: generators $a_n$ of $\Hc_{\rm CM}$ given by the equality $$ \vp (x) = x + \sum_{n \geq 2} a_n (\vp)

767: \, x^n \, . \leqno (7) $$ and define the coproduct in $\Hc_{\rm CM}$

768: by the equality $$   \lgl \D a_n  \, , \, \vp_1 \ot \vp_2   \rgl = a_n(\vp_2 \circ \vp_1)\leqno (8) $$

769: We then define uniquely the algebra homomorphism

770: $$ \Phi : \Hc_{\rm CM} \ra \Hc $$ by the condition,

771: $$ \Phi (a_n) =\a_{n}  \, . \leqno (9) $$  By construction

772: $\Phi$ is a morphism of algebras. We shall show that it is

773: comultiplicative, i.e. $$ (\Phi \ot \Phi) \, \D \, x = \D \, \Phi

774: (x) \qquad \fl \, x \in \Hc_{\rm CM} \leqno (10) $$ and comes from

775: a group morphism, $$ \rho : G_c \ra G_2 \, . \leqno (11) $$ where $G_2$ is the group

776: of characters of $\Hc_{\rm CM}$ which is by construction the opposite of the group

777: of formal diffeomorphisms. In

778: fact we shall first describe the corresponding Lie algebra

779: morphism, $ \rho$.

780: 

781: Let us first recall from part I that a 1PI graph $\G$ defines a

782: primitive element $(\G)$

783:  of $\Hc_{\rm gr}^*$ which only pairs

784: nontrivially with the monomial $\G$ of $\Hc$ and satisfies $\lgl (\G) \, ,\, \G \rgl =1$.

785:  We take the

786: following natural basis $\underline \G =S(\G) (\G)$ for the Lie

787: algebra of primitive elements of $\Hc_{\rm gr}^*$, labelled by 1PI

788: graphs with two or three external legs. By part I, theorem 2, their Lie

789: bracket is given by, $$ [\underline\G , \underline\G'] = \sum_{v'}

790: \underline{\G' \circ^{v'} \G} - \sum_v \underline{\G \circ^v \G' }

791: \, , \leqno (12) $$ where $\G' \circ_{v'} \G $ is the graph

792: obtained by grafting $ \G $ at the vertex $v'$ of $\G' $. (Our

793: basis differs from the one used in loc.~cit.~by an overall - sign,

794: but the present choice will be more convenient). In our context of

795: the simplified Hopf algebra the places where a given graph $\G$

796: can be inserted in another graph $\G'$ are no longer always

797: labelled by vertices of $\G'$. They are when $\G$ is a vertex

798: graph but when $\G$ is a self energy graph such places are just

799: labelled by the internal lines of $\G'$, as we could discard the

800: use of external structures and two-point vertices for self energy

801: graphs.

802: 

803: We also let $Z'_n$ be the natural basis of the Lie algebra of

804: primitive elements of $\Hc_{\rm CM}^*$ which corresponds to

805:  the vector fields $ x^{n+1} \ \frac{\partial}{\partial x}$.

806:  More precisely, $Z'_n$ is given as the linear form on $\Hc_{\rm CM}$ which only pairs

807: with the monomial $a_{n+1}$, $$\lgl Z'_n \, , \, a_{n+1} \rgl \, = 1 ,\leqno (13) $$

808: and the Lie bracket is given by, $$ [Z'_n ,

809: Z'_m] = (m-n) \, Z'_{n+m} \, . \leqno (14) $$ We then first prove,

810: 

811: \medskip

812: 

813: \noindent {\bf Lemma 8.} {\it  Let $\rho_{\G} = \frac{3}{2}$ for

814: $2$-point graphs and $\rho_{\G}=1$ for $3$-point graphs. The

815: equality $\rho (\underline\G) = \rho_{\G} \, Z'_{2\ell}$, where

816: $\ell = L (\G)$ is the loop number, defines a Lie algebra

817: homomorphism.}

818: 

819: \medskip

820: 

821: \noindent {\it Proof.} We just need to show that $\rho$ preserves

822: the Lie bracket. Let us first assume that $\G_1 , \G_2$ are vertex

823: graphs and let $V_i$ be the vertex number of $\G_i$. One has, $$ V

824: = 2L + 1 \leqno (15) $$ for any vertex graph $\G$. Thus the Lie

825: bracket $\rho \, [\underline\G_1 , \underline\G_2]$ provides $V_2

826: - V_1 = 2 (L_2 - L_1)$ vertex graph contributions all equal to

827: $Z'_{2 (L_1 + L_2)}$ so that $$ \rho \, ([\underline\G_1 ,

828: \underline\G_2]) = 2 (L_2 - L_1) \, Z'_{2 (L_1 + L_2)} \leqno (16)

829: $$ which is exactly $[\rho \, (\underline\G_1) , \rho \,

830: (\underline\G_2)]$ by (14).

831: 

832: Let then $\G_1$ and $\G_2$ be 2-point graphs. For any such graph

833: one has, $$ I = 3L - 1 \leqno (17) $$ where $I$ is the number of

834: internal lines of $\G$. Thus $\rho \, ([\underline\G_1 ,

835: \underline\G_2])$ gives $I_2 - I_1 = 3 (L_2 - L_1)$ 2-point graph

836: contributions, each equal to $\frac{3}{2} \ Z'_{2(L_1 + L_2)}$.

837: Thus, $$ \rho \, ([\underline\G_1 , \underline\G_2]) = \frac{3}{2}

838: \ 3 (L_2 - L_1) \, Z'_{2 (L_1 + L_2)} \leqno (18) $$ but the right

839: hand side is $\rho_{\G_1} \, \rho_{\G_2} \, 2 (L_2 - L_1) \, Z'_{2

840: (L_1 + L_2)}$ so that, $$ \rho \, ([\underline\G_1 ,\underline

841: \G_2]) = [\rho \, (\underline\G_1) , \rho \, (\underline\G_2)] \,

842: , \leqno (19) $$ as required.

843: 

844: Finally if say $\G_1$ is a 3-point graph and $\G_2$ a 2-point

845: graph, we get from $[\underline\G_1 , \underline\G_2]$ a set of

846: $V_2$ 2-point graphs minus $I_1$ 3-point graphs which gives, $$

847: \left( \frac{3}{2} \, V_2 - I_1 \right) \, Z'_{2 (L_1 + L_2)} \, .

848: \leqno (20) $$ One has $V_2 = 2 \, L_2$, $I_1 = 3 \, L_1$ so that

849: $\frac{3}{2} \, V_2 - I_1 = 3 (L_2 - L_1) = \rho_{\G_1} \,

850: \rho_{\G_2} \, 2 (L_2 - L_1)$ which gives (19) as required.~\xx

851: 

852: \bigskip

853: 

854: We now have the Lie algebra morphism $\rho$ and the algebra

855: morphism $\Phi$. To $\rho$ corresponds a morphism of groups, $$

856: \rho : G_c \ra G_2 \leqno (21) $$ and we just need to check that

857: the algebra morphism $\Phi$ is the transposed of $\rho$ on the

858: coordinate algebras, $$ \Phi (a) = a \circ \rho \qquad \fl \, a

859: \in \Hc_{\rm CM} \, . \leqno (22) $$ To prove (22) it is enough to

860: show that $\Phi$ is equivariant with respect to the action of the

861: Lie algebra $L$ of primitive elements of $\Hc_{\rm gr}^*$. More

862: precisely, given a primitive element $$ Z \in \Hc_{\rm gr}^* \ , \

863: \D Z = Z \ot 1 + 1 \ot Z \, , \leqno (23) $$ we let $\partial_Z$

864: be the derivation of the algebra $\Hc$ given by, $$

865: \partial_Z (y) = \lgl Z \ot {\rm id} , \D y \rgl \in \Hc \qquad \fl \, y

866: \in \Hc \, . \leqno (24) $$ What we need to check is the

867: following,

868: 

869: \medskip

870: 

871: \noindent {\bf Lemma 9.} {\it For any $a \in \Hc_{\rm CM}$, $Z \in

872: L$ one has $\partial_Z \, \Phi (a) = \Phi (\partial_{(\rho Z)}

873: (a))$.}

874: 

875: \medskip

876: 

877: \noindent {\it Proof.} It is enough to check the equality when $Z$

878: is of the form $\underline\G=S(\G) (\G)$ with the above notations.

879: Thus we let $\G$ be a 1PI graph and let $\partial_{\G}$ be the

880: corresponding derivation of $\Hc$ given by (24) with $Z = S(\G)

881: (\G)$. Now by definition of the primitive element $(\G)$ one has

882: (cf. (48) section 2 of part~I), $$ \lgl (\G) \ot {\rm id} , \D \G' \rgl = \sum n (\G

883: , \G'' ; \G') \, \G'' \leqno (25) $$ where the integer $n (\G ,

884: \G'' ; \G')$ is the number of subgraphs of $\G'$ which are

885: isomorphic to $\G$ while $\G' / \G \cong \G''$. By Theorem~2 of

886: part~I we have, $$ S (\G) \, S (\G'') \, n (\G , \G'' ; \G') = i

887: (\G , \G'' ; \G') \, S (\G') \leqno (26) $$ where $i (\G , \G'' ;

888: \G')$ is the number of times $\G'$ appears in $\G'' \circ \G$. We

889: thus get, $$

890: \partial_{\G} \, \frac{\G'}{S (\G')} = \sum i (\G , \G'' ; \G') \,

891: \frac{\G''}{S (\G'')} \leqno (27) $$ which shows that

892: $\partial_{\G}$ admits a very simple definition in the generators

893: $\frac{\G'}{S (\G')}$ of $\Hc$.

894: 

895: The derivation $\partial_{(\rho Z)}$ of $\Hc_{\rm CM}$ is also

896: very easy to compute. One has by construction, (Lemma~8), $$ \rho

897: (Z) = \rho_{\G} \, Z'_{2\ell} \qquad \ell = L (\G) \leqno (28) $$

898: and the derivation $d_k$ of $\Hc_{\rm CM}$ associated to the

899: primitive element $Z'_k$ of $\Hc_{\rm CM}^*$ is simply given, in

900: the basis $a_n \in \Hc_{\rm CM}$ by, $$ d_k (a_n) = (n-k) \,

901: a_{n-k} \, . \leqno (29) $$ We thus get, $$

902: \partial_{(\rho Z)} = \rho_{\G} \, d_{2\ell} \ , \quad \ell = L (\G) \, .

903: \leqno (30) $$ Now by construction both $\partial_Z \, \Phi$ and

904: $\Phi \, \circ \,

905: \partial_{(\rho Z)}$ are derivations from the algebra $\Hc_{\rm CM}$ to $\Hc$ viewed as a

906:  bimodule over $\Hc_{\rm CM}$, i.e. satisfy,

907: $$ \d (ab) = \d (a) \, \Phi (b) + \Phi (a) \, \d (b) \, . \leqno

908: (31) $$ Thus, to prove the lemma we just need to check the

909: equality, $$

910: \partial_{\G} \, \Phi (a_n) = \rho_{\G} \, \Phi (d_{2\ell} (a_n))

911: \qquad \ell = L(\G) \leqno (32) $$ or equivalently using the

912: generating function $$ g_0 = x + \sum \Phi (a_n) \, x^n \, ,

913: \leqno (33) $$ that $$

914: \partial_{\G} \, g_0 = \rho_{\G} \, x^{2\ell + 1} \,

915: \frac{\partial}{\partial x} \, g_0 \, . \leqno (34) $$ Now by

916: construction of $\Phi$ we have, $$ g_0 = (x \, Z_1) (Z_3)^{-3/2} \

917: , \ Z_3 = 1 - \d Z \leqno (35) $$ where, $$

918: Z_1=1+\sum_{\epsfysize=3mm\epsfbox{fig1a.eps}}

919: x^{2l}\frac{\Gamma}{S(\Gamma)},\;\delta Z=

920: \sum_{\epsfysize=3mm\epsfbox{fig1b.eps}}

921: x^{2l}\frac{\Gamma}{S(\Gamma)} \leqno (36) $$

922: %$$ \hbox{

923: %\epsfbox{Fig-5.eps} } \leqno \matrix{(36) \cr \cr \cr \cr}

924: %$$

925: %$$ Z_1 = \quad \sum \quad x^{2\ell} \, \frac{\G}{S(\G)} \ , \ \d Z

926: %= \quad \sum \quad x^{2\ell} \, \frac{\G}{S(\G)} \, . \leqno (36)

927: %$$

928: Thus, since both $\partial_{\G}$ and $\frac{\partial}{\partial x}$

929: are derivations we can eliminate the denominators in (34) and

930: rewrite the desired equality, after multiplying both sides by

931: $(1-\d Z)^{5/2}$ as, $$ \ldisplaylinesno{ \left( x \,

932: \frac{\partial}{\partial \G} \, Z_1 \right) (1-\d Z) + \frac{3}{2}

933: \left( \frac{\partial}{\partial \G} \, \d Z \right) (x \, Z_1)

934: &(37)  \cr = \rho_\Gamma \left( x^{2\ell + 1} \left(

935: \frac{\partial}{\partial x} \, (x \, Z_1) \right) (1-\d Z) +

936: \frac{3}{2} ( x \, Z_1) \, x^{2\ell + 1} \,

937: \frac{\partial}{\partial x} \, \d Z \,\right) . \cr } $$ Both

938: sides of this formula are bilinear expressions in the 1PI graphs.

939: We first need to compute $\frac{\partial}{\partial \G} \, Z_1$ and

940: $\frac{\partial}{\partial \G} \, \d Z$. One has $$

941: \frac{\partial}{\partial\Gamma}Z_1=\sum_{\epsfysize=3mm

942: \epsfbox{fig1a.eps}}x^{2l+2l^\prime}c(\Gamma,\Gamma^\prime)

943: \frac{\Gamma^\prime}{S(\Gamma^\prime)}\leqno (38) $$

944: %$$ \hbox{

945: %\epsfbox{Fig-6.eps} } \leqno \matrix{(38) \cr \cr \cr \cr}

946: %$$

947: %$$ \frac{\partial}{\partial \G} \, Z_1 = \quad \sum \quad x^{2\ell

948: %+ 2\ell'} \, c (\G , \G') \, \frac{\G'}{S(\G')} \leqno (38) $$

949: and, $$ \frac{\partial}{\partial\Gamma}\delta

950: Z=\sum_{\epsfysize=3mm

951: \epsfbox{fig1b.eps}}x^{2l+2l^\prime}c(\Gamma,\Gamma^\prime)

952: \frac{\Gamma^\prime}{S(\Gamma^\prime)},\leqno (39) $$

953: %$$ \hbox{ \epsfbox{Fig-7.eps} } \leqno \matrix{(39) \cr \cr

954: %\cr \cr} $$

955: %$$ \frac{\partial}{\partial \G} \, \d Z = \quad

956: %\sum \quad x^{2\ell + 2\ell'} \, c (\G , \G') \,

957: %\frac{\G'}{S(\G')} \, , \leqno (39) $$

958: where $\ell = L(\G)$, $\ell' = L (\G')$ are the loop numbers and

959: the integral coefficient $c (\G , \G')$ is given by, $$ c(\G,\G')

960: = V' \ \hbox{if} \ \rho_{\G} = 1 \quad \hbox{and} \quad c(\G,\G')

961: = I' \ \hbox{if} \ \rho_{\G} = 3/2 \leqno (40) $$ (where $V'$ and

962: $I'$ are respectively the number of vertices and of internal lines

963: of $\G'$).

964: 

965: To prove (38) and (39) we use (27) and we get in both cases

966: expressions like (38), (39) with $$ c(\G,\G') = \sum_{\G''} \ i

967: (\G , \G' ; \G'') \, . \leqno (41) $$ But this is exactly the

968: number of ways we can insert $\G$ inside $\G'$ and is thus the

969: same as (40). Let now $\G_1$ be a 3-point graph and $\G_2$ a

970: 2-point graph. The coefficient of the bilinear term, $$

971: \frac{\G_1}{S(\G_1)} \ \frac{\G_2}{S(\G_2)} \, , \leqno (42) $$ in

972: the left hand side of (37) is given by $$ \left( \frac{3}{2} \ c

973: (\G , \G_2) - c (\G , \G_1) \right) \, x^{2\ell + 2\ell_1 +

974: 2\ell_2 + 1} \, . \leqno (43) $$ Its coefficient in the right hand

975: side of (37) is coming from the terms, $$ \ldisplaylinesno{

976: x^{2\ell + 1} \, \frac{\partial}{\partial x} \left( x^{2\ell_1 +

977: 1} \, \frac{\G_1}{S(\G_1)} \right) \left( \frac{-\G_2}{S(\G_2)}

978: \right) \, x^{2\ell_2} \, + &(44)  \cr \frac{3}{2} \ x^{2\ell_1 +

979: 1} \, \frac{\G_1}{S(\G_1)} \, x^{2\ell + 1} \,

980: \frac{\partial}{\partial x} \left( x^{2\ell_2} \,

981: \frac{\G_2}{S(\G_2)} \right) \cr } $$ which gives, $$ (3 \ell_2 -

982: 2 \ell_1 - 1) \, x^{1+2\ell + 2\ell_1 + 2 \ell_2} \, . \leqno (45)

983: $$ We thus only need to check the equality, $$ \frac{3}{2} \ c (\G

984: , \G_2) - c (\G , \G_1) = (3 \ell_2 - 2\ell_1 - 1) \, \rho_{\G} \,

985: . \leqno (46) $$ Note that in general, for any graph with $N$

986: external legs we have, $$ V = 2 (L-1) + N \, , \ I = 3 (L-1) + N

987: \, . \leqno (47) $$ Let us first take for $\G$ a 3-point graph so

988: that $\rho_{\G} = 1$. Then the left hand side of (46) gives

989: $\frac{3}{2} \, V_2 - V_1 = \frac{3}{2} \, (2\ell_2) - (2 (\ell_1

990: - 1) + 3) = 3\ell_2 - 2\ell_1 - 1$.

991: 

992: Let then $\G$ be a 2-point graph, i.e. $\rho_{\G} = \frac{3}{2}$.

993: Then the left hand side of (46) gives $\frac{3}{2} \, I_2 - I_1 =

994: \frac{3}{2} \, (3\ell_2 - 1) - 3\ell_1 = \rho_{\G} \, (3 \ell_2 -

995: 2\ell_1 - 1)$ which gives the desired equality.

996: 

997: Finally we also need to check the scalar terms and the terms

998: linear in $\G_1$ or in $\G_2$. The only scalar terms in the left

999: hand side of (37) are coming from $x \, \frac{\partial}{\partial

1000: \G} \, Z_1 + \frac{3}{2} \, x \, \frac{\partial}{\partial \G} \,

1001: \d Z$ and this gives, $$ x^{2\ell + 1} \, \rho_{\G} \, . \leqno

1002: (48) $$ The only scalar term in the right hand side of (37) comes

1003: from $x^{2\ell + 1}$ thus they fulfill (37).

1004: 

1005: The terms linear in $\G_1$ in the left hand side of (37) come only

1006: from $x \, \frac{\partial}{\partial \G} \, Z_1$ if $\G$ is a

1007: 3-point graph and the coefficient of $\G_1 / S (\G_1)$ is thus, $$

1008: c (\G , \G_1) \, x^{1 + 2\ell_1 + 2\ell} \, . \leqno (49) $$ In

1009: the right hand side of (37) we just get $$ (2\ell_1 + 1) \, x^{1 +

1010: 2\ell_1 + 2\ell} \, . \leqno (50) $$ We thus need to check that $c

1011: (\G , \G_1) = 2\ell_1 + 1$ which follows from (40) and (47) since

1012: $V_1 = 2\ell_1 + 1$.

1013: 

1014: Similarly, if $\G$ is a 2-point graph the left side of (37) only

1015: contributes by $x \, \frac{\partial}{\partial \G} \, Z_1 +

1016: \frac{3}{2} \, x^{2\ell + 1} \, Z_1$, so that the coefficient of

1017: $\G_1 / S(\G_1)$ is $$ \left(c (\G , \G_1) + \frac{3}{2} \right)

1018: \, x^{1+2\ell + 2\ell_1} \, . \leqno (51) $$ In the right hand

1019: side of (37) we get just as above $$ (2\ell_1 + 1) \, x^{1+2\ell +

1020: 2\ell_1} \leqno (52) $$ multiplied by $\rho_{\G} = 3/2$.

1021: 

1022: Now here, since $\G$ is a 2-point graph, we have $c (\G , \G_1) =

1023: I_1 = 3 (\ell_1 - 1) + 3 = 3 \ell_1$ so that $$ c (\G , \G_1) +

1024: \frac{3}{2} = \frac{3}{2} \, (2\ell_1 + 1) = \rho_{\G} (2\ell_1 +

1025: 1) $$ as required.

1026: 

1027: The check for terms linear in $\G_2$ is similar.~\xx

1028: 

1029: \bigskip

1030: 

1031: We can now state the main result of this section,

1032: 

1033: \medskip

1034: 

1035: \noindent {\bf Theorem 10.} {\it The map $\Phi = \Hc_{\rm CM} \ra

1036: \Hc$ given by the effective coupling is a Hopf algebra

1037: homomorphism. The transposed Lie group morphism is $\rho : G_c \ra

1038: G_2$.}

1039: 

1040: \medskip

1041: 

1042: The proof follows from Lemma 9 which shows that the map from $G_c$ to $G_2$

1043: given by the transpose of the algebra morphism $\Phi$ is the Lie

1044: group morphism $\rho$. By construction the morphism

1045: $\Phi$ is compatible with the grading $\Theta$ of ${\cal H}$

1046: and $\alpha$ of ${\cal H}_{CM}$ given by ${\rm deg}(a_n)=n-1$

1047: (cf.~\cite{CM}), one has indeed,

1048: $$ \Phi\circ\alpha_t=\Theta_{2t}\circ\Phi,\;\forall t\in \Rb. \leqno (51)

1049: $$

1050: Finally we remark that our proof of Theorem 10

1051: is similar to the proof of the equality $$

1052: F_{\phi_1\phi_2}=F_{\phi_2}\circ F_{\phi_1}\leqno (52)$$

1053: for the Butcher series used in the numerical integration of differential equations, but that the presence

1054: of the $Z_3$ factor makes it much more involved in our case.

1055: 

1056: 

1057: 

1058: 

1059: \section{The $\beta$-function and the

1060: Birkhoff decomposition of the unrenormalized effective

1061: coupling in the diffeomorphism group}\label{sec5}

1062: 

1063: 

1064: 

1065: 

1066: Let us first recall our notations from part~I concerning the

1067: effective action. We work in the Euclidean signature of space time

1068: and in order to minimize the number of minus signs we write the

1069: functional integrals in the form, $$ {\cal N} \int e^{S(\vp)} \, P

1070: (\vp) \, [D \vp] \leqno (1) $$ so that the Euclidean action

1071: is\footnote{We know of course that the usual sign convention is

1072: better to display the positivity of the action functional.}, $$ S

1073: (\vp) = - \frac{1}{2} \, (\partial_{\mu} \, \vp)^2 - \frac{1}{2}

1074: \, m^2 \, \vp^2 + \frac{g}{6} \, \vp^3 \, . \leqno (2) $$ The

1075: effective action, which when used at tree level in (1) gives the

1076: same answer as the full computation using (2), is then given in

1077: dimension $d = 6 - \ve$ by, $$ \ldisplaylinesno{ S_{\rm eff} (\vp)

1078: = S (\vp) + \sum_{{\rm 1PI}} \, \frac{(\mu^{\ve / 2} \,

1079: g)^{n-2}}{n!}  \int \, \frac{\G}{S(\G)} \ (p_1 , \ldots , p_n)

1080: &(3)\cr \times \vp (p_1) \ldots \vp (p_n) \, \Pi \, dp_i \cr } $$

1081: where, as in part~I, we do not consider tree graphs as 1PI and the

1082: integral is performed on the hyperplane $\sum p_i = 0$. To be more

1083: precise one should view the right hand side of (3) as a formal

1084: power series  with values in the Hopf algebra $\Hc$. The theory

1085: provides us with a loop $$ \g_{\mu} (\ve) = \g_- (\ve)^{-1} \,

1086: \g_{\mu_+} (\ve) \leqno (4) $$ of characters of $\Hc$. When we

1087: evaluate $\g_{\mu} (\ve)$ (resp.~$\g_-$, $\g_{\mu^+}$) on the

1088: right hand side of (3) we get respectively the unrenormalized

1089: effective action, the bare action and the renormalized effective

1090: action (in the MS scheme).

1091: 

1092: Our notation is hiding the $g$-dependence of $\g_{\mu} (\ve)$, but this

1093: dependence

1094:  is entirely governed by the grading. Indeed with $t= \log(g)$

1095: one has, with obvious notations,

1096: $$

1097: \g_{\mu,g} (\ve) = \Theta_{2t}(\g_{\mu,1} (\ve)) \leqno (5)

1098: $$

1099: Since $\Theta_{t}$ is an automorphism the same equality holds for both $\g_{\mu_+}$ and $\g_{-}$.

1100: 

1101: As in section 4 we restrict ourselves

1102: to the massless case  and

1103: let $\g_{\mu} (\ve) = \g_- (\ve)^{-1} \, \g_{\mu_+}(\ve)$

1104:  be the Birkhoff decomposition of $\g_{\mu} (\ve)=\g_{\mu,1} (\ve)$.

1105: \medskip

1106: 

1107: \noindent {\bf Lemma 11.} {\it Let $\rho : G_c \ra G_2$ be given by Theorem~10. Then

1108: $\rho(\g_{\mu} (\ve))(g)$ is the unrenormalized  effective coupling constant,

1109: $\rho(\g_{\mu_+} (0))(g)$ is the renormalized  effective coupling constant and

1110: $\rho(\g_{-} (\ve))(g)$ is the bare coupling constant $g_0$.}

1111: 

1112: \medskip

1113: 

1114: This follows from (3) and Theorem~10. It is now straightforward to translate the results

1115:  of the previous sections in terms of diffeomorphisms. The only subtle point to remember is that

1116: the group $G_2$ is the opposite of the group of diffeomorphisms so that if we view $\rho$ as a map

1117:  to diffeomorphisms it is an antihomomorphism,

1118: $$ \rho(\g_1 \, \g_2)=\, \rho(\g_2) \circ \rho(\g_1)\, . \leqno

1119: (6) $$

1120: \medskip

1121: 

1122: \noindent {\bf Theorem 12.} {\it The renormalization group flow is the

1123: image $\rho (F_{t})$ by $\rho : G_c \ra {\rm Diff}$

1124: of the one parameter group $F_{t} \in

1125: G_c$.}

1126: \medskip

1127: 

1128: \noindent {\it Proof.} The {\it bare coupling constant}  $g_0$ governs the bare

1129: action,

1130: $$

1131:  S_{\rm bare} (\vp_0) = -\frac{1}{2} \, (\partial_{\mu} \, \vp_0)^2 + \mu^{\ve /2} \

1132: \frac{g_0}{6} \, \vp_0^3 \leqno (7)

1133: $$

1134: in terms of the bare field $\vp_0$.

1135: 

1136: Now when we replace $\mu$ by

1137: $$

1138: \mu' = e^t \, \mu \leqno (8)

1139: $$

1140: we can keep the bare action, and hence the physical theory, unchanged

1141: provided we replace the

1142: renormalized coupling constant $g$ by $g'$ where

1143: $$

1144: g_0 (\ve , g') = e^{-\ve \, \frac{t}{2}} \, g_0 (\ve , g) \, . \leqno

1145: (9)

1146: $$

1147: By construction we have,

1148: $$

1149: g' = \psi_{\ve}^{-1} (e^{-\ve \frac{t}{2}} \, \psi_{\ve}(g)) \leqno (10)

1150: $$

1151: where $\psi_{\ve}$ is the formal diffeomorphism given by

1152:  $$\psi_{\ve} = \,\rho \, (\g_- (\ve)) \, . \leqno (11)$$

1153:  Now the behaviour for

1154: $\ve \ra 0$ of $g'$ given by (10) is the same as for,

1155: $$

1156: \psi_{\ve}^{-1} \, \a_{\ve t/2} (\psi_{\ve}) \leqno (12)

1157: $$

1158: where $\a_s$ is the grading of ${\rm Diff}$ given as above by

1159: $$

1160: \a_s (\psi)(x) = e^{-s }\, \psi (e^{s}x) \, . \leqno (13)

1161: $$

1162: Thus, since the map $\rho$ preserves the grading,

1163: $$

1164: \rho \, (\t_t (\g)) = \a_{t/2} \, \rho (\g) \leqno (14)

1165: $$

1166: (by (51) of section~4), we see by Corollary~6 of section~3 that,

1167: $$

1168: g' \ra \rho (F_{t})g \quad \hbox{when} \ \ve \ra 0 \, . \leqno (15)

1169: $$ ~\xx

1170: 

1171: 

1172: 

1173: 

1174: \medskip

1175: 

1176: As a corollary we get of course,

1177: 

1178: \medskip

1179: 

1180: \noindent {\bf Corollary 13.} {\it The image by $\rho$ of $\b \in L$ is

1181: the $\b$-function of the theory.}

1182: 

1183: \medskip

1184: 

1185: In fact all the results of section~3 now translate to the group $G_2$.

1186: We get the formula for the bare coupling constant in terms of

1187: the $\b$-function, namely,

1188: $$

1189: \psi_{\ve} = \lim_{t \ra \ify} e^{-t Z_0} \, e^{t \left( \frac{\b}{\ve}

1190: + Z_0 \right)} \leqno (16)

1191: $$

1192: where $Z_0 = x \ \frac{\partial}{\partial x}$ is the generator of

1193: scaling. But we can also express the main result of part~I independently

1194: of the group $G$ or of its Hopf algebra $\Hc$. Indeed the group

1195: homomorphism $\rho : G \ra G_2$ maps the Birkhoff decomposition of

1196: $\g_{\mu} (\ve)$ to the Birkhoff decomposition of $\rho (\g_{\mu} (\ve))$.

1197: But we saw above that $\rho \, (\g_{\mu} (\ve))$ is just the {\it

1198: unrenormalized} effective coupling constant. We can thus state,

1199: 

1200: \medskip

1201: 

1202: \noindent {\bf Theorem 14.} {\it Let the unrenormalized effective

1203: coupling constant $g_{\rm eff} (\ve)$  viewed as a formal power series

1204: in $g$ be considered as a loop of formal

1205: diffeomorphisms

1206: and let $g_{\rm eff} (\ve)= (g_{{\rm eff}_-})^{-1}(\ve) \, g_{{\rm eff}_+}(\ve)$

1207: be its Birkhoff decomposition in the group  of formal

1208: diffeomorphisms. Then the loop $g_{{\rm eff}_-} (\ve)$ is the bare

1209: coupling constant and $g_{{\rm eff}_+} (0)$ is the renormalized

1210: effective coupling.}

1211: 

1212: \medskip

1213: 

1214: Note that $G_2$ is naturally isomorphic to the opposite group of Diff so we

1215: used the opposite order in the Birkhoff decomposition.

1216: 

1217: This result is very striking since it no longer involves the Hopf algebra

1218: $\Hc$ or the group $G$ but only the idea of thinking of the effective

1219: coupling constant as a formal diffeomorphism. The proof is immediate, by

1220: combining Lemma~11, Theorem~10 of section 4 with Theorem~4 of part I.

1221: 

1222: Now in the same way as the Riemann-Hilbert problem and the Birkhoff

1223: decomposition for the group $G = {\rm GL} (n,\Cb)$ are intimately related

1224: to the classification of holomorphic $n$-dimensional vector bundles on $P_1

1225: (\Cb) = C_+ \cup C_-$, the Birkhoff decomposition for the group $G_2 = {\rm

1226: Diff}^0$ is related to the classification of one dimensional complex (non

1227: linear) bundles,

1228: $$

1229: P = (C_+ \ts X) \cup_{g_{\rm eff}} (C_- \ts X) \, . \leqno (17)

1230: $$

1231: Here $X$ stands for a formal one dimensional fiber and $C_{\pm}$ are, as in

1232: part I, the components of the complement in $P_1 (\Cb)$ of a small circle

1233: around $D$. The total space $P$ should be thought of as a 2-dimensional

1234: complex manifold which blends together the $\ve = D-d$ and the coupling

1235: constant of the theory.

1236: 

1237: 

1238: 

1239: 

1240: 

1241: 

1242: 

1243: 

1244: 

1245: 

1246: 

1247: 

1248: 

1249: 

1250: 

1251: 

1252: \section{Conclusions}\label{sec6}

1253: We showed in this paper that the group $G$

1254: of characters of the Hopf algebra ${\cal H}$ of Feynman graphs

1255: plays a key role in the geometric understanding of the basic ideas of

1256: renormalization

1257: including the renormalization group and the $\beta$-function.

1258: 

1259: We showed in particular that the group $G$

1260: acts naturally on the complex space

1261: $X$ of dimensionless coupling constants of the theory.

1262: Thus, elements of $G$ are a refined form

1263: of diffeomorphisms of $X$ and as such should be called

1264: diffeographisms.

1265: 

1266: The action of these diffeographisms on the space of coupling

1267: constants allowed us first of all to read off directly

1268: the bare coupling constant and

1269: the renormalized one from the Riemann--Hilbert

1270: decomposition of the unrenormalized effective coupling constant

1271: viewed as a loop of formal

1272: diffeomorphisms. This showed that renormalization is intimately

1273: related

1274: with the theory of non-linear complex bundles on the Riemann sphere

1275: of the dimensional regularization parameter $\ve$.

1276: It also allowed us to lift

1277: both the renormalization group and the $\beta$-function

1278: as the asymptotic scaling in the group of diffeographisms.

1279: This used the full power of the Riemann--Hilbert decomposition

1280: together

1281: with the invariance of $\gamma_-(\ve)$ under a change of unit

1282: of mass. This gave us a completely streamlined proof of the existence

1283: of the renormalization group and more importantly a closed

1284: formula of scattering nature,

1285: delivering the full higher pole structure of minimal

1286: subtracted counterterms in terms of the residue.

1287: 

1288: In the light of the predominant role of the residue in NCG

1289: we expect this type of formula to help us to decipher the message

1290: on space-time geometry burried in the need for renormalization.

1291: 

1292: Moreover, thanks to \cite{K3} the previous results no longer

1293: depend upon dimensional regularization but can be formulated in

1294: any regularization or renormalization scheme. Also, we could

1295: discard a detailed discussion of anomalous dimensions, since it is

1296: an easy corollary \cite{BK} of the knowledge of the

1297: $\beta$-function.

1298: 

1299: 

1300: For reasons of simplicity our analysis was limited to the case of

1301: one coupling constant. The generalization to a higher dimensional

1302: space $X$ of coupling constants is expected to involve the same

1303: ingredients as those which appear in higher dimensional

1304: diffeomorphism groups and Gelfand-Fuchs cohomology \cite{CM}. We

1305: left aside the detailed study of the Lie algebra of

1306: diffeographisms and its many similarities with the Lie algebra of

1307: formal vector fields. This, together with the interplay between

1308: Hopf algebras, rational homotopy theory, BRST cohomology, rooted

1309: trees and shuffle identities will be topics of future joint work.

1310: 

1311: 

1312: 

1313: 

1314: 

1315: \section{Appendix: up to three loops}\label{sec7}

1316: We now want to check the Hopf algebra homomorphism ${\cal

1317: H}_{CM}\to{\cal H}$ up to three loops as an example. We regard

1318: $g_0$ as a series in a variable $x$ (which can be thought

1319: of as a physical coupling) up to order $x^6$, making use of

1320: $g_0=xZ_1Z_3^{-3/2}$ and the expression of the $Z$-factors in

1321: terms of 1PI Feynman graph. The challenge is then to confirm that

1322: the coordinates $\delta_n$ on $G_2$, implicitly defined by \cite{CM}

1323: $$\log\left[g_0(x)^\prime\right]^{(n)}$$  commute with the

1324: Hopf algebra homomorphism: calculating the coproduct $\Delta_{CM}$

1325: of $\delta_n$ and expressing the result in Feynman graphs must

1326: equal the application of the coproduct $\Delta$ applied to

1327: $\delta_n$ expressed in Feynman graphs.

1328: 

1329: 

1330: By (2) of section \ref{sec4} we write $g_0=xZ_1Z_3^{-3/2}$,

1331: \[

1332: Z_1=1+\sum_{k=1}^\infty z_{1,2k}x^{2k},

1333: \]

1334: \[

1335: Z_3=1-\sum_{k=1}^\infty z_{3,2k}x^{2k},

1336: \]

1337: and

1338: \[

1339: Z_g=Z_1Z_3^{-3/2},\;z_{i,2k}\in {\cal H}_c,\;i=1,3,

1340: \]

1341: as formal series in $x^2$. Using

1342: \[

1343: \log\left(\frac{\partial}{\partial x}

1344: xZ_g\right)=\sum_{k=1}^\infty\frac{\delta_{2k}}{(2k)!}x^{2k},

1345: \]

1346: which defines $\delta_{2k}$ as the previous generators $a_n(\phi)$

1347: of coordinates of $G_2$, we find

1348: \begin{eqnarray}

1349: \frac{1}{2!}\delta_2 \equiv\tilde{\delta_2} & = &

1350: 3z_{1,2}+\frac{9}{2}z_{3,2},\label{e1}\\ \frac{1}{4!}\delta_4

1351: \equiv\tilde{\delta_4} & = &

1352: 5[z_{1,4}+\frac{3}{2}z_{3,4}]-\frac{9}{2}z_{1,2}^2

1353: -6z_{1,2}z_{3,2}-\frac{3}{4}z_{3,2}^2,\label{e2}\\

1354: \frac{1}{6!}\delta_6 \equiv\tilde{\delta_6} & = &

1355: 9z_{1,2}^3+18z_{1,2}^2z_{3,2}-5[3z_{1,2}z_{1,4}+\frac{3}{2}z_{3,2}z_{3,4}]

1356: \label{e3}\\ & & +12[ z_{1,2}z_{3,2}^2-z_{1,2}

1357: z_{3,4}-z_{1,4}z_{3,2}]+7[z_{1,6}+\frac{1}{2}z_{3,2}^3

1358: +\frac{3}{2}z_{3,6}].\nonumber

1359: \end{eqnarray}

1360: The algebra homomorphism ${\cal H}_c\to{\cal H}$

1361: of section(\ref{sec4})

1362: is effected by expressing the $z_{i,2k}$ in Feynman graphs,

1363: with 1PI graphs

1364: with three external legs contributing to $Z_1$, and 1PI graphs

1365: with two external legs, self-energies, contributing to $Z_3$.

1366: 

1367: Explicitly, we have  {\large

1368: \begin{eqnarray*} z_{1,2} & = & \v,\\ z_{3,2}

1369: &= & \frac{1}{2}\p,\\ z_{1,4} &= &

1370: \vlv+\vuv+\vdv+\frac{1}{2}\left[\vrp+\vup+\vdp\right]+\frac{1}{2}\w,\\

1371: z_{3,4} &= & \frac{1}{2}\left[\pdp+\pv\right].\\

1372: \end{eqnarray*}}

1373: The symmetry factor {\large $$2=S\left(\w\right)$$} is most

1374: obvious if we redraw $$\epsfysize=6mm\epsfbox{figure1.eps}.$$

1375: Further, we have

1376:  {\large\begin{eqnarray*}

1377: z_{1,6} & = &

1378: \vlvlv+\vlvuv+\vlvdv+\vuvlv+\vuvuv+\vuvdv+\vdvlv+\vdvuv+\vdvdv\\

1379:  & &

1380: +\frac{1}{2}\left[\vlvup+\vlvdp+\vlvrp+\vuvup+\vuvdp+\vuvrp\right.\\

1381:  & & \left.+\vdvup+\vdvdp+\vdvrp

1382: \right] +\vluvv+\vldvv+\vudvv\\ & &

1383: +\frac{1}{2}\left[\vlrvp+\vldvp+\vluvp+\vurvp+\vudvp+\vuuvp\right.\\

1384:  & & \left.+\vdrvp+\vddvp+\vduvp\right]\\

1385:  & &

1386: +\frac{1}{4}\left[\vuupp+\vrrpp+\vddpp+\vurpp+\vdrpp+\vudpp\right]\\

1387:  & &

1388: +\frac{1}{2}\left[\wlv+\wurv+\wdrv\right]+\wulv\\

1389:  & &

1390: +\frac{1}{2}\left[\wulp+\wurp+\wdlp+\wdrp+\wmlp+\wmrp\right]\\

1391:  & &

1392: +\frac{1}{2}\left[\vlw+\vuw+\vdw\right]\\

1393:  & &

1394: +\frac{1}{2}\left[\vupv+\vrpv+\vdpv+\vupdp+\vrpdp+\vdpdp\right]+\mbox{\small

1395: primitive terms},

1396: \end{eqnarray*}}

1397: and

1398:  {\large\begin{eqnarray*}

1399: z_{3,6} & = &

1400: \frac{1}{2}\pdpdp+\frac{1}{8}\pdupp+\frac{1}{4}\pddpp

1401: +\frac{1}{2}\pdpv+\frac{1}{2}\pw\nonumber\\

1402:  & &

1403: +\frac{1}{2}\plrvv+\frac{1}{4}

1404: \pvmp+\frac{1}{2}\pvlp+\frac{1}{2}\pvrp+\pvuv.

1405: \end{eqnarray*}}

1406: Here, primitive terms refers to 1PI three-loop vertex graphs

1407: without subdivergences. They fulfil all desired identities below

1408: trivially, and are thus not explicitly given.

1409: On the level of diffeomorphisms, we have the coproducts

1410: \begin{eqnarray}

1411: \Delta_{CM}[\delta_4] & = & \delta_4\ot1+1\ot\delta_4+4\delta_2\ot\delta_2,\\

1412: \Delta_{CM}[\delta_6] & = & \delta_6\ot1+1\ot\delta_6+

1413: 20\delta_2\ot\delta_4+6\delta_4\ot\delta_2\nonumber\\

1414:  & & +28\delta_2^2\ot\delta_2,\label{ed6}

1415: \end{eqnarray}

1416: where we skip odd gradings.

1417: 

1418: We have to check that the coproduct $\Delta$ of Feynman graphs

1419: reproduces these results.

1420: 

1421: Applying $\Delta$ to the rhs of (\ref{e2}) gives, using the

1422: expressions for $z_{i,k}$ in terms of Feynman graphs,

1423:  {\large\begin{eqnarray*}

1424: \Delta(\tilde{\delta_4}) & = &

1425: 6\v\ot\v+\frac{9}{2}\left[\v\ot\p+\p\ot\v\right]\\

1426:  & & +\frac{27}{8}\p\ot\p+\tilde{\delta_4}\ot 1+1\ot

1427: \tilde{ \delta_4}.

1428: \end{eqnarray*}}

1429: This has to be compared with $\tilde{\delta_4}\ot 1+1\ot \tilde{

1430: \delta_4}+\frac{2!2!}{4!}4\tilde{\delta_2}\ot\tilde{\delta_2}$,

1431: which matches perfectly, as

1432:  {\large\begin{eqnarray*}

1433: \tilde{\delta_2}\ot\tilde{\delta_2} & = & 9\v\ot\v+\frac{27}{4}

1434: \left[\v\ot\p\;+\;\p\ot\v\right]\\ & & +\frac{81}{16}\p\ot\p.

1435: \end{eqnarray*}}

1436: 

1437: After this warming up, let us do the check at order $g^6$, which

1438: will be much more demanding, as the coproduct will be

1439: noncocommutative now.

1440: 

1441: We need that $\Delta_{CM}(\delta_6)$ in (\ref{ed6}) is equivalent

1442: to

1443: \begin{eqnarray}

1444: \Delta_{CM}(\tilde{\delta_6}) & = & \tilde{\delta_6}\ot

1445: 1+1\ot\tilde{\delta_6}+20\frac{2!4!}{6!}\tilde{\delta_2}

1446: \ot\tilde{\delta_4}+6\frac{2!4!}{6!}\tilde{\delta_4}\ot\tilde{\delta_2}

1447: \nonumber\\

1448:  & & +28\frac{2!2!2!}{6!}\tilde{\delta_2}^2\ot\tilde{\delta_2}.

1449: \end{eqnarray}

1450: Applying the Hopf algebra homomorphism to Feynman graphs

1451: on both sides of the tensor product delivers

1452:  {\large\begin{eqnarray}

1453: \Delta(\tilde{\delta_6}) & = & \tilde{\delta_6}\ot

1454: 1+1\ot\tilde{\delta_6}\nonumber\\

1455:  & & +\frac{28}{90}\left[3\v+\frac{9}{4}\p\right]^2\ot\left[3\v+

1456: \frac{9}{4}\p\right]\nonumber\\

1457:  & &

1458:  +\frac{6}{15}\left[5\left[\vlv+\vuv+\vdv+

1459:  \frac{1}{2}\left[\vrp+\vup+\vdp+\w\right]\right.\right.\nonumber\\

1460:   & &

1461: \left.\left.

1462: +\frac{3}{2}\left[\frac{1}{2}\pdp+\frac{1}{2}\pv\right]\right]\right.\nonumber\\

1463:  & & \left. -3\left[\frac{3}{2}\v\v+\v\p+\frac{1}{16}\p\p\right]\right]\nonumber\\

1464:  & & \ot \left[3\v+\frac{9}{4}\p\right]\nonumber\\

1465:  & & +\frac{20}{15} \left[3\v+\frac{9}{4}\p\right]\nonumber\\

1466:  & & \ot

1467: 5\left[\vlv+\vuv+\vdv+\frac{1}{2}\left[\vrp+\vup+\vdp+\w\right]

1468: \right.\nonumber\\

1469:  & & \left.

1470: +\frac{3}{2}\left[\frac{1}{2}\pdp+\frac{1}{2}\pv\right]\right]

1471: \nonumber\\

1472:  & &  -3\left[\frac{3}{2}\v\v+\v\p+\frac{1}{16}\p\p\right].

1473: \end{eqnarray}}

1474: Multiplying this out, we find the following result

1475:  {\large\begin{eqnarray}

1476: \Delta(\tilde{\delta_6})  =  \tilde{\delta_6} & \ot & 1\nonumber\\

1477: +1 & \ot & \tilde{\delta_6}\nonumber\\ +10\v & \ot & \w\nonumber\\

1478: +3\w & \ot & \v\nonumber\\ +\frac{9}{4}\w & \ot &

1479: \p\nonumber\\+\frac{15}{2}\p & \ot & \w\nonumber\\

1480:  +\frac{9}{4}\left[\vrp+\vup+\vdp\right] & \ot & \p\nonumber\\

1481: +\frac{15}{2}\p & \ot & \left[\vrp+\vup+\vdp\right]\nonumber\\

1482: +\frac{9}{2}\left[\vlv+\vuv+\vdv\right] & \ot & \p\nonumber\\

1483: +15\p & \ot & \left[\vlv+\vuv+\vdv\right]\nonumber\\

1484: +\frac{9}{2}\pv & \ot & \v\nonumber\\+15\v & \ot & \pv\nonumber\\

1485: +\frac{9}{2}\pdp & \ot & \v\nonumber\\+15\v & \ot & \pdp\nonumber\\

1486: +\frac{27}{8}\pv & \ot & \p\nonumber\\

1487: +\frac{45}{4}\p & \ot & \pv\nonumber\\

1488: +\frac{27}{8}\pdp & \ot & \p\nonumber\\

1489: +\frac{45}{4}\p & \ot & \pdp\nonumber\\

1490: +3\left[\vrp+\vup+\vdp\right] & \ot & \v\nonumber\\

1491: +10\v & \ot & \left[\vrp+\vup+\vdp\right]\nonumber\\

1492: +6\left[\vlv+\vuv+\vdv\right] & \ot & \v\nonumber\\

1493: +20\v & \ot & \left[\vlv+\vuv+\vdv\right]\nonumber\\

1494: -9\p & \ot & \v\p\nonumber\\-\frac{9}{2}\p\p & \ot & \v\nonumber\\

1495: -\frac{3}{4}\v & \ot & \p\p\nonumber\\

1496: -\frac{9}{16}\p & \ot & \p\p\nonumber\\

1497: -\frac{27}{2}\p & \ot & \v\v\nonumber\\-12\v & \ot & \v\p\nonumber\\

1498: -18\v & \ot & \v\v\nonumber\\+\frac{27}{4}\p\v & \ot & \p\nonumber\\

1499: +\p\p & \ot & \p\nonumber\\+9\p\v & \ot & \v\nonumber\\

1500: +\frac{9}{4}\v\v & \ot & \p\nonumber\\+3\v\v & \ot & \v\label{e4}.

1501: \end{eqnarray}}

1502: Now we have to compare with $\Delta(\tilde{\delta_6})$,

1503: so we first apply the homomorphism to graphs and use the coproduct

1504: $\Delta$ on them. For this,

1505: we need

1506:  {\large\begin{eqnarray}

1507: \Delta[z_{1,2}] & = & \v\ot 1+1\ot \v,\\ \Delta[z_{3,2}] & = &

1508: \p\ot 1+1\ot \p,\\ \Delta[z_{1,4}] & = & z_{1,4}\ot 1+1\ot

1509: z_{1,4}+3\v\ot\v+\frac{3}{2}\p\ot\v,\\ \Delta[z_{3,2}] & = &

1510: z_{3,4}\ot 1+1\ot z_{3,4}+\frac{1}{2}\p\ot\p+\v\ot\p,\\

1511: \Delta[z_{1,6}] & = & z_{1,6}\ot 1+1\ot

1512: z_{1,6}+3\v\ot\left[\vlv+\vuv+\vdv\right] \\

1513:  & &

1514: +3\left[\vlv+\vuv+\vdv\right] \ot\v\nonumber\\

1515:  & & +2\v\ot\left[\vlv+\vuv+\vdv\right]\nonumber\\

1516:  & &

1517: +3\v\v\ot\v+\p\ot\left[\vrp+\vup+\vdp\right]\nonumber\\

1518:  & & +\frac{3}{2}\p\p\ot\v\nonumber\\

1519:  & & +\frac{3}{2}\v\ot\left[\vrp+\vup+\vdp\right]+\frac{3}{2}\pv\ot\v\nonumber\\

1520:  & & +\frac{1}{2}\p\ot\left[\vrp+\vup+\vdp\right]+\frac{3}{2}\pdp\ot\v\nonumber\\

1521: & &

1522: +\frac{3}{2}\p\ot\left[\vlv+\vuv+\vdv\right]\nonumber\\

1523:  & & +\v\ot\left[\vrp+\vup+\vdp\right]

1524: \nonumber\\

1525:  & & +\frac{9}{2}\v\p\ot\v+\frac{3}{2}\p\ot\left[\vlv+\vuv+\vdv\right]\nonumber\\

1526:  & & +\frac{3}{2}\left[\vrp+\vup+\vdp\right]\ot\v+\frac{3}{2}\w\ot\v\nonumber\\

1527:  & & +\frac{5}{2}\v\ot\w+3\p\ot\w,\nonumber\\

1528: \Delta[z_{3,6}] & = & z_{3,6}\ot 1+1\ot

1529: z_{3,6}+\frac{1}{2}\p\ot\pdp\\

1530:  & & +\frac{1}{2}\pdp\ot\p+\frac{1}{4}\p\ot\pdp\nonumber\\

1531:  & & +\frac{1}{8}\p\p\ot\p

1532: \nonumber\\

1533:  & & +\frac{1}{2}\p\ot\pdp+\frac{1}{4}\p\p\ot\p+\v\ot\pdp\nonumber\\

1534:  & & +\frac{1}{2}\pv\ot\p+\w\ot\p+\frac{1}{2}\v\v\ot\p\nonumber\\

1535:  & & +\v\ot\pv+\frac{1}{4}\p\ot\pv\nonumber\\

1536:   & & +\frac{1}{2}\vrp\ot\p+\p\ot\pv

1537: \nonumber\\

1538:  & & +\frac{1}{2}\left[\vup+\vdp\right]\ot\p+\v\ot\pdp+\v\ot\pv\nonumber\\

1539:   & & +\left[\vlv+\vuv+\vdv\right]\ot\p.\nonumber

1540: \end{eqnarray}}

1541: It is now only a matter of using the rhs of

1542: (\ref{e3}) for $\tilde{\delta_6}$

1543: to confirm that we reproduce the result (\ref{e4}). For example,

1544: for the contribution to $\v\ot\w$ in $\Delta(\tilde{\delta_6})$ we

1545: find

1546:  {\large\[

1547: -\frac{5\times 3}{2}\v\ot\w+7\times \frac{5}{2}\v\ot\w=10\v\ot\w,

1548: \]}

1549: as desired. Similarly, one checks all of the 32 tensorproducts of

1550: (\ref{e4}).

1551: 

1552: \section*{Acknowledgements}

1553: Both authors thank the IHES for generous support during this

1554: collaboration. D.K.~is grateful to the DFG for a Heisenberg

1555: Fellowship.

1556: \begin{thebibliography}{999}

1557: \bibitem{I}

1558: A.~Connes, D.~Kreimer, Comm.~Math.~Phys.~{\bf 210} 249 (2000),

1559: hep-th/9912092.

1560: \bibitem{CK}

1561: A.~Connes, D.~Kreimer, Comm.~Math.~Phys.~{\bf 199} 203 (1998),

1562: hep-th/9808042.

1563: \bibitem{Ar} H.~Araki,

1564: %Expansionals in Banach Algebras

1565: Ann.~Sci.~\'Ecole Norm.~Sup.~(4) {\bf 6} 67 (1973).

1566: %\bibitem{4}

1567: %\bibitem{B}

1568: %A.~Beauville, {\sl Monodromie des syst\`emes diff\'erentiels

1569: %lin\'eaires \`a p\^oles simples sur la sph\`ere de Riemann}, {\it

1570: %S\'eminaire Bourbaki} 45\`eme ann\'ee, 1992-1993, n.{\bf 765}.

1571: %\medskip

1572: %\bibitem{Bo}

1573: %A.~Bolibruch, {\sl Fuchsian systems with reducible monodromy and

1574: %the Riemann-Hilbert problem}, {\it Lecture Notes in Math.}{\bf

1575: %1520}, 139-155(1992).

1576: %\medskip

1577: % \bibitem{BP}

1578: %N.~N.~Bogoliubov, O. Parasiuk {\it Acta Math. } {\bf 97}, 227

1579: %(1957);\\ K.~Hepp, {\it Commun.Math.Phys.} {\bf 2}, 301 (1966).

1580: %\medskip

1581: %\bibitem{K4} %Applications

1582: %D.J.~Broadhurst, D.~Kreimer, {\it J. Symb. Comput.}{\bf 27}, 581

1583: %(1999); hep-th/9810087;\\

1584: %\medskip

1585: %D.~Kreimer, R.~Delbourgo,  {\it Phys. Rev.} {\bf D60}, 105025

1586: %(1999); hep-th/9903249.

1587: %\medskip

1588: \bibitem{Col} J.~Collins, {\it Renormalization},  Cambridge University Press (1984).

1589: %\medskip

1590: %\bibitem{CK0}

1591: %A.~Connes, D.~Kreimer, {\it J.High Energy Phys.}{\bf 09}, 024

1592: %(1999); hep-th/9909126.

1593: %\medskip

1594: \bibitem {CM}

1595: A.~Connes, H.~Moscovici, Commun.~Math.~Phys.~{\bf 198}

1596:  199 (1998), math.dg/9806109.

1597: %\medskip

1598: %\bibitem{MD}

1599: %M.~Dresden, {\sl Renormalization in historical perspective - The

1600: %first stage}, in {\it Renormalization}, ed.~L.~Brown,

1601: %Springer-Verlag, New York, Berlin, Heidelberg (1994).

1602: %\medskip

1603: %\bibitem{EG}

1604: %H.~Epstein, V.~Glaser, {\sl The role of locality in perturbation

1605: %theory}, {\it Ann. Inst. H. Poincar\'e} A {\bf 19} (1973) 211-295.

1606: %\medskip

1607: %\bibitem{FulMacP} Fulton, MacPherson.

1608: %\medskip

1609: %\bibitem{GK}

1610: %K.~Gawedzki, A.~Kupiainen,  {\sl Exact renormalization of the Gross-Neveu

1611: %model of quantum fields},  {\it Phys. Rev. Lett.}{\bf 54}(1985); \\

1612: %J.~Feldman, J.~Magnen, V.~Rivasseau, R.~Seneor,  {\sl Massive Gross-Neveu

1613: %model: a rigorous perturbative construction},

1614: %{\it Phys. Rev. Lett.}{\bf 54}(1985).

1615: %\medskip

1616: %\bibitem{GJ}

1617: %J.~Glimm, A.~Jaffe, {\sl Quantum Physics}, Springer Verlag, New

1618: %York, Berlin, Heidelberg (1987).

1619: %\medskip

1620: %\bibitem{H}

1621: %G.~'t Hooft, {\it Nuclear Physics} {\bf B 61} 455 (1973).

1622: %\medskip

1623: %\bibitem{KW} T.~Krajewski, R.~Wulkenhaar, {\it Eur.Phys.J.} {\bf C7}, 697-708

1624: %(1999); hep-th/9805098.

1625: %\medskip

1626: %\bibitem{K1} %Basics

1627: %D.~Kreimer,  {\it Adv. Theor. Math. Phys.} {\bf 2.2}, 303 (1998);

1628: %q-alg/9707029;\\ A.~Connes, D.~Kreimer,  {\it Lett. Math. Phys.}

1629: %{\bf 48}, 85 (1999); hep-th/9904044.

1630: %\medskip

1631: %\bibitem{Max} M.~Kontsevich, Lett.Math.Phys.

1632: %\medskip

1633: %\bibitem{K2} %Overlapping Divs

1634: %D.~Kreimer,  {\it Commun. Math. Phys.}{\bf 204}, 669 (1999); hep-th/9810022.

1635: %\medskip

1636: \bibitem{K3} %Chen paper

1637: D.~Kreimer,  Adv.~Theor.~Math.~Phys.~{\bf 3.3} (1999),

1638: hep-th/9901099.

1639: %\medskip

1640: \bibitem{BK}

1641: D.J.~Broadhurst, D.~Kreimer, Phys.~Lett.~{\bf B475} 63 (2000),

1642: hep-th/9912093.

1643: %\bibitem{Ki}

1644: %\bibitem{KW}

1645: %\bibitem{LD}

1646: %I.~Lappo-Danilevskii, {\sl M\'emoires sur la th\'eorie des

1647: %syst\`emes des \'equations diff\'erentielles lin\'eaires,}

1648: %Chelsea, New York (1953).

1649: \end{thebibliography}

1650: 

1651: \end{document}

1652: