scieee Open visual document viewer

A flatness property for filtered D-modules

Castro Jiménez, Francisco Jesús; Granger, Michel

Abstract

Let M be a coherent module over the ring DX of linear differential operators on an analytic manifold X and let Z1, · · · , Zk be k germs of transverse hypersurfaces at a point x ∈ X. The Malgrange-Kashiwara V-filtrations along these hypersurfaces, associated with a given presentation of the germ of M at x, give rise to a multifiltration U•(M) of Mx as in Sabbah’s paper [9] C. Sabbah, Proximité evanescente I. La structure polaire d’un D–module Compositio Math. 62 (1987) 283-319 and to an analytic standard fan in a way similar to [3] A. Assi., F. Castro-Jiménez and M. Granger, The analytic standard fan of a D-module, J. Pure Appl. Algebra 164 (2001) 3-21. We prove here that this standard fan is adapted to the multifiltration, in the sense of C. Sabbah. This result completes the proof of the existence of an adapted fan in [9] C. Sabbah, Proximité evanescente I. La structure polaire d’un D–module Compositio Math. 62 (1987) 283-319, for which the use of [8] C. Sabbah and F. Castro, Appendice à “proximité evanescente” I. La structure polaire d’un D–module, Compositio Math. 62 (1987) 320-328. is not possible.

Full text

a Xi :ma h/0505465 1 [ma h.AG] 23 May 2005 A la ness p ope y o il e ed D-modules F.J. Cas o-Jim´enez and M. G ange Abs ac Le Mbe a cohe en module o e he ing DXo linea di e en ial ope a o s on an analy ic mani old Xand le Z1,· · · , Zkbe kge ms o ans e se hype su aces a a poin x∈X. The Malg ange-Kashiwa a V- il a ions along hese hype su aces, associa ed wi h a gi en p esen a ion o he ge m o Ma x, gi e ise o a mul i il a ion U•(M) o Mxas in Sabbah’s pape [9] and o an analy ic s anda d an in a way simila o [3]. We p o e he e ha his s anda d an is adap ed o he mul i il a ion, in he sense o C. Sabbah. This esul comple es he p oo o he exis ence o an adap ed an in [9], o which he use o [8] is no possible. 1 In oduc ion Le us conside a cohe en module Mo e he ing DXo di e en ial ope a o s on an analy ic mani old X. Fo any smoo h hype su ace Zo X, Malg ange and Kashiwa a de ined a il a ion along Z o DXand he no ion o a good il a ion o M. Gi en a se o ans e se smoo h hype su aces Z1,...,Zk, Sabbah conside ed in [9] mul i il a ions o Mindexed by k-uples o ela i e in ege s. To be p ecise he deal wi h linea combina ions o e Q+o he il a ions V(j)along each hype su ace Zjand wi h e inemen s VΓo he o iginal mul i il a ion associa ed wi h each a ional polyed al simplicial cone in he posi i e quad an o (Qk)⋆. The o iginal mul i il a ion is he one which co esponds o he case Γ = Nk. The aim o his pape is o cla i y he la ness p ope ies which appea in [9], namely o p o e he exis ence o a an E, such ha o any cone Γ in his an, he Rees module o he il a ion VΓis la o e he a ine ing AΓo he o ic blowing-up o Ckassocia ed wi h he an E. Such a an is called an adap ed an in [9]. The eason o his cla i ica ion is ha he p oo in [9] depends on he appendix [8] in which he main ool is a di ision heo em in Rees ings o di e en ial ope a o s which is no co ec as s a ed. Indeed, he in ini e p ocess ha i s p oo sugges s, would equi e monomials o unbounded deg ees in he di e en ial a iables o he ing o ope a o s. One o he main consequences o he exis ence o an adap ed an as de eloped in [9] is ha we hus comple e he p oo o he exis ence o non i ial unc ional equa ions o Be ns ein-Sa o ype o a k–uple o unc ions, ollowing he a gumen o Sabbah in [9]. I should be emphasized he e ha his p oblem in ol es a DX×Ck- module na u ally associa ed wi h a k-uple o unc ions on X, and ha in his si ua ion a mul i il a ion, along he ans e se hype su aces j= 0 due o he ac o Ck, appea s in a na u al way. We mus no e he e ha he p oo o he exis ence o Be ns ein-Sa o equa ions has al eady been comple ed in Bahloul’s pape [4], by a di e en me hod which a oids he e e ence o a la ness p ope y in ol ed in he no ion o an adap ed an. Bahloul uses ins ead he analy ic s anda d an as de ined in [3], and desc ibed also in [5]. I is he e o e no comple ely su p ising ha he adap ed an wan ed in [9] u ns ou in ac o be he analy ic s anda d an. This emphasizes he in e es o Bahloul’s p oo which also has he ad an age o being cons uc i e. This cons uc i eness is aluable in an algeb aic se ing also. In bo h p oo s, Bahloul’s and Sabbah’s , he la e which is comple ed by his pape , he main s ep is he p oo o he goodness o he so-called sa u a ed il a ion. The de ails a e gi en in sec ion 2.3. The s a ing poin o he p oo o ou main heo em (Theo em 4.3) is a c i e ion o la ness by M. He mann and U. O banz (see [7]), o g aded modules o e g aded ings, whe e he g ading is indexed by an a bi a y commu a i e g oup. In he s a emen o his esul he e is no e e ence o any ini eness p ope y. In ou case he indexing g oup will be Zk, and he ing a conical sub ing o he ing o Lau en polynomials in k a iables wi h he ob ious mul ig ading by monomials. The main ing edien o he p oo o Theo em 4.3 is hen 1 he exis ence o a simul aneous L–s anda d basis o a submodule N ⊂ D o all Lin a cone o he analy ic s anda d an o N(see [3]). The plan o he pape is as ollows. In sec ion 2, we ecall wi h mo e de ails wha V-mul i il a ions, and Rees modules a e, and he e inemen s o hese mul i il a ions wi h espec o a a ional polyhed al simplicial cone. We end his sec ion by de ining he ibe a ze o o hese Rees modules seen as modules o e he ing o an a ine o ic blowing up. We aim o p o e ha his is he ibe o a la de o ma ion i he cone is aken in he analy ic s anda d an. In sec ion 3 we ecall he no ion o an analy ic s anda d an as de eloped in [3], and we ske ch he easy gene alisa ion which we need om he cyclic case ea ed in [3] o he gene al case. In sec ion 4 we inally p o e ha he analy ic s anda d an is an adap ed an in he sense o [9]. Acknowledgemen s.- Du ing he p epa a ion o his pape bo h au ho s ha e been pa ially suppo ed by Acci´on In eg ada-Picasso HF2004-0117. Fi s au ho has been also pa ially suppo ed by MTM2004-01165. Bo h au ho s a e g a e ul o he D´epa emen de Ma h´ema iques e Applica ions o he ´ Ecole No male Sup´e ieu e (Pa is) whe e he i s was in i ed du ing he p epa a ion o he inal pa o his wo k. They also hank Jeni e G ange o he p oo eading. 2 Mul i il a ions, Rees ings and Rees modules Le us deno e by O=C{x1,...,xn} he complex con e gen powe se ies ing and by D he ing o ge ms o linea di e en ial ope a o s wi h holomo phic coe icien s (i.e. D=O[∂] = O[∂1, . . . , ∂n] whe e ∂iis he pa ial de i a i e wi h espec o xiand he p oduc in Dis de ined by he Leibni z’s ule: ∂ia=a∂i+∂i(a), o each a∈ O). An ope a o Pin Dcan be w i en as P=X β∈Nn pβ(x)∂β=X α,β∈Nn pαβxα∂β whe e he i s sum is ini e, α= (α1,...,αn), β = (β1,...,βn), xα=xα1 1···xαn n,∂β=∂β1 1···∂βn n,pβ(x)∈ O and pαβ ∈C. Fo each i= 1,...,n, le us emembe ha he V- il a ion on D, wi h espec o he hype su ace xi= 0, was de ined by Malg ange and Kashiwa a as: V(i) ℓ(D) = V(i) ℓ={P=X α,β∈Nn pαβxα∂β∈ D | βi−αi≤ℓ} o each ℓ∈Z. The amily (V(i) ℓ)ℓ∈Zis an inc easing exhaus i e il a ion on he ing D. Fo i= 1, he associa ed g aded ing g V(1) (D) = M ℓ V(1) ℓ V(1) ℓ−1 is isomo phic o he ing C{x2, . . . , xn}[x1, ∂1, ∂2,...,∂n], g aded by he so-called V(1)–g adua ion, whe e he homogeneous elemen s o V(1)–deg ee ℓa e X α,β∈Nn;β1−α1=ℓ pαβxα∂β. We ha e simila desc ip ions o i= 2,...,k. We can also conside a V(i)- il a ion on he ee module D jus by de ining V(i) ℓ(D ) = (V(i) ℓD) and mo e gene ally, o any ec o m= (m1,...,m )∈Z , one can de ine he shi ed il a ion V(i)[m]ℓ(D ) = ⊕ j=1V(i) ℓ−mjD. Such a il e ed module is called a V(i)- il e ed ee module. All he D-modules conside ed will be le modules. 2 De ini ion 2.1 Le Mbe a D-module. We say ha a il a ion U(i)(M)indexed by Zis a good V(i)- il a ion i he e exis s a p esen a ion M=D Nas a quo ien by a le submodule No D , and a weigh ec o msuch ha U(i) ℓ(M) = π(V(i)[m]ℓ(D )) whe e πis he p ojec ion D → M. 2.1 V-mul i il a ions and VΓ-mul i il a ions Le us ix an in ege ksuch ha 1 ≤k≤n. Fo each s= (s1,...,sk)∈Zk, we shall deno e Vs(D) = Tk i=1 V(i) si(D). The amily {Vs(D)}s∈Zkde ines a mul i- il a ion on he ing D. To simpli y we shall say ha V•(D) is a k– il a ion on Do e en, i no con usion is possible, a il a ion on D. Le us conside a a ional simplicial cone Γ in he i s quad an o he dual space (Qk)∗=HomQ(Qk,Q). We deno e by ˇ Γ he dual cone o Γ, i.e. ˇ Γ = {a∈Qk|γ(a)≥0,∀γ∈Γ}. We associa e wi h such a cone Γ he a ine a ie y, deno ed by SΓ, wi h coo dina e ing equal o C[ˇ Γ∩Zk]. We will deno e AΓ=C[ˇ Γ∩Zk] and A=C[Nk]. We deno e by L(Γ) he se o p imi i e elemen s in he 1-skele on o Γ. The mul i il a ion VΓon Dis de ined as ollows: Fo each s∈Zkwe de ine VΓ s(D) = X σ∈Zk|L(σ)≤L(s);∀L∈L(Γ) Vσ(D). No ice ha he sum is indexed by σ∈s−ˇ Γ and ha we ha e he inclusion Vs(D)⊂VΓ s(D). The amily VΓ •(D) is a mul i il a ion o he ing D, indexed by s∈Zk. This means ha : VΓ s(D).V Γ s′(D)⊂VΓ s+s′(D) and [VΓ s(D) = D. We may de ine, in a simila way o he case o one il a ion, he no ion o a ee mul i- il e ed module and ha o a good mul i il a ion o a ini ely gene a ed D-module M. Fo ha pu pose we chose a shi mul i ec o n= (n(1),··· , n( ))∈(Zk) called also a shi ma ix, wi h columns n(i)∈Zk, and a p esen a ion M=D No M. De ini ion 2.2 The mul i il e ed ee module associa ed wi h nis he module D endowed wi h he mul i il a ion indexed by s∈Zk, and de ined as: V[n]s(D ) = M i=1 Vs−n(i)(D) De ini ion 2.3 A good mul i il a ion o M, is a il a ion indexed by s∈Zk, o he ype Us(M) = π(V[n]s(D )) = V[n]s(D ) + N N o some p esen a ion π:D → M o M. In he ob ious sense hese mul i il a ions a e compa ible wi h he mul i il e ed s uc u e on he ing D. Rema k ha he i- h gene a o eio D is hen o mul ideg ee n(i)∈Zk. We may also endow Nwi h he induced il a ion Us(N) = V[n]s(D )∩ N, so ha we also ha e Us(M) = V[n]s(D ) Us(N). We may obse e ha he mul i il a ion on he ee module D is de ined as he in e sec ion o he V(i)– il a ion wi h espec o he ow ec o s o n,ni= (n(1) i,··· , n( ) i)∈Z , ha is V[n]s(D ) = V(1)[n1]s1(D )∩ · · · ∩ V(k)[nk]sk(D ), 3 bu he analogue o M, wi h espec o he good V(i)- il a ions as de ined in de ini ion 2.1 is no ue since he inclusion: Us(M) = V(1)[n1]s1(D )∩ · · · ∩ V(k)[nk]sk(D ) + N N⊂ k i=1 V(i)[ni]si(D ) + N N= k i=1 U(i) si(M) may be s ic . Fo each good il a ion he e is an associa ed Γ- il a ion compa ible wi h he mul i il a ion VΓ •(D) on D: Fo each s∈Zkle us conside UΓ s(M) = X σ∈Zk|L(σ)≤L(s);∀L∈L(Γ) Uσ(M). The mul i il a ion UΓ •(M) is a good il a ion wi h espec o VΓ •(D). The goodness means he e ha we can e i y, wi h he p esen a ion o Mas abo e, ha UΓ s(M) = π(V[n]Γ s(D )), is s ill he quo ien o a il a ion on he ee module D which is a di ec sum o con enien shi s o he il a ion VΓ •on he ing D. Mo e p ecisely he in ol ed il a ion V[n]Γ •is de ined by V[n]Γ s(D ) = X σ∈Zk|L(σ)≤L(s);∀L∈L(Γ) V[n]σ(D )( = M i=1 (VΓ s−n(i)(D)) and i emains o be ema ked ha : UΓ s(M) = X σ∈Zk|L(σ)≤L(s);∀L∈L(Γ) Uσ(M) = X σ∈Zk|L(σ)≤L(s);∀L∈L(Γ) V[n]σ(D ) + N N=V[n]Γ s(D ) + N N so ha he goodness o he il a ion U•(M) implies he goodness o UΓ •(M). 2.2 Rees ings and Rees modules. 2.2.1 De ini ion o Rees ings RV(D),RΓ(D)and o ela ed Rees modules The Rees ing associa ed wi h he V-mul i il a ion on Dis de ined by: RV(D) = M s∈Zk Vs(D)us whe e u= (u1,...,uk) a e a iables and he p oduc in he Rees ing is induced by he na u al p oduc in he Lau en polynomial ing D[u±1] = D[u1, u−1 1,...,uk, u−1 k]. By de ini ion he Rees ing RV(D) is a g aded C-algeb a wi h alues g oup Zk, whose homogeneous elemen s a e P us o P∈Vs(D) and s∈Zk. Simila ly gi en a D-module M=D /Nand a shi ma ix n, we de ine om he good il a ions V[n]s(D ), and Us(M), he Rees modules RV[n](D ) = ⊕s∈ZkV[n]s(D )us RU(M) = M s∈Zk Us(M)us. Bo h ha e a na u al s uc u e o mu ig aded le RV(D)-module and we shall conside RU(M) as a sub-g oup o M[u±1], he Lau en polynomials wi h coe icien s in M. As Nis a le submodule o D , we can also conside on N he induced V- il a ion: Vs(N) := N ∩ Vs[n](D ) 4 o each s∈Zkas de ined be o e (see 2.1). The abelian g oup RV(N) := M s∈Zk Vs(N)us is in ac an homogeneous submodule o RV[n](D ). The RV(D)–module RU(M) is na u ally isomo phic o he quo ien RV[n](D )/RV(N). De ini ion 2.4 We call he module RU(M) he Rees module associa ed wi h he il a ion U•(M). Le us now conside he il a ion VΓ •(D) as in 2.1. Then he Rees ing associa ed wi h his il a ion is de ined in a simila way as RΓ(D) = M s∈Zk VΓ s(D)us⊂ D[u±1] he p oduc being induced by he one o D[u±1]. I M=D Nis a ini ely gene a ed D-module and U•(M) is a good il a ion on Mwi h espec o V•(D) hen, o any cone Γ as in 2.1, he abelian g oup RΓ(M) = M s∈Zk UΓ s(M)us⊂ M[u±1] is a le g aded RΓ(D)-module. De ini ion 2.5 We call he module RΓ(M) he Rees module associa ed wi h he il a ion U•(M)and he cone Γ. The ac ha a il a ion is good in he sense o p e ious sec ion 2.1 is hen equi alen o he ac ha he Rees module o Mis ini ely gene a ed o e RV(D). Simila ly he goodness o UΓ •(M) is equi alen o he ac ha he Rees module RΓ(M) as de ined abo e, is ini ely gene a ed o e RΓ(D). Since RΓDis a Noe he ian ing, his implies in pa icula an A in ype p ope y: I M′⊂ M is a submodule o a D-module Mendowed wi h a good il a ion U•(M) hen he induced il a ion on M′, namely UΓ •(M)∩ M′is good. The ing RΓ(D) con ains as a sub ing he C–algeb a AΓ=C[ˇ Γ∩Zk]⊂C[u, u−1], all hese ings being included in D[u, u−1]. Theo em 2.6 The e is a an Ein (Qk)∗ +such ha o each cone Γ∈ E he Rees module RΓ(M)is la o e AΓ. We will p o e a mo e p ecise o m o his esul we in sec ion 4, see heo em 4.3, a e ha ing ecalled in sec ion 3 he no ion o an analy ic s anda d an. 2.2.2 The ibe a 0 Le us deno e by m he ideal o RV(D) gene a ed by (u1,...,uk). This is a wo sided ideal since i is gene a ed by cen al elemen s. The ibe a ze o o RV(D) ( esp. o RV[n](D )) is by de ini ion he quo ien ing ( esp. he quo ien module), RV(D) m( esp. RV[n](D ) m.RV[n](D )). Mo e gene ally he ibe a ze o o RU(M) = RV[n](D ) RV(N)is he quo ien RU(M) mRU(M) which is na u ally isomo phic o he quo ien RV[n](D ) RV(N) + m.RV[n](D ). 5 No ice ha he ibe a 0 o he module RU(M) can be ze o o a non-ze o RU(M), as shown in he ollowing example. Le us deno e by I he p incipal ideal o Dgene a ed by he di e en ial ope a o P= 1 + x2 1∂1and le us suppose k≥1. Then RV(I) is he p incipal ideal o RV(D) gene a ed by P u0 1= 1 + (x2 1∂1u−1 1)u1and he ibe a ze o o RV(D/I) is hen ze o since RV(I) + m.RV(D) con ains P−(x2 1∂1u−1 1)u1= 1, so ha i is equal o RV(D). Simila ly o any k-dimensional cone Γ he C–algeb a AΓ=C[ˇ Γ∩Zk] has a maximal ideal mΓ=C[ˇ Γ∩Zk {0}] because ˇ Γ is s ic ly con ex. Then we de ine he ibe a ze o o he Rees module RΓ(M) as RΓ(M) mΓRΓ(M).As is explained in [9], his ibe is a module o e he ing RΓ(D) mΓRΓ(D)≃g ΓD. 2.2.3 Desc ip ion o Rees ings RV(D)and RΓ(D) I is use ul o desc ibe he Rees ing RV=RV(D) ( esp. RΓ(D)) as sub ings o he ing o ela i e di e en ial ope a o s DCn×Ck/Ck( esp DCn×SΓ/SΓ). This is no hing bu an explici e sion o he in e p e a ion o a Rees ing as a ing o ela i e di e en ial ope a o s on he de o ma ion o Y×0Γ o i s no mal cone in Cn×SΓ,see [9]. We deno e by SΓ he algeb aic a ie y associa ed o he ing AΓ. Le us de ine i s A=C{X′U, X′′}[X′, U, ∆] = C{X1U1,...,XkUk, Xk+1,...,Xn}[X′, U, ∆] whe e U= (U1,...,Uk), X′= (X1,...,Xk), X′′ = (Xk+1,...,Xn), X= (X1, . . . , Xn), ∆ = (∆1,...,∆n) a e new a iables sa is ying he ollowing ela ions, o i= 1,...,n: ∆iXi=Xi∆i+ 1 he o he ela ions being i ial. The ing Ais g aded wi h Zkas a alues g oup (we will say ha Ais a Zk–g aded ing) , namely we ha e A=M s∈Zk As whe e Asis he se As={X α,β∈Nn;σ∈Nk αβσXα∆βUσ∈ A | αβσ ∈C{X′U, X′′}, σi+βi−αi=si;i= 1,...,k}. P oposi ion 2.7 The e is an isomo phism o Zk–g aded ings i=iV:RV(D)→ A de ined by: •i(uj) = Uj o j= 1,...,k. •i(xju−1 j) = Xj o j= 1, . . . , k and i(xj) = Xj o j=k+ 1,...,n. •i(∂juj) = ∆j o j= 1,...,k and i(∂j) = ∆j o j=k+ 1,...,n. 6 P oo .I is clea ha iis injec i e and ha , by he o mula i(xα∂βus) = Xα∆β(Πk i=1Usi+αi−βi i), i(Vs(D)us) = As o all s∈Zk. Le us gi e now he same desc ip ion o he ing RΓ(D). We shall do i in he only case o in e es o us, when he cone Γ is basic, which means ha i i is de ined by kindependen linea o ms {L1,...,Lk}gene a ing he la ice Zk. Fo he sake o simplici y le us suppose k=n. In his si ua ion we a e going o make a change o a iables in o de o w i e AΓas a polynomial ing. Fo any iwe can w i e Li= (ℓi1,...,ℓik) and we can suppose by a sui able o de ing o he linea o ms Li ha he de e minan o he ma ix L= (ℓij) equals 1. Le us w i e L′ o he in e se ma ix o L, hen he columns {C1,...,Ck}o L′ o m a basis o he dual cone ˇ Γ. The amily uC1,...,uCkgene a es he C–algeb a AΓ=C[ˇ Γ∩Zk]. He e we w i e uCj=uc1j 1···uckj kwhe e he cij’s a e he en ies o he column Cj. We will w i e Wi=UCi o i= 1,...,k. Simila ly, we ha e Uj=WCj(L) o j= 1,...,k whe e Cj(L) is he j− h column o he ma ix L. In sec ion 4 we will only use he column Cj=Cj(L′). Le us ema k he e ha i k < n hen we can s ill de ine W, he ela ionship be ween Wand Ubeing exac ly he same. P oposi ion 2.8 The ing RΓ(D)is isomo phic o he sub ing o A′=C{X, W}[∆], which con ains all polynomials and in which only con e gen powe se ies wi h espec o he amily o monomials XiUi=XiWCi(L) o i= 1,···k, and Xk+1,··· , Xn, in he gene al case k≤na e allowed. P oo .A monomial Xα∆βUs−β+αcan be w i en in e ms o X, ∆, W as Xα∆βWL(s−β+α).All ha emains o be done is o enume a e he monomials de i ed om he powe se ies a iables x1,··· , xn The ing AΓis iden i ied o he sub ing C[W](= C[W1,...,Wk]) ⊂ RΓ(D) and his inclusion is la . We shall p o e his s a emen in 4.2. The ibe o RΓ(D) a he o igin is by de ini ion RΓ(D)⊗C[W] C[W] (W)which is isomo phic o a Weyl algeb a, namely he Weyl algeb a C[X, ∆]. This Weyl algeb a is endowed wi h a Zk–g adua ion by weigh (Xi) = −ǫi and weigh (∆i) = ǫi o i= 1, . . . , k whe e ǫiis he ec o in Zkwhose j− h coo dina e is δij. 2.3 Mul i il a ions and Be ns ein-Sa o unc ional equa ions Le us de ail he unc ional equa ion p oblem aised in he in oduc ion. I has al eady been ema ked ha i gi es ise o a si ua ion whe e a mul i il a ion along ans e se hype su aces comes ou in a na u al way. These equa ions a e o he ype P(λ) λ=b(λ) λ1+1 1··· λk+1 k whe e λ= (λ1,··· , λk) is a k-uple o inde e mina es, and = ( 1,··· , k) is a k-uple o analy ic unc ions on X. They a e na u ally w i en in he module OXλ1,··· , λk,1 1··· k λ, endowed wi h a s uc u e o a DX[λ]- module, which can be ex ended in a way disco e ed by B. Malg ange o a DX×Ck-module s uc u e wi h an ac ion o he 2k a iables j, ∂ jsuch ha λj=−∂ j j.The module M ha we ha e hen o conside is he module gene a ed o e he ing DX×Ckby λ, wi h i s na u ally de ined mul i il a ion V•(DX×Ck)· λalong he hype su aces 1= 0,··· , k= 0. The p oo o Sabbah in [9] can be ske ched as ollows: le us de ine o any linea o m L∈(Qk)⋆wi h posi i e a ional coe icien s ℓj≥0, a il a ion o UL •(M), associa ed wi h he linea combina ion PℓjV(j)o he basic V- il a ions (see no a ions a 2). P ecisely we se (see also he no a ion 3.1 below): VL •(DX×Ck) = {P=X µ,ν∈Nk Pµν(x, ∂x) µ∂ν ,Xℓi(νi−µi)≤ •} UL •(M) = VL •(DX×Ck)· λ Then he exis ence o a unc ional equa ion, in which b(λ) is a p oduc o a ine o ms L(λ) + ccomes ou om he ollowing h ee s eps: 7 1) The e a e Be ns ein-Sa o polynomials bL(λ), ela i e o each L, wi h unc ional equa ions bL(λ) λ∈VL <0(DX×Ck)· λ. 2) The sa u a ed il a ion Us(M) = L∈(Qk)⋆ UL L(s)(M) can be de ined by using only a ini e numbe o ixed linea o ms. 3) The sa u a ed il a ion is good which is equi alen o he exis ence o a k-uple o in ege s κ, such ha ∀s∈Nk, Us(M)⊂Us(M)⊂Us+κ(M). I is s ep 2) which, in he p oo in [9], makes an essen ial use o he no ion o an adap ed an, whose exis ence is p o ed in his pape , see heo em 2.6. In Bahloul’s pape [4], his s ep is made by cons uc i e me hods which do no use he la ness p ope y. 3 The analy ic s anda d an In his sec ion we will summa ize he main esul s o [3]. Since hese esul s a e only gi en o a module D/I o e Dwhich is a quo ien by an ideal we will hen show b ie ly how o adap hem o a module o he ype D N. Le us ema k ha he an used in his pape is ob ained by a es ic ion o he an in [3] o a subse o linea o ms o he se Ude ined below, namely he linea combina ions o he il a ions V(i). Le Ube he se o linea o ms Λ : R2n→R, Λ(α, β) = Pn i=1 eiαi+Pn i=1 iβiwi h ei+ i≥0 and ei≤0 o i= 1,...,n. I P=X αβ pαβxα∂β is an elemen in Dwe de ine o dΛ(P) – he Λ-o de o P– o be he maximal alue o Λ(α, β) o α, β such ha pαβ 6= 0. The Λ- il a ion FΛ,•(D) is de ined by FΛ,ℓ =FΛ,ℓ(D) = {P∈ D | o dΛ(P)≤ℓ} o any ℓ∈R. We will w i e FΛ,<ℓ(D) := {P∈ D | o dΛ(P)< ℓ}. I ei= 0, i= 1 o i= 1,...,n he co esponding Λ– il a ion is no hing bu he usual il a ion by he o de o di e en ial ope a o s. We shall deno e i simply F•(D). Le us ecall ha we ha e ixed k≤n. Fo each linea o m L∈(Qk)∗ +(i.e. he coe icien s o La e non-nega i e) we deno e by e L he linea o m on R2nde ined by e L(α, β) = L(β)−L(α) whe e L(α1,...,αn) = L(α1,...,αk). No a ion 3.1 We de ine he il a ion VL •on D, indexed by he se o alues L(Zk), as: VL ℓ(D) = Fe L,ℓ(D) = {P∈ D | o de L(P)≤ℓ} Le us ix L∈(Qk)∗ +. The g aded ing associa ed o he il a ion VL=Fe Lon D, is by de ini ion g e L(D) = M ℓ∈e L(Z2n) Fe L,ℓ Fe L,<ℓ . I no con usion is possible we shall w i e simply o dLand g L(D) ins ead o o de Land g e L(D). The g aded ing g L(D) is a ing o di e en ial ope a o s and i s s uc u e is he ollowing: suppose he coe icien s o he o m La e (e1,...,ek) and suppose we also ha e o de ed he a iables o ha e ei>0 o 1≤i≤ℓ o some ℓ≤k. 8 Then he g aded ing g L(D) is isomo phic o he ing C{xℓ+1,...,xn}[x1,...,xℓ, ∂1,...,∂n]. In his ing he g adua ion is induced by he weigh s weigh (xi) = −ei,weigh (∂i) = ei o 1 ≤i≤ℓand weigh (xj) = weigh (∂j) = 0 o he wise. The e is only a ini e numbe o ypes o hese ings, one o each pa i ion o {1,...,k}in o wo se s. No a ion 3.2 Fo each P∈ D and o each d∈L(Zk)wi h o dL(P)≤dwe deno e by σL d(P)– he L–symbol o Po o de d– he class o Pin Fe L,d/Fe L,<d. The p incipal symbol o Pis by de ini ion σL(P) = σL d(P)i d= o dL(P). Fo P, Q ∈ D we ha e σL(PQ) = σL(P)σL(Q). Fo any le ideal Iin Dwe deno e by g L(I) he g aded ideal o g L(D)gene a ed by he se {σL(P)|P∈I}. We se σL(0) = 0. We deno e by D[ ] he C-algeb a O[∂, ] = C{x1,...,xn}[∂1,...,∂n, ] wi h ela ions ( being a new a iable) •[ , a] = [ , ∂i] = [a, b] = [∂i, ∂j] = 0, •[∂i, a] = ∂a ∂xi , o a, b ∈ O and i= 1,...,n. The ing D[ ] is isomo phic o he Rees ing associa ed wi h he o de il a ion F•on D. Since his Rees ing is by de ini ion RF(D) = M ℓ∈Z Fℓ(D) ℓ⊂ D[ ] = D ⊗CC[ ] o a new a iable , we can de ine an isomo phism o g aded ings ι:D[ ]→ RF(D) by ι(a) = a=a 0, ι(∂j) = ∂j , and ι( ) = 1 = . The na u al g aded s uc u e o RF(D) can be ansla ed on D[ ]. An homogeneous elemen o deg ee d∈Z in D[ ] is no hing bu an exp ession X ℓ+|β|=d aℓ β∂β ℓ o some aℓ β ∈ O. Fo P=Pβpβ(x)∂β∈ D, he elemen P o d(P)∈ RF(D) is called he homogeniza ion o P, whe e o d(P) is he usual o de o P. I is use ul o see P o d(P)as an elemen o D[ ]. The homogeniza ion o Pis hen deno ed by h(P) and we ha e h(P) = X β pβ(x)∂β d−|β| o d= o d(P) he usual o de o P. Fo each linea o m L∈(Qk)∗ +we can de ine in a na u al way a il a ion VL •(D[ ]) on D[ ], he L–o de –deno ed o dL(R)– o an elemen R=X ℓαβ ℓαβxα∂β ℓ being he maximal alue o L(β)−L(α) o ℓαβ 6= 0. The associa ed g aded ing is deno ed by g L(D[ ]). We ha e a na u al ing isomo phism om g L(D[ ]) on o g L(D)[ ], whe e wi h he no a ions as be o e we ha e [∂i, xj] = δij ,δij being he K onecke symbol. Fo each le ideal Iin Dwe deno e by h(I) he le homogeneous ideal o D[ ] gene a ed by {h(P)|P∈I}. As in he case o D, we deno e by g L(h(I)) he homogeneous ideal –in g L(D[ ])– gene a ed by he se o p incipal symbols σL(G) o he elemen s Gin h(I). The main esul o [3] is he ollowing Theo em 3.3 Le Ibe a non-ze o le ideal o Dand le h(I)be he associa ed homogenized ideal in D[ ]. Then he e exis s a pa i ion Eo Uin o con ex a ional polyhed al cones such ha o any Γ∈ E he ideals g Λ(h(I)) and g Λ(I)do no depend on Λ∈Γ. 9