1: %%%%%%%%%% espcrc2.tex %%%%%%%%%%

2: %

3: % $Id: espcrc2.tex,v 1.2 2004/02/24 11:22:11 spepping Exp $

4: %

5: \documentclass[fleqn,twoside]{article}

6: % \usepackage{espcrc2}

7: % Use the option 'headings' if you want running headings

8: \usepackage[headings]{espcrc2}

9: 

10: % identification

11: \readRCS

12: $Id: espcrc2.tex,v 1.2 2004/02/24 11:22:11 spepping Exp $

13: \ProvidesFile{espcrc2.tex}[\filedate \space v\fileversion

14:      \space Elsevier 2-column CRC Author Instructions]

15: 

16: % change this to the following line for use with LaTeX2.09

17: % \documentstyle[twoside,fleqn,espcrc2]{article}

18: 

19: % if you want to include PostScript figures

20: \usepackage{graphicx}

21: % if you have landscape tables

22: \usepackage[figuresright]{rotating}

23: 

24: % put your own definitions here:

25: %   \newcommand{\cZ}{\cal{Z}}

26: %   \newtheorem{def}{Definition}[section]

27: %   ...

28: \newcommand{\ttbs}{\char'134}

29: \newcommand{\AmS}{{\protect\the\textfont2

30:   A\kern-.1667em\lower.5ex\hbox{M}\kern-.125emS}}

31: 

32: % add words to TeX's hyphenation exception list

33: \hyphenation{author another created financial paper re-commend-ed Post-Script}

34: 

35: % set the starting page if not 1

36: % \setcounter{page}{17}

37: 

38: 

39: % ---------- our definitions ----------

40: 

41: \usepackage{amsmath}

42: \usepackage{amssymb}

43: \usepackage{amsthm}

44: \usepackage{bbm}

45: 

46: \usepackage{maple2e}

47:  \def\emptyline{\vspace{12pt}}

48: \DefineParaStyle{Maple Output}

49: \DefineCharStyle{2D Math}

50: \DefineCharStyle{2D Output}

51: \DefineParaStyle{Bullet Item}

52: \DefineParaStyle{Heading 1}

53: \DefineParaStyle{Heading 2}

54: \DefineParaStyle{Warning}

55: \DefineCharStyle{2D Comment}

56: \DefineCharStyle{Help Normal}

57: \DefineCharStyle{Hyperlink}

58: 

59: \newcommand{\Gr}{Gr{\"o}bner }

60: \newcommand{\GR}{GR{\"O}BNER }

61: \newcommand{\cJ}{\cal{J}}

62: \newcommand{\cM}{\cal{M}}

63: \newcommand{\cI}{\cal{I}}

64: \newcommand{\cNM}{\cal{NM}}

65: \newcommand{\cL}{\cal{L}}

66: 

67: \newcommand{\D}{\mathbb{D}}

68: \newcommand{\Q}{\mathbb{Q}}

69: \newcommand{\Z}{\mathbb{Z}}

70: \newcommand{\K}{\mathbb{K}}

71: \newcommand{\N}{\mathbb{N}}

72: \newcommand{\M}{\mathbb{M}}

73: \newcommand{\R}{\mathbb{R}}

74: \newcommand{\X}{\mathbb{X}}

75: 

76: 

77: % declarations for front matter

78: \title{A Maple Package for Computing \Gr Bases for Linear Recurrence Relations}

79: 

80: \author{Vladimir P. Gerdt\address[JINR]{Laboratory of Information Technologies,

81:        Joint Institute for Nuclear Research,

82:        141980 Dubna, Russia}\thanks{gerdt@jinr.ru}

83:        and

84:        Daniel Robertz\address[LBFM]{Lehrstuhl B f{\"u}r Mathematik, RWTH Aachen,

85:        Templergraben 64, D-52062 Aachen,

86:        Germany}\thanks{daniel@momo.math.rwth-aachen.de}}

87: 

88: % If you use the option headings,

89: % the title is also used as the running title,

90: % and the authors are also used as the running authors.

91: % You can change that by using \runtitle and \runauthor.

92: 

93: \runtitle{A Maple Package for Computing \Gr bases for Linear Recurrence Relations}

94: \runauthor{V. P. Gerdt and D. Robertz}

95: 

96: \begin{document}

97: 

98: \begin{abstract}

99: A Maple package for computing \Gr bases of linear difference ideals is described.

100: The underlying algorithm is based on Janet and Janet-like monomial divisions associated with

101: finite difference operators. The package can be used, for example, for automatic

102: generation of difference schemes for linear partial differential equations and for reduction of multiloop

103: Feynman integrals. These two possible applications are illustrated by simple examples

104: of the Laplace equation and a one-loop scalar integral of propagator type.

105: \end{abstract}

106: 

107: % typeset front matter (including abstract)

108: \maketitle

109: 

110: %

111: % ------ INTRODUCTION ------

112: %

113: \section{INTRODUCTION}

114: 

115: As it is argued in~\cite{G04}, \Gr bases form the most universal

116: algorithmic tool for reduction of loop integrals determined by

117: linear recurrence relations derived from the integration by parts

118: method. These recurrence relations can be considered as generators

119: of an ideal in the ring of finite difference polynomials. Hence, to

120: determine the minimal set of basis (master) integrals and to

121: express other integrals in terms of basis integrals it suffices to

122: compute any \Gr basis, i.e., for any ranking (ordering) on

123: difference operators.

124: 

125: Another important application of difference \Gr bases is generation of

126: difference schemes for partial differential equations~\cite{MB01,GBM05}. A difference

127: scheme is constructed by elimination of partial derivatives from the system of difference

128: equations obtained by discretization of the differential equations and certain integral

129: relations. Therefore, the corresponding elimination ranking is to be used.

130: 

131: In this paper we present a Maple package for computing \Gr bases

132: of linear difference ideals. The package is a modified version

133: of our earlier package~\cite{Daniel} oriented towards commutative and

134: linear differential algebra and based on the involutive basis

135: algorithm~\cite{InvAlg}. The modified version is specialized to

136: linear difference ideals and uses both Janet and Janet-like

137: divisions~\cite{GB05} adopted to linear difference

138: polynomials~\cite{G-ACAT}.

139: 

140: 

141: %

142: % ------ THE MAPLE PACKAGE LDA ------

143: %

144: \section{THE MAPLE PACKAGE LDA}

145: 

146: \subsection{Outline of the package}

147: 

148: The package LDA (abbreviation for {\bf{L}}inear {\bf{D}}ifference

149: {\bf{A}}lgebra) implements the involutive basis algorithm

150: \cite{InvAlg}\footnote{However, Janet trees for the fast search of

151: involutive divisors are not yet implemented since Maple does not

152: provide efficient data structures for trees.} for linear systems

153: of difference equations using Janet division. In addition, the package

154: implements a modification of the algorithm

155: oriented towards Janet-like division~\cite{GB05} and, thus, computes Janet-like

156: \Gr bases of linear difference ideals~\cite{G-ACAT}.

157: 

158: Table~\ref{table:LDA}

159: collects the most important

160: commands of LDA. Its main procedure {\tt JanetBasis} converts a

161: given set of difference polynomials into its Janet basis or

162: Janet-like \Gr basis form. More precisely, let $\R$ be the

163: difference ring \cite{G-ACAT} of polynomials in the variables

164: $\theta^\mu \circ y^k$, $\mu \in \Z_{\ge 0}^n$, $k = 1$, \ldots, $m$, with

165: coefficients in a difference field $\K$

166: containing $\mathbbm{Q}$ for which the field operations can be

167: carried out constructively in Maple. We denote by $\R_L$ the set of linear

168: polynomials in $\R$.

169: Given a generating set $F \subset \R_L$ for a

170: linear difference ideal $I$ in $\R$, {\tt JanetBasis}

171: computes the minimal Janet(-like Gr{\"o}bner) basis $J$ of $I$

172: w.r.t.\ a certain monomial order (ranking).

173: The input for {\tt JanetBasis} consists of the left hand

174: sides of a linear system of difference equations in the

175: dependent variables $y^1$, \ldots, $y^m$, e.g. functions of

176: $x_1$, \ldots, $x_n$. The difference

177: ring $\R$ is specified by the lists of independent variables

178: $x_1$, \ldots, $x_n$ and dependent

179: variables given to {\tt JanetBasis}. The output %of {\tt JanetBasis}

180: is a list containing the Janet(-like

181: Gr{\"o}bner) basis $J$ and the lists of independent and dependent

182: variables.

183: 

184: After $J$ is computed, the involutive/${\cJ}-$normal form of any element of

185: $\R_L$ modulo $J$ can be computed using {\tt InvReduce}.

186: Given $p \in \R_L$ representing a residue class $\overline{p}$ of

187: the difference residue class ring $\R / I$, {\tt InvReduce}

188: returns the unique representative $q \in \R_L$ of $\overline{p}$ which

189: is not involutively/${\cJ}-$reducible modulo $J$.

190: A $\K$-basis of the vector space $\R_L / (I \cap \R_L)$ is returned by

191: {\tt ResidueClassBasis} as a list if it is finite or is enumerated by

192: a formal power series \cite{JanetsApproach}

193: in case it is infinite. For examples of how to apply

194: these two commands, cf.\ Section~\ref{section:Feynman}.

195: 

196: Given an affine (i.e. inhomogeneous) linear system of difference

197: equations, a call of {\tt CompCond} after the application of

198: {\tt JanetBasis} returns a generating set of

199: compatibility conditions for the affine part of the system,

200: i.e. necessary conditions for the right hand sides of the

201: inhomogeneous system for solvability.

202: 

203: Moreover, combinatorial devices to compute the

204: Hilbert series and polynomial and function etc. \cite{Daniel}

205: are included in LDA.

206: 

207: For the application of LDA to the reduction of Feynman

208: integrals, a couple of special commands were implemented

209: to impose further relations on the master integrals:

210: By means of {\tt AddRelation} an infinite sequence

211: of master integrals parametrized by indeterminates which

212: are not contained in the list of independent variables

213: is set to zero. Subsequent calls of {\tt InvReduce}

214: and {\tt ResidueClassBasis} take these additional

215: relations into account (cf.\ Section~\ref{section:Feynman}).

216: 

217: Finally, LDA provides several tools for dealing with

218: difference operators. Difference operators represented by

219: polynomials can be applied to (lists of) expressions containing

220: $y^1$, \ldots, $y^m$

221: %the dependent variables

222: as functions of the independent variables.

223: Conversely, the difference operators can be extracted

224: from systems of difference equations.

225: Leading terms of difference equations

226: %w.r.t.\ chosen rankings

227: can be selected.

228: 

229: 

230: \begin{table*}[htb]

231: \caption{Main commands of LDA}

232: \label{table:LDA}

233: \renewcommand{\tabcolsep}{2pc} % enlarge column spacing

234: \renewcommand{\arraystretch}{1.2} % enlarge line spacing

235: \begin{tabular}{@{}ll}

236: \hline

237: {\tt JanetBasis}                  & Compute Janet(-like Gr{\"o}bner) basis\\

238: {\tt InvReduce}                   & Involutive / ${\cJ}-$reduction modulo Janet(-like Gr{\"o}bner) basis\\

239: {\tt CompCond}                    & Return compatibility conditions for inhomogeneous system\\

240: {\tt HilbertSeries} etc.          & Combinatorial devices\\

241: {\tt Pol2Shift}, {\tt Shift2Pol}  & Conversion between shift operators and equations\\

242: \hline

243: \multicolumn{2}{@{}l}{Some interpretations of commands for the reduction of Feynman integrals:} \\

244: \hline

245: {\tt ResidueClassBasis}           & Enumeration of the master integrals\\

246: {\tt AddRelation}                 & Definition of additional relations for master integrals\\

247: {\tt ResidueClassRelations}       & Return the relations defined for master integrals\\

248: \hline

249: \end{tabular}

250: \end{table*}

251: 

252: 

253: 

254: \subsection{Implementation details}

255: 

256: We consider difference rings containing

257: shift operators which act in one direction only.

258: If a linear system of difference equations is given

259: containing functions shifted in both directions,

260: then the system needs to be shifted by the

261: maximal negative shift in order to obtain a

262: difference system with shifts in one direction only.

263: However, LDA allows to change the shift direction

264: globally.

265: 

266: Unnecessary computations of involutive reductions

267: to zero are avoided using the four involutive criteria

268: described in \cite{Criteria}, \cite{UnnecessaryReduction}.

269: Fine-tuning is possible by selecting the criteria

270: individually.

271: 

272: The implemented monomial orders/rankings are the

273: (block) degree-reverse-lexicographical

274: and the lexicographical one. In the case of more

275: than one dependent variable, priority of comparison

276: can be either given to the difference operators

277: (``term over position'') or to the dependent variables

278: (``position over term''/elimination ranking).

279: 

280: The

281: ranking is controlled via options given to

282: each command separately. The other options described above

283: can be set for the entire LDA session using

284: the command {\tt LDAOptions} which also allows

285: to select Janet or Janet-like division.

286: 

287: 

288: %

289: % ------ GENERATION OF FINITE DIFFERENCE SCHEMES FOR PDES ------

290: %

291: \section{GENERATION OF FINITE DIFFERENCE SCHEMES FOR PDES}

292: 

293: \input Laplace.tex

294: 

295: 

296: %

297: % ------ REDUCTION OF FEYNMAN INTEGRALS ------

298: %

299: \section{REDUCTION OF FEYNMAN INTEGRALS} \label{section:Feynman}

300: 

301: \input Feynman.tex

302: 

303: 

304: %

305: % ------ CONCLUSION ------

306: %

307: \section{CONCLUSION}

308: 

309: We presented the Maple package LDA implementing the Janet/Janet-like

310: division algorithm for computing \Gr bases of linear difference ideals. Generation

311: of difference schemes for linear PDEs and reduction of loop Feynman integrals are

312: important applications of the package.

313: 

314: These two kinds of applications were illustrated by rather simple

315: examples. The first difference system (discrete Laplace equation

316: and integral relations) contains two independent variables $(x,y)$

317: and three dependent variables $(u,u_x,u_y)$. The second system

318: (recurrence relations for one-loop Feynman integral) also contains

319: two independent variables/indices $(k,n)$, but the only dependent

320: variable $f$. The second system, however, is computationally

321: slightly harder than the first one because of explicit dependence

322: of the recurrence relations on the indices and three parameters

323: $(d, m^2, q^2)$ involved in the dependence on indices.

324: 

325: Dependence on index variables and parameters is an attribute of recurrence

326: relations for Feynman integrals. Similar dependence may occur in the generation of

327: difference

328: schemes for PDEs with variable coefficients containing parameters. Theoretically

329: established exponential and superexponential (depends on the ideal and ordering)

330: complexity of constructing polynomial \Gr bases implies

331: that construction of difference \Gr bases is at least exponentially hard in the number of

332: independent variables (indices). Besides, in the presence of parameters

333: the volume of computation grows very rapidly as the number of parameters increases.

334: 

335: Thus, for successful application of the \Gr basis technique to multiloop

336: Feynman integrals with masses and to multidimensional PDEs with multiparametric

337: variable coefficients we are going not only to improve our Maple code

338: but also to implement the algorithms for computing Janet and/or Janet-like difference bases in

339: C++ as a special module of the open source software available on the Web page {\tt http://invo.jinr.ru}.

340: 

341: 

342: 

343: 

344: 

345: 

346: 

347: 

348: 

349: 

350: 

351: \section{ACKNOWLEDGEMENTS}

352: 

353: The authors are grateful to Vladimir Smirnov

354: for useful

355: discussion, comments and suggestions on the example of

356: Section~\ref{section:Feynman}. The contribution of the first

357: author (V.P.G.) was partially supported by the grants 04-01-00784

358: and 05-02-17645 from the Russian Foundation for Basic Research and

359: the grant 2339.2003.2 from the Russian Ministry of Science and

360: Education.

361: 

362: 

363: \begin{thebibliography}{99}

364: 

365: \bibitem{G04} V.P.Gerdt.

366: {\em \Gr Bases in Perturbative Calculations}. Nuclear Physics B

367: (Proc. Suppl.) 135, 2004, pp. 232--237. arXiv:/hep-ph/0501053.

368: 

369: \bibitem{MB01} V.V.Mozzhilkin, Yu.A.Blinkov. {\em Methods of Constructing Difference

370: Schemes in Gas Dynamics}. Transactions of Saratov University, 1(2), 2001, pp. 145--156

371: (in Russian).

372: 

373: \bibitem{GBM05} V.P.Gerdt, Yu.A.Blinkov, V.V.Mozzhilkin. {\em Linear Difference Ideals and

374: Generation of Difference Schemes for PDEs}. Symmetry,

375: Integrability and Geometry: Methods and Applications (SIGMA),

376: Institute of Mathematics, Kiev, 2005, to appear.

377: 

378: \bibitem{Daniel} Yu. A. Blinkov, V. P. Gerdt, C. F. Cid, W.Plesken, D.Robertz.

379: {\em The Maple Package "Janet": I. Polynomial Systems}. Computer Algebra in

380: Scientific Computing / CASC 2003, V. G. Ganzha, E. W. Mayr, and E.

381: V. Vorozhtsov (Eds.), Institute of Informatics, Technical

382: University of Munich, Garching, 2003, pp. 31--40; {\em II. Linear

383: Partial Differential Equations}, ibid., pp. 41--54.

384: 

385: \bibitem{InvAlg} V.P.Gerdt.

386: {\em Involutive Algorithms for Computing \Gr Bases}.

387: Computational commutative and non-commutative algebraic

388: geometry, IOS Press, Amsterdam, 2005, pp. 199--225.

389: arXiv:math.AC/0501111, 2005.

390: 

391: \bibitem{GB05} V.P.Gerdt, Yu.A.Blinkov.

392: {\em Janet-like Monomial Division.}

393: Computer Algebra in Scientific Computing / CASC 2005, V. G. Ganzha, E. W.

394: Mayr, E. V. Vorozhtsov (Eds.), Springer-Verlag, Berlin, 2005, pp. 174--183;

395: {\em Janet-like \Gr Bases}, ibid., pp. 184-195.

396: 

397: \bibitem{G-ACAT} V.P.Gerdt.

398: {\em On Computation of \Gr Bases for Linear Difference Systems}.

399: This volume.

400: 

401: \bibitem{Criteria} V.P.Gerdt, D.A.Yanovich.

402: {\em Experimental Analysis of Involutive Criteria}. Algorithmic

403: Algebra and Logic 2005, A.Dolzmann, A.Seidl, T.Sturm (Eds.), BOD

404: Norderstedt, Germany, 2005, pp. 105--109.

405: 

406: \bibitem{UnnecessaryReduction} J.Apel, R.Hemmecke.

407: {\em Detecting unnecessary reductions in an involutive basis computation},

408: RISC Linz Report Series 02-22, 2002.

409: 

410: \bibitem{JanetsApproach} W.Plesken, D.Robertz.

411: {\em Janet's approach to presentations and resolutions for

412: polynomials and linear pdes}.

413: Archiv der Mathematik, 84 (1), 2005, pp. 22--37.

414: 

415: \bibitem{Tarasov} O.V.Tarasov.

416: {\em Reduction of Feynman graph amplitudes to a minimal set of

417: basic integrals}. Acta Physica Polonica B29, 1998, pp. 2655--2666. arXiv:/hep-ph/9812250

418: 

419: \bibitem{Smirnov} V.A.Smirnov.

420: {\em Evaluating Feynman Integrals},

421: STMP 211, Springer, Berlin, 2004.

422: \end{thebibliography}

423: 

424: \end{document}

425: