%%%%%%%%%%%%%%%%%%%%%%%%% %% FICHIER PLAIN TEX %% %%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%% %% COMMENTAIRES %% %%%%%%%%%%%%%%%%%%%% % % Pour les th\'eor\`emes, propositions, lemmes, % corollaires, assertions, conjectures, probl\`emes, % ... et d\'efinitions, en Fran\c{c}ais ou en Anglais, % les taper en "petites capitales" (small capitals) % et leurs textes en italique, sauf pour les % d\'efinitions o\`u il est en romain, avec, % \'eventuellement, l'objet ou le concept \`a % d\'efinir en italique. Les faire pr\'ec\'eder % par un "\medskip" et suivre par un "\smallskip". % Les remarques, exemples, cas etc... sont \`a % taper en italique et leurs textes en romain. % % Exemple : % % \medskip % % {\sc Th\'eor\`eme~2.1.~--~}{\it \'Enonc\'e du % th\'eor\`eme.} % % \smallskip % % {\it Remarque\/}~2.2.~--~}Texte de la remarque % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \magnification=1100 \overfullrule=0pt %%%%%%%%%%%%%%% %\catcode`@=11 %---> FOURTEENPOINT %\newfam\gothfam %\newfam\gothbfam %\newfam\bbfam %\newskip\ttglue %\def\twelvepoint{\def\rm{\fam0\twelverm}% %\textfont0=\fourteenrm \scriptfont0=\twelverm \scriptscriptfont0=\ninerm %\textfont1=\fourteeni \scriptfont1=\twelvei \scriptscriptfont1=\ninei %\textfont2=\fourteensy \scriptfont2=\twelvesy \scriptscriptfont2=\ninesy %\textfont3=\tenex \scriptfont3=\tenex \scriptscriptfont3=\tenex %\textfont\itfam=\fourteenit \def\it{\fam\itfam\fourteenit}% %\textfont\slfam=\fourteensl \def\sl{\fam\slfam\fourteensl}% %\textfont\ttfam=\fourteentt \def\tt{\fam\ttfam\fourteentt}% %\textfont\bffam=\fourteenbf \scriptfont\bffam=\twelvebf \scriptscriptfont\bffam=\ninebf % \def\bf{\fam\bffam\fourteenbf}% %\textfont\gothfam=\fourteeneufm \scriptfont\gothfam=\twelveeufm \scriptscriptfont\gothfam=\nineeufm % \def\goth{\fam\gothfam\fourteeneufm}% %\textfont\gothbfam=\fourteeneufb \scriptfont\gothbfam=\twelveeufb \scriptscriptfont\gothbfam=\nineeufb % \def\gthb{\fam\gothbfam\fourteeneufb}% %\textfont\bbfam=\fourteenbb \scriptfont\bbfam=\twelvebb \scriptscriptfont\bbfam=\ninebb % \def\bb{\fam\bbfam\fourteenbb}% %\tt\ttglue=0.8em plus .5em minus .3em %\normalbaselineskip=18pt %\setbox\strutbox=\hbox{\vrule height15pt depth6pt width0pt}% %\font\sc=cmcsc10 scaled\magstep2 %\normalbaselines\rm}%\fourteenpoint %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % MACRO % \font\eighttt=cmtt8 % \font\eightsl=cmsl8 % \font\eightex=cmex10 at 8pt %%%%%%%%%%%%%%%% %% FONT CMR %% %%%%%%%%%%%%%%%% % \font\fiverm=cmr5 % \font\sixrm=cmr6 \font\seventeenrm=cmr17 \font\fourteenrm=cmr12 at 14pt \font\twelverm=cmr12 \font\ninerm=cmr10 at 9pt \font\eightrm=cmr8 \font\sixrm=cmr10 at 6pt % \textfont0=\tenrm \scriptfont0=\sevenrm \scriptscriptfont0=\fiverm \def\rm{\fam0\tenrm} %%%%%%%%%%%%%%%% %% FONT CMM %% %%%%%%%%%%%%%%%% % % \font\sixi=cmmi6 % \font\eighti=cmmi8 % \font\ninei=cmmi9 % \font\twelvei=cmmi12 % \font\fourteeni=cmmi10 at 14pt % \font\teni=cmmi10 \font\seveni=cmmi7 \font\fivei=cmmi5 \textfont1=\teni \scriptfont1=\seveni \scriptscriptfont1=\fivei \def\mit{\fam1} \def\oldstyle{\fam1\teni} %%%%%%%%%%%%%%%%% %% FONT CMSY %% %%%%%%%%%%%%%%%%% % \font\fivesy=cmsy6 at 5pt \font\sixsy=cmsy6 \font\sevensy=cmsy7 \font\eightsy=cmsy8 \font\ninesy=cmsy9 \font\tensy=cmsy10 \font\twelvesy=cmsy10 at 12pt \font\fourteensy=cmsy10 at 14pt %%%%%%%%%%%%%%%%% %% FONT CMTI %% %%%%%%%%%%%%%%%%% % \font\nineit=cmti10 at 9pt \font\eightit=cmti8 \font\sevenit=cmti7 %%%%%%%%%%%%%%%%%%% %% FONT CMBFTI %% %%%%%%%%%%%%%%%%%%% % \font\eightbfti=cmbxti10 at 8pt \font\bfti=cmbxti10 \font\twelvebfti=cmbxti10 at 12pt \font\fourteenbfti=cmbxti10 at 14pt %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% FONT CMCSC (Petites capitales) %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \font\sc=cmcsc10 %%%%%%%%%%%%%%%%% %% FONT MSBM %% %%%%%%%%%%%%%%%%% \font\sixbboard=msbm7 at 6pt % \font\ninebb=msbm9 % msbm10 at 9pt % \font\eightbb=msbm8 % msbm10 at 8pt % \font\sixbb=msbm6 % msbm10 at 6pt % \font\sixbboard=msbm7 at 6pt % \font\tenbb=msbm10 \font\sevenbb=msbm7 \font\fivebb=msbm5 \newfam\bbfam \textfont\bbfam=\tenbb \scriptfont\bbfam=\sevenbb \scriptscriptfont\bbfam=\fivebb \def\bb{\fam\bbfam\tenbb} \let\oldbb=\bb \def\bb #1{{\oldbb #1}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% FONT EUFM (Gothique) %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \font\sixgoth=eufm5 at 6pt \font\eightgoth=eufm7 at 8pt % \font\tengoth=eufm10 \font\sevengoth=eufm7 \font\fivegoth=eufm5 \newfam\gothfam \textfont\gothfam=\tengoth \scriptfont\gothfam=\sevengoth \scriptscriptfont\gothfam=\fivegoth \def\goth{\fam\gothfam\tengoth} % %%%%%%%%%%%%%%%%% %% FONT CMBX %% %%%%%%%%%%%%%%%%% % % \font\eightbf=cmbx10 at 8pt \font\twelvebf=cmbx12 \font\fourteenbf=cmbx10 at 14pt % \font\ninebf=cmbx9 \font\eightbf=cmbx8 % \font\sevenbf=cmbx10 at 7pt \font\sixbf=cmbx5 at 6pt % \font\sixbf=cmbx5 % \font\fivebf=cmbx10 at 5pt \font\tenbf=cmbx10 \font\sevenbf=cmbx7 \font\fivebf=cmbx5 \newfam\bffam \textfont\bffam=\tenbf \scriptfont\bffam=\sevenbf \scriptscriptfont\bffam=\fivebf \def\bf{\fam\bffam\tenbf} % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% HAUT-DE-PAGE (EX. : AUTEUR COURANT ET TITRE COURANT) %% %% BAS-DE-BAGE %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \newif\ifpagetitre \pagetitretrue \newtoks\hautpagetitre \hautpagetitre={\hfil} \newtoks\baspagetitre \baspagetitre={\hfil} \newtoks\auteurcourant \auteurcourant={\hfil} \newtoks\titrecourant \titrecourant={\hfil} \newtoks\hautpagegauche \hautpagegauche={\hfil\the\auteurcourant\hfil} \newtoks\hautpagedroite \hautpagedroite={\hfil\the\titrecourant\hfil} \newtoks\baspagegauche \baspagegauche={\hfil\tenrm\folio\hfil} \newtoks\baspagedroite \baspagedroite={\hfil\tenrm\folio\hfil} \headline={\ifpagetitre\the\hautpagetitre \else\ifodd\pageno\the\hautpagedroite \else\the\hautpagegauche\fi\fi } \footline={\ifpagetitre\the\baspagetitre \global\pagetitrefalse \else\ifodd\pageno\the\baspagedroite \else\the\baspagegauche\fi\fi } \def\nopagenumbers{\def\folio{\hfil}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% UN SEMBLANT DES GUILLEMETS FRANCAIS %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \def\og{\leavevmode\raise.3ex \hbox{$\scriptscriptstyle\langle\!\langle$~}} \def\fg{\leavevmode\raise.3ex \hbox{~$\!\scriptscriptstyle\,\rangle\!\rangle$}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% TRUC AVEC MACHIN AU-DESSUS ET BIDULE AU-DESSOUS %% %% (MODE MATHEMATIQUE) %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \def\build #1_#2^#3{\mathrel{\mathop{\kern 0pt #1} \limits_{#2}^{#3}}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \def\scs{\scriptstyle} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% FLECHES HORIZONTALES ET VERTICALES AVEC TRUC %% %% AU-DESSUS ET MACHIN AU-DESSOUS (MODE MATH) %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \def\hfl[#1][#2][#3]#4#5{\kern -#1 \sumdim=#2 \advance\sumdim by #1 \advance\sumdim by #3 \smash{\mathop{\hbox to \sumdim {\rightarrowfill}} \limits^{\scriptstyle#4}_{\scriptstyle#5}} \kern-#3} \def\vfl[#1][#2][#3]#4#5% {\sumdim=#1 \advance\sumdim by #2 \advance\sumdim by #3 \setbox1=\hbox{$\left\downarrow\vbox to .5\sumdim {}\right.$} \setbox1=\hbox{\llap{$\scriptstyle #4$}\box1 \rlap{$\scriptstyle #5$}} \vcenter{\kern -#1\box1\kern -#3}} \def\Hfl #1#2{\hfl[0mm][12mm][0mm]{#1}{#2}} \def\Vfl #1#2{\vfl[0mm][12mm][0mm]{#1}{#2}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% DIAGRAMMES COMMUTATIFS %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%% \newdimen\sumdim \def\diagram#1\enddiagram {\vcenter{\offinterlineskip \def\tvi{\vrule height 10pt depth 10pt width 0pt} \halign{&\tvi\kern 5pt\hfil$\displaystyle##$\hfil\kern 5pt \crcr #1\crcr}}} %% SYNTAXE %% $$ %% \diagram %% ... & & ... & \cr %% ... & & ... & \cr %% ................. %% ... & & ... & \cr %% \enddiagram %% $$ %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %\def\Doteq{\build{=}_{\scriptscriptstyle\bullet}^{\scriptscriptstyle\bullet}} \def\Doteq{\build{=}_{\hbox{\bf .}}^{\hbox{\bf .}}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% ESPACE VERTICAL EN MODE MATH %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \def\vspace[#1]{\noalign{\vskip #1}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%% %% ENCADREMENT %% %%%%%%%%%%%%%%%%%%% \def\boxit [#1]#2{\vbox{\hrule\hbox{\vrule \vbox spread #1{\vss\hbox spread #1{\hss #2\hss}\vss}% \vrule}\hrule}} \def\carre{\boxit[5pt]{}} %% INDIQUE LA FIN D'UNE PREUVE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \catcode`\@=11 \def\Eqalign #1{\null\,\vcenter{\openup\jot\m@th\ialign{ \strut\hfil$\displaystyle{##}$&$\displaystyle{{}##}$\hfil &&\quad\strut\hfil$\displaystyle{##}$&$\displaystyle{{}##}$ \hfil\crcr #1\crcr}}\,} \catcode`\@=12 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %\font\tengoth=eufm10 \font\tenbboard=msbm10 %\font\eightrm=cmr8 \font\eighti=cmmi8 %\font\eightsy=cmsy8 \font\eightbf=cmbx8 % \font\eighttt=cmtt8 \font\eightit=cmti8 % \font\eightsl=cmsl8 \font\eightgoth=eufm7 % \font\eightbboard=msbm7 \font\eightex=cmex10 at 8pt %\font\eightgoth=eufm7 \font\sevenbboard=msbm7 % \font\sixrm=cmr6 \font\sixi=cmmi6 %\font\sixsy=cmsy6 \font\sixbf=cmbx6 % \font\sixgoth=eufm5 at 6pt \font\sixbboard=msbm7 at 6pt %\font\fivegoth=eufm5 \font\fivebboard=msbm5 \vsize= 20.5cm \hsize=15cm \hoffset=3mm \voffset=3mm %%%%%%%%%%%%%%%%% %% DEFINITIONS %% %%%%%%%%%%%%%%%%% \def\qed{\quad\raise -2pt\hbox{\vrule\vbox to 10pt{\hrule width 4pt \vfill\hrule}\vrule}} \def\qeda{\quad\raise -2pt\hbox{\vrule\vbox to 10pt{\hrule width 4pt height9pt \vfill\hrule}\vrule}} \def\arti#1#2{\setbox50=\null\ht50=#1\dp50=#2\box50} \def\tend_#1^#2{\mathrel{ \mathop{\kern 0pt\hbox to 1cm{\rightarrowfill}} \limits_{#1}^{#2}}} \def\tendps{\tend_{n\to\infty}^{\hbox{p.s.}}} \def\lfq{\leavevmode\raise.3ex\hbox{$\scriptscriptstyle \langle\!\langle$}\thinspace} \def\rfq{\leavevmode\thinspace\raise.3ex\hbox{$\scriptscriptstyle \rangle\!\rangle$}} \font\msamten=msam10 \font\msamseven=msam7 \newfam\msamfam \textfont\msamfam=\msamten \scriptfont\msamfam=\msamseven \def\ams {\fam\msamfam\msamten} \def\hexnumber #1% {\ifcase #1 0\or 1\or 2\or 3\or 4\or 5 \or 6\or 7\or 8\or 9\or A\or B\or C\or D\or E\or F \fi} \mathchardef\leadsto="3\hexnumber\msamfam 20 \def\frac #1#2{{#1\over #2}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% FIN DE DEFINITIONS PROPRES A CE FICHIER %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% suites exactes %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \def\rarr#1#2{\smash{\mathop{\hbox to .5in{\rightarrowfill}} \limits^{\scriptstyle#1}_{\scriptstyle#2}}} \def\larr#1#2{\smash{\mathop{\hbox to .5in{\leftarrowfill}} \limits^{\scriptstyle#1}_{\scriptstyle#2}}} \def\darr#1#2{\llap{$\scriptstyle #1$}\left\downarrow \vcenter to .5in{}\right.\rlap{$\scriptstyle #2$}} \def\uarr#1#2{\llap{$\scriptstyle #1$}\left\uparrow \vcenter to .5in{}\right.\rlap{$\scriptstyle #2$}} \def\diagram#1{{\normallineskip=8pt \normalbaselineskip=0pt \matrix{#1}}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% AUTEUR(S) COURANT(S) SUR PAGES PAIRES %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \auteurcourant{\eightbf B.~Lass \hfill} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% TITRE COURANT SUR PAGES IMPAIRES %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \titrecourant{\hfill\eightbf On the combinatorics of the graph-complex} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% {\parindent=0pt\eightbf C. R. Acad. Sci. Paris, t.~334, S\'erie~I, p.~1--6, 2002 %%%%%%%%%%%%%%%%% %% RUBRIQUE(S) %% %%%%%%%%%%%%%%%%% Alg\`ebres de Lie/\hskip -.5mm{\eightbfti Lie Algebras} (Combinatoire/\hskip -.5mm{\eightbfti Combinatorics}) \vskip 1.5cm %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%% %% TITEL %% %%%%%%%%%%%%% \noindent {\fourteenbf On the combinatorics of the graph-complex } %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% MENTION "NOTE PR\'ESENT\'EE PAR ... %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \parindent=-1mm\footnote{}{ Note pr\'esent\'ee par Michel D{\sixbf UFLO.} } \vskip 15pt %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% Pr\'enoms NOMS (DES AUTEURS) %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \noindent{\bf Bodo LASS} \vskip 6pt %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% ADRESSES DES AUTEURS ET COURRIEL OU E-MAIL %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% {\parindent=0mm D\'epartement de math\'ematique, Universit\'e Louis-Pasteur, 67084 Strasbourg, France Courriel~: lass@math.u-strasbg.fr \par} \medskip {\parindent=0mm Lehrstuhl II f\"ur Mathematik, RWTH Aachen, 52062 Aachen, Allemagne Courriel~: lass@math2.rwth-aachen.de \par} \medskip %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% MENTION "(RE\C{C}U LE ..., ACCEPT\'E LE ...)" %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% (Re\c{c}u le 26 octobre 2001, accept\'e le 13 novembre 2001) } \vskip3pt \line{\hbox to 2cm{}\hrulefill\hbox to 1.5cm{}} \vskip3pt %%%%%%%%%%%%%%%%%% %% Abstract %% %%%%%%%%%%%%%%%%%% {\parindent=2cm\rightskip 1.5cm\baselineskip=9pt \item{\bf Abstract.~}{\eightrm In their recent preprint~[3] Kontsevich and Shoikhet have introduced two graph-com\-plexes: the complex on the even (resp.~odd) space in order to study the cohomology of the Lie algebra Ham$\scs _0$ (resp.~Ham$\scs _0^{odd}$) of Hamiltonian vector fields vanishing at the origin on the infinite-dimensional even (resp.~odd) space. We construct an isomorphism between those two graph-complexes, proving in particular that their cohomologies coincide. This solves a problem posed by Shoikhet. \copyright~Acad\'emie des Sciences/Elsevier, Paris} \par} \vskip 10pt %%%%%%%%%%%%%%%%%%%%%%%% %% TITRE EN FRANCAIS %% %%%%%%%%%%%%%%%%%%%%%%%% {\parindent=2cm\rightskip 1.5cm {\bfti Sur la combinatoire du complexe de graphes} \par} \vskip 10pt %%%%%%%%%%%%%%%% %% RESUME %% %%%%%%%%%%%%%%%% {\parindent=2cm\rightskip 1.5cm\baselineskip=9pt \item{\bf R\'esum\'e.~}{\eightit Dans leur pr\'epublication r\'ecente~[3], Kontsevich et Shoikhet ont introduit deux complexes de graphes~: le complexe sur l'espace pair (resp.~impair) pour \'etudier la cohomologie de l'alg\`ebre de Lie Ham$\scs _0$ (resp.~Ham$\scs _0^{odd}$) des champs vectoriels hamiltoniens sans terme constant sur l'espace pair (resp.~impair) de dimension infinie. Nous construisons un isomorphisme entre ces deux complexes de graphes, d\'emontrant notamment que leur cohomologies co\"\i ncident. Ceci r\'esoud un probl\`eme pos\'e par Shoikhet.~}{\eightrm\copyright~Acad\'emie des Sciences/Elsevier, Paris} \par} \vskip1.5pt \line{\hbox to 2cm{}\hrulefill\hbox to 1.5cm{}} \medskip \parindent=3mm \vskip 5mm %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% Version fran\c caise abr\'eg\'ee %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \noindent {\bf Version fran\c caise abr\'eg\'ee} \bigskip Le probl\`eme de calculer les groupes de cohomologie de beaucoup d'alg\`ebres de Lie peut \^etre ramen\'e au calcul des groupes de cohomologie de diff\'erents complexes de graphes (voir~[1],[2],[3]). Un {\it complexe (cohomologique) de graphes} est un complexe \vskip-10pt $$\diagram{ 0 & \rarr{}{} & {\bb C} & \rarr{0}{} & C^1 & \rarr{\delta}{} & C^2 & \rarr{\delta}{} & C^3 & \rarr{\delta}{} & \dots \cr }$$ \vskip-5pt \noindent o\`u $C^n$ est l'espace vectoriel engendr\'e par les {\it classes d'isomorphie} de paires $(G,\hbox{or}_G)$ telles que $G$ est un (multi)graphe de $n$~sommets et $\hbox{or}_G$ est une {\it orientation} d'un espace vectoriel r\'eel (ou plut\^ot d'un ensemble fini) associ\'e \`a $G$. Un {\it (multi)graphe} $G = (H;V,E)$ est un ensemble de {\it demi-ar\^etes} $H$ ($|H| = 2m$) muni de deux partitions $V$ et $E$. Les $n$ blocs $v \in V$ sont appel\'es {\it sommets} tandis que les $m$ blocs $e \in E$ de la deuxi\`eme partition sont des sous-ensembles de $H$ de cardinalit\'e deux, appel\'es {\it ar\^etes}. Une ar\^ete est une {\it boucle} si et seulement si ses deux demi-ar\^etes font partie du m\^eme sommet. Le {\it degr\'e} d'un sommet $v \in V$ est sa cardinalit\'e, c'est-\`a-dire le nombre de ses demi-ar\^etes. On dit que $v$ est {\it pair} (resp.~{\it impair}) si et seulement si son degr\'e est pair (resp.~impair). Notons $Q$ (resp.~$O$) l'ensemble des sommets pairs (resp.~impairs) de~$G$. On {\it oriente} un ensemble fini (ou bien l'espace r\'eel engendr\'e par cet ensemble) en le munissant d'une {\it forme alternante} (multilin\'eaire) $[ \; \cdot \; ]$ prenant ses valeurs dans l'ensemble~$\{1,-1\}$. \'Evidemment, le nombre d'orientations diff\'erentes vaut deux. L'union disjointe de plusieurs ensembles orient\'es est naturellement {\it pseudo-orient\'ee} (i.e.~orient\'ee par rapport \`a la partition) par le produit ordinaire des formes alternantes des blocs de la partition. Une {\it pseudo-orientation} d'un ensemble partitionn\'e est alors une forme alternante par rapport \`a toutes les permutations respectant la partition. \'Evidemment, le nombre de pseudo-orientations est aussi \'egal \`a deux. % L'orientation d'un graphe dite {\it paire} est la pseudo-orientation de l'union disjointe $V \uplus (\biguplus_{e \in E} e)$; autrement dit, on \hbox{oriente} l'ensemble~$V$ et, pour chaque ar\^ete $e \in E$, l'ensemble des deux demi-ar\^etes de~$e$ (voir~[2],[3],[4]). L'orientation d'un graphe dite {\it impaire}, cependant, est la pseudo-orientation de l'union disjointe $Q \uplus (\biguplus_{v \in V} v)$, ce qui veut dire que l'on oriente l'ensemble~$Q$ des sommets pairs et, pour chaque sommet $v \in V$, l'ensemble des demi-ar\^etes de~$v$ (voir~[3]). Un {\it automorphisme}~$P$ du graphe~$G$ est une permutation~$P_H$ de l'ensemble des demi-ar\^etes~$H$ pr\'eservant les structures de~$V$ et de~$E$. Voil\`a pourquoi $P$ agit sur l'orientation paire (resp.~impaire) par une multiplication avec un nombre not\'e $\Theta_{even}(P) \in \{1,-1\}$ (resp.~$\Theta_{odd}(P)$). Pour la permutation~$T$ des deux demi-ar\^etes d'une boucle, on a, par exemple, $\Theta_{even}(T) = \Theta_{odd}(T) = -1$, puisque l'orientation d'une seule ar\^ete (pour l'orientation paire) et d'un seul sommet (pour l'orientation impaire) est renvers\'ee. De fa\c con plus g\'en\'erale, nous d\'emontrerons pour chaque automorphisme~$P$ que $\Theta_{even}(P) = \Theta_{odd}(P) = \hbox{sign}(P_H) \cdot \hbox{sign}(P_V)$, si $P_V$ d\'esigne la permutation induite sur l'ensemble des sommets de~$G$. % On impose la relation $(G,-\hbox{or}_G) = -(G,\hbox{or}_G)$, de sorte que $(G,\hbox{or}_G) = 0$ si et seulement si $G$ a un automorphisme tel que $\Theta_{even}(P) = \Theta_{odd}(P) = -1$, comme dans le cas des boucles (d'o\`u $C^1 = 0$). Par cons\'equent, les g\'en\'erateurs du complexe de graphes sont (\`a une multiplication par $\pm 1$ pr\`es) les m\^emes pour l'orientation paire et impaire. En fait, nous montrerons que ces deux orientations correspondent de fa\c con naturelle \`a une orientation ordinaire de l'ensemble $H \uplus V$, que nous appelons {\it universelle}. Cette correspondance permet \'egalement de comparer les diff\'erentielles, d\'efinies par Kontsevich et Shoikhet~[3] de la mani\`ere suivante~: % Pour $\delta_{even}(G)$ et $\delta_{odd}(G)$ on prend toujours la somme sur {\it tous} les {\lfq gonflages\rfq} de {\it tous} les sommets de~$G$, o\`u un gonflage d'un sommet veut dire que l'ensemble de ses demi-ar\^etes est partitionn\'e (de fa\c con non-ordonn\'ee) en deux blocs (ce sont des sommets nouveaux), qui sont reli\'es par une ar\^ete nouvelle (c'est-\`a-dire chaque bloc nouveau obtient une demi-ar\^ete nouvelle, et ces deux demi-ar\^etes nouvelles forment une ar\^ete). \'Evidemment, chaque gonflage augmente le nombre de sommets d'une unit\'e. Soit $G = (H;V,E)$ avec $V = \{v_1,v_2,\dots,v_n\}$; $E = \{e_1,\dots,e_m\}$; $v_1 = \{h_{a_1},\dots,h_{x_1},h_{y_1},\dots,h_{b_1}\}$, $v_2 = \{h_{a_2},\dots,h_{b_2}\}$, \dots, $v_n = \{h_{a_n},\dots,h_{b_n}\}$; $e_1 = \{h_{i_1},h_{t_1}\}$, \dots, $e_m = \{h_{i_m},h_{t_m}\}$; et $H = \biguplus_{k=1}^n v_k = \biguplus_{k=1}^m e_k$. Consid\'erons un gonflage de $v_1$ avec $v'_1 = \{h_{a_1},\dots,h_{x_1}\}$, $v''_1 = \{h_{y_1},\dots,h_{b_1}\}$ et avec la nouvelle ar\^ete $\{h'_1,h''_1\}$. Pour l'orientation, Kontsevich et Shoikhet~[3] d\'efinissent alors $$ [v'_1,v''_1,v_2,\dots,v_n] \cdot [h'_1,h''_1] \prod_{k=1}^m [h_{i_k},h_{t_k}] \; := \; [v_1,v_2,\dots,v_n] \cdot \prod_{k=1}^m [h_{i_k},h_{t_k}] \quad \hbox{sur l'espace pair et} $$ $$ [v'_1,v''_1,v_2,\dots,v_n]_Q \cdot [h'_1,h_{a_1},\dots,h_{x_1}] [h''_1,h_{y_1},\dots,h_{b_1}] \prod_{k=2}^n [h_{a_k},\dots,h_{b_k}] $$ $$ := \; [v_1,v_2,\dots,v_n]_Q \cdot \prod_{k=1}^n [h_{a_k},\dots,h_{b_k}] \quad \hbox{sur l'espace impair,} $$ o\`u $[ \; \cdot \; ]_Q$ signifie qu'il faut supprimer tous les sommets impairs entre les crochets. % Nous d\'emontrons le th\'eor\`eme suivant ($o := |O|$, $n := |V|$). \medskip {\sc Th\'eor\`eme.~-} {\it Du point de vue de l'orientation universelle, il suffit de multiplier l'orientation de chaque graphe~$G$ par $(-1)^{n-o/2}$ pour obtenir un isomorphisme entre l'orientation paire et impaire qui est compatible avec les diff\'erentielles. C'est pourquoi le complexe sur l'espace pair et le complexe sur l'espace impair sont isomorphes.} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \vskip 6pt \centerline{\hbox to 2cm{\hrulefill}} \vskip 9pt %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \bigskip %%%%%%%%%%%%%%%%%%%%%%% %% TEXTE PRINCIPAL %% %%%%%%%%%%%%%%%%%%%%%%% \noindent {\bf 1. Introduction} \medskip The underlying vector space of the Lie algebra Ham$_0(N)$ of Hamiltonian vector fields vanishing at the origin {\it on the even space} is equal to $\bigoplus_{d \ge 2} S^d V$, where $V$ is the defining $2N$-dimensional representation of sp$(2N)$, $S^2 V = \hbox{sp}(2N)$, etc. It is classical (see [1] and [2]) that the cochain complex $$ C_{Lie}^\bullet(\hbox{Ham}_0(N);\bb C) = \bigoplus_{d_2 \ge 0, \, d_3 \ge 0, \, \dots} \textstyle \bigwedge^{d_2}(S^2 V) \wedge \bigwedge^{d_3}(S^3 V) \wedge \cdots $$ is quasi-isomorphic to its sp-invariant part $[C_{Lie}^\bullet(\hbox{Ham}_0(N);\bb C)]^{sp}$. Let us consider $\bigwedge^{d_2}(S^2 V) \wedge \bigwedge^{d_3}(S^3 V) \wedge \cdots$ as a quotient space of $(V^{\otimes 2})^{\otimes d_2} \otimes (V^{\otimes 3})^{\otimes d_3} \otimes \cdots$. If $2d_2 + 3d_3 + \cdots =: 2m$ is even, then any perfect matching on $H := \{1,2,\dots,2m\}$ defines a (multi)graph $G = (H;V,E)$ with $2m$ half-edges in~$H$, $m$ edges in~$E$ and $n := d_2 + d_3 + \cdots$ vertices in~$V$, $d_2$ of which have degree~$2$, $d_3$ of which have degree~$3$, etc. There exists a canonical symplectic invariant in $\bigwedge^2 V \hookrightarrow V^{\otimes 2}$, that can be chosen in $V_i \otimes V_j$ for every edge $\{i,j\} \in E$ ($i,j \in H$) to obtain an invariant in $V^{\otimes 2m}$. By the main theorem of invariant theory those tensors form a basis of $[V^{\otimes 2m}]^{sp}$ and there are no invariants for $(2d_2 + 3d_3 + \cdots)$ odd, if $2N = \hbox{dim}(V)$ is sufficiently large. The (multi)graphs $G = (H;V,E)$ carry the {\it even orientation} corresponding to the exterior product over the set of vertices (in the definition of the cochain complex) and to the exterior products over the set of half-edges of any edge (in the definition of the canonical symplectic invariants). In fact, the consideration of those oriented graphs modulo the action of symmetric groups provides a bijection with the elements of the complex $[C_{Lie}^\bullet(\hbox{Ham}_0(N);\bb C)]^{sp}$. The differentials of this complex can be easily expressed in graph-theoretical terms (see Section 3). Therefore the cohomology of the resulting {\it graph-complex on the even space} is equal to the cohomology of Hamiltonian vector fields vanishing at the origin on the infinite-dimensional even space. The underlying vector space of the Lie algebra Ham$_0^{odd}(N)$ of Hamiltonian vector fields vanishing at the origin {\it on the odd space} is equal to $\bigoplus_{d \ge 2} \bigwedge^d V$, where $V$ is the defining $N$-dimensional representation of $\hbox{sp}^{odd}(N) = \hbox{\tengoth o}(N)$, $\bigwedge^2 V = \hbox{sp}^{odd}(N)$, etc. The main difference from the even case is that the canonical invariant lies in $S^2 V \hookrightarrow V^{\otimes 2}$ and that the cochain complex is given by $$ C_{Lie}^\bullet(\hbox{Ham}_0^{odd}(N);\bb C) = \bigoplus_{d_2 \ge 0, \, d_3 \ge 0, \, \dots} \textstyle \bigwedge_{super}^{d_2}(\bigwedge^2 V) \wedge_{super} \bigwedge_{super}^{d_3}(\bigwedge^3 V) \wedge_{super} \cdots , $$ where $\wedge_{super}$ denotes the graded exterior product (which is anticommutative if the degrees of both factors are even and commutative otherwise). Therefore the cohomology of Hamiltonian vector fields vanishing at the origin on the infinite-dimensional odd space is equal to the cohomology of the {\it graph-complex on the odd space}, which differs from the graph-complex on the even space just by its orientation. In fact, the (multi)graphs $G = (H;V,E)$ carry the {\it odd orientation} corresponding to the graded exterior product over the set of vertices and to the exterior products over the set of half-edges of any vertex, and the differentials of the graph-complex on the odd space can be expressed in graph-theoretical terms with the help of the odd orientation. With respect to this situation Shoikhet suggested that {\it ``you obtain at first look a different graph-complex, but our idea with Kontsevich was that the two cohomologies coincide - and to prove it by something like an isomorphism of orientations. {\dots} So, up to me, a really interesting problem in this direction is to prove that the even and the odd graph-complexes give the same orientation.'' } We solve this problem in the next two sections using still another orientation, which we call {\it universal} and which is more classical from the point of view of algebraic topology. \bigskip \noindent {\bf 2. The generators of the graph complexes on the even space and on the odd space} \medskip A (multi)graph $G = (H;V,E)$ is a set of {\it half-edges}~$H$ consisting of $2m$ elements, together with two partitions of this set into disjoint unions of subsets. The $n$ blocks of the first partition $v \in V$ are called {\it vertices} whereas the $m$ blocks of the second partition $e \in E$ are two-element subsets of~$H$ called {\it edges}. The {\it degree} of a vertex $v \in V$ is the number of its half-edges and $v$ is called {\it even} (resp.~{\it odd}) iff its degree is even (resp.~odd). % Let $[ \; \cdot \; ]$ be an alternating (multilinear) form on (the real space generated by) the disjoint union of all half-edges $h_1,h_2,\dots,h_{2m}$ and all vertices $v_1,v_2,\dots,v_n$. The form $[ \; \cdot \; ]$ is uniquely defined by determining the value $[h_1,h_2,\dots,h_{2m};v_1,v_2,\dots,v_n] = [v_1,v_2,\dots,v_n;h_1,h_2,\dots,h_{2m}] \in \{1,-1\}$. A (multi)graph with such a form will be called {\it universally oriented}. In order to determine $[h_1,h_2,\dots,h_{2m};v_1,v_2,\dots,v_n]$ we can write the elements $h_1,h_2,\dots,h_{2m};v_1,v_2,\dots,v_n$ in different orders, all of which will start with the half-edges. One natural order makes use of the block structure determined on the set of half-edges~$H$ by the set of edges~$E$, i.e.~we consider $[h_{i_1},h_{t_1};h_{i_2},h_{t_2};\dots;h_{i_m},h_{t_m};v_{j_1},v_{j_2},\dots,v_{j_n}]$, where $\{h_{i_1},h_{t_1}\}$, $\{h_{i_2},h_{t_2}\}$, \dots, $\{h_{i_m},h_{t_m}\}$ are the $m$ edges of the graph $G$. Permuting whole edges does not change the value of $[h_{i_1},h_{t_1};h_{i_2},h_{t_2};\dots;h_{i_m},h_{t_m};v_{j_1},v_{j_2},\dots,v_{j_n}]$ whereas permuting two half-edges within an edge does change this value. Therefore, orienting a graph universally is canonically the same as orienting the set $V$ of vertices and orienting, for every edge $e \in E$, the set of half-egdes of~$e$. This corresponds to the orientation in the case of the graph-complex on the even space (see [2],[3],[4]). Another natural order makes use of the block structure determined on the set of half-edges~$H$ by the set~$Q$ ($|Q|=q$) of even vertices and by the set~$O$ ($|O|=o$) of odd vertices. In other words, we consider $$ w \; := \; [h_{a_1},\dots,h_{b_1}; \dots; h_{a_q},\dots,h_{b_q}; h_{c_1},\dots,h_{d_1}; \dots; h_{c_o},\dots,h_{d_o}; v_{f_1},\dots,v_{f_o}; v_{g_1},\dots,v_{g_q}], \leqno (*) $$ where $\{h_{a_1},\dots,h_{b_1}\}$, \dots, $\{h_{a_q},\dots,h_{b_q}\}$ as well as $v_{g_1},\dots,v_{g_q}$ are the $q$ even vertices of $G$ and where $v_{f_1} = \{h_{c_1},\dots,h_{d_1}\}$, \dots, $v_{f_o} = \{h_{c_o},\dots,h_{d_o}\}$ are its $o$ odd vertices. Permuting whole even vertices (i.e.~their sets of half-edges) does not change the value of $w$. But permuting whole odd vertices (i.e.~their sets of half-edges) does not change $w$ either, because in that case the vertices $v_{f_1},\dots,v_{f_o}$ have to be permuted in the same way by our convention. On the other hand, permutations within whole vertices (i.e.~within their sets of half-edges) as well as permutations of $Q = \{v_{g_1},\dots,v_{g_q}\}$ do change $w$. Therefore, orienting a graph universally is equivalent to orienting the set $Q$ of even vertices and for every vertex $v \in V$ orienting the set of half-egdes of~$v$. This corresponds to the orientation in the case of the graph-complex on the odd space (see~[3]). An {\it automorphism} $P$ of the graph~$G$ is an arbitrary permutation $P_H$ of the set~$H$ which preserves the structures of~$V$ and~$E$. In particular, it induces permutations $P_Q$ of $Q$, $P_O$ of $O$ and $P_V$ of the set~$V$ of vertices. The automorphism $P$ acts on the even (resp.~odd) orientation by a multiplication by $$ \Theta_{even}(P) \; = \; \hbox{sign}(P_V) \, \cdot \, \prod_{e \in E}\epsilon_P(e), \quad \Theta_{odd}(P) \; = \; \hbox{sign}(P_Q) \, \cdot \, \prod_{v \in V}\epsilon_P(v), \quad \hbox{respectively}, $$ where $\epsilon_P(e)$ equals $1$ (resp.~$-1$) if and only if $P$ is sign-preserving (resp.~sign-reversing) on (the set of half-edges of) $e$, and where $\epsilon_P(v)$ equals $1$ (resp.~$-1$) if and only if $P$ is sign-preserving (resp.~sign-reversing) on (the set of half-edges of) $v$. Finally, $P$ acts on the universal orientation by a multiplication by $\hbox{sign}(P_H) \cdot \hbox{sign}(P_V)$. Now our established correspondence between those three orientations implies: $$ \Theta_{even}(P) \; = \; \Theta_{odd}(P) \; = \; \hbox{sign}(P_H) \cdot \hbox{sign}(P_V). $$ The $n$-dimensional generators of the different graph-complexes are always {\it equivalence classes} of pairs $(G,\hbox{or}_G)$, where $G$ is a graph with $n$ vertices (whose degrees are all greater than one), and where $\hbox{or}_G$ is one of our orientations. One imposes the relation $(G,-\hbox{or}_G) = -(G,\hbox{or}_G)$, so that $(G,\hbox{or}_G) = 0$ if and only if $G$ has an automorphism whose sign is $-1$. We have established the following theorem. \medskip {\sc Theorem 1.~-} {\it There is a canonical correspondence between the generators of the graph-complex on the even space, the generators of the graph-complex on the odd space and the generators of the graph-complex using the universal orientation.}\qed \bigskip \noindent {\bf 3. The differentials of the graph complexes on the even space and on the odd space} \medskip In order to get an isomorphism of complexes we have to verify whether our canonical correspondences between the generators are compatible with the differentials. Let us first define a differential for the universal orientation. It is essentially given by the dual of the operation of contracting each edge. More precisely, some vertex $v_s \in V$ has to be split up into two vertices $v'_s$ and $v''_s$, i.e.~the set of its half-edges has to be partitioned. Moreover, a new edge $\{h'_s,h''_s\}$ will be introduced, where $h'_s \in v'_s$ and $h''_s \in v''_s$. Finally, the orientation can be defined by $$\openup2pt\eqalign{ &[h_1,h_2,\dots,h_{2m},h'_s,h''_s;v_1,v_2,\dots,v_{s-1},v'_s,v''_s,v_{s+1},\dots,v_n] \cr &:= \; (-1)^{s-1} [h_1,h_2,\dots,h_{2m};v_1,v_2,\dots,v_{s-1},v_s,v_{s+1},\dots,v_n]. \cr }$$ It is evident that this definition is well defined and leads in fact to a differential $\delta$ (for the differential, the sum over all splittings is taken; if we split twice in the same way but in a different order, then the corresponding terms cancel, i.e.~$\delta^2 = 0$). The expression of the definition in terms of the block structure determined by the set of edges of the graph~$G$ is immediate. It leads precisely to the definition of~[3] (and~[2]) in the case of the graph-complex on the even space (corresponding to the cochain complex $C_{Lie}^\bullet(\hbox{Ham}_0;\bb C)$). In order to find the expression for the definition in terms of the block structure determined by the set of vertices of~$G$, it is sufficient to consider what happens to $w$ (see equation $(*)$ in the preceding section) if we partition $v_{f_1}$ or $v_{g_1}$ (these vertices are both on odd places because the number~$o$ of odd vertices of $G$ is even). First suppose that $v_{f_1} = \{ h_{c_1},\dots,h_{x_1},h_{y_1},\dots,h_{d_1} \}$ and that the half-edges of $v_{f_1}$ are partitioned into the blocks $v'_{f_1} = \{ h'_{f_1},h_{c_1},\dots,h_{x_1} \}$ and $v''_{f_1} = \{ h''_{f_1},h_{y_1},\dots,h_{d_1} \}$, where $\{h'_{f_1},h''_{f_1}\}$ is the new edge. If $|v'_{f_1}|$ is odd and $|v''_{f_1}|$ is even, then, according to our universal orientation, $w$ is transformed into $$\openup2pt\eqalign{ [&h_{a_1},\dots,h_{b_1}; \dots; h_{a_q},\dots,h_{b_q}; h'_{f_1},h''_{f_1}; h_{c_1},\dots,h_{x_1}; h_{y_1},\dots,h_{d_1}; h_{c_2},\dots,h_{d_2}; \dots; h_{c_o},\dots,h_{d_o}; \cr &v'_{f_1},v''_{f_1};v_{f_2},\dots,v_{f_o}; v_{g_1},\dots,v_{g_q}] \; = \cr [&h_{a_1},\dots,h_{b_1}; \dots; h_{a_q},\dots,h_{b_q}; h''_{f_1},h_{y_1},\dots,h_{d_1}; h'_{f_1},h_{c_1},\dots,h_{x_1}; h_{c_2},\dots,h_{d_2}; \dots; h_{c_o},\dots,h_{d_o}; \cr &v'_{f_1},v_{f_2},\dots,v_{f_o}; v''_{f_1},v_{g_1},\dots,v_{g_q}] \cdot (-1). \cr }$$ If $|v'_{f_1}|$ is even and $|v''_{f_1}|$ is odd, we get $$\openup2pt\eqalign{ [&h_{a_1},\dots,h_{b_1}; \dots; h_{a_q},\dots,h_{b_q}; h'_{f_1},h_{c_1},\dots,h_{x_1}; h''_{f_1},h_{y_1},\dots,h_{d_1}; h_{c_2},\dots,h_{d_2}; \dots; h_{c_o},\dots,h_{d_o}; \cr &v''_{f_1},v_{f_2},\dots,v_{f_o}; v'_{f_1},v_{g_1},\dots,v_{g_q}] \cdot (-1). \cr }$$ Now suppose that $v_{g_1} = \{ h_{a_z},\dots,h_{x_z},h_{y_z},\dots,h_{b_z}\}$ and that the half-edges of $v_{g_1}$ are partitioned into the blocks $v'_{g_1} = \{ h'_{g_1},h_{a_z},\dots,h_{x_z} \}$ and $v''_{g_1} = \{ h''_{g_1},h_{y_z},\dots,h_{b_z} \}$. If $|v'_{g_1}|$ and $|v''_{g_1}|$ are odd, then $$\openup2pt\eqalign{ [&h_{a_1},\dots,h_{b_1}; \dots; h'_{g_1}, h''_{g_1}; h_{a_z},\dots,h_{x_z}; h_{y_z},\dots,h_{b_z}; \dots; h_{a_q},\dots,h_{b_q}; h_{c_1},\dots,h_{d_1}; \dots; h_{c_o},\dots,h_{d_o}; \cr &v_{f_1},\dots,v_{f_o}; v'_{g_1}, v''_{g_1}; v_{g_2},\dots,v_{g_q}] \; = \cr [&h_{a_1},\dots,h_{b_1}; \dots; h_{a_{z-1}},\dots,h_{b_{z-1}}; h_{a_{z+1}},\dots,h_{b_{z+1}}; \dots; h_{a_q},\dots,h_{b_q}; \cr &h'_{g_1},h_{a_z},\dots,h_{x_z}; h''_{g_1},h_{y_z},\dots,h_{b_z}; h_{c_1},\dots,h_{d_1}; \dots; h_{c_o},\dots,h_{d_o}; \cr &v'_{g_1},v''_{g_1},v_{f_1},\dots,v_{f_o}; v_{g_2},\dots,v_{g_q}]. \cr }$$ Finally, if $|v'_{g_1}|$ and $|v''_{g_1}|$ are even, we get $$\openup2pt\eqalign{ [&h_{a_1},\dots,h_{b_1}; \dots; h'_{g_1},h_{a_z},\dots,h_{x_z}; h''_{g_1},h_{y_z},\dots,h_{b_z}; \dots; h_{a_q},\dots,h_{b_q}; h_{c_1},\dots,h_{d_1}; \dots; h_{c_o},\dots,h_{d_o}; \cr &v_{f_1},\dots,v_{f_o}; v'_{g_1},v''_{g_1},v_{g_2},\dots,v_{g_q}] \cdot (-1). \cr }$$ If there were no factors $-1$, the preceding relations would be precisely those used by Kontsevich and Shoikhet~[3] for defining their differential of the graph-complex on the odd space (corresponding to the cochain complex $C_{Lie}^\bullet(\hbox{Ham}_0^{odd};\bb C)$). In order to get rid of those factors $-1$, however, {\it it suffices to use the isomorphism which multiplies every graph~$G$ (i.e.~its orientation) by $(-1)^{n-o/2}$, if $G$ has $n$ vertices, $o$ of which are odd}. In fact, the multiplication by $(-1)^n$ provides a minus for every differential, whereas the multiplication by $(-1)^{o/2}$ (note that $o$ is always even) kills this minus if two new odd vertices are created. Therefore we have proved the following theorem. \medskip {\sc Theorem 2.~-} {\it The cohomological graph-complex on the even space, the cohomological graph-complex on the odd space and the universally oriented cohomological graph-complex are isomorphic. In particular, their cohomologies coincide.}\qed \medskip By the classical arguments reproduced in the introduction this implies our final corollary, a surprising duality relation which (according to Feigin) seems to be difficult to derive by different means. \medskip {\sc Corollary.~-} {\it The cohomologies of the Lie algebras $\hbox{Ham}_0$ and $\hbox{Ham}_0^{odd}$ of Hamiltonian vector fields vanishing at the origin on the infinite-dimensional even and odd spaces coincide.}\qed \bigskip {\it Acknowledgements.} I should like to thank B.~Shoikhet for kindly sending me the interesting preprint~[3], for asking me the question that led to this Note and for very useful discussions; I am also grateful to P.~Cartier and D.~Foata for all their help and encouragement. \bigskip %%%%%%%%%%%%%%%%%%%%%%% %% BIBLIOGRAPHIE %% %%%%%%%%%%%%%%%%%%%%%%% {\bf References} \medskip {\baselineskip=9pt\eightrm \item{[1]} Fuchs D.B., Kogomologii beskonechnomernykh algebr Lie, Nauka, Moskva, 1984. \smallskip \item{[2]} Kontsevich M., Formal (non)-commutative symplectic geometry, in~: Corwin L., Gelfand I., Lepowsky J.~(Eds.), The Gelfand Mathema\-tical Seminars 1990-1992, Birkh\"auser, 1993, pp.~173-187. \smallskip \item{[3]} Kontsevich M., Shoikhet B., Formality conjecture, geometry of complex manifolds and combinatorics of the graph-complex, preprint. \smallskip \item{[4]} Lass B., Une conjecture de Kontsevich et Shoikhet et la caract\'eristique d'Euler, Rev.~Math.~Pures Appl.~(\`a para\^\i tre). \smallskip } \end