DISCONTINUOUS GALERKIN METHODS FOR THE ONE-DIMENSIONAL VLASOV-POISSON SYSTEM
|
|
- Jason Patrick
- 5 years ago
- Views:
Transcription
1 DSCONTNUOUS GALERKN METHODS FOR THE ONE-DMENSONAL VLASOV-POSSON SYSTEM BLANCA AYUSO, J. A. CARRLLO, AND CH-WANG SHU Abstract. We construct a new famly of sem-dscrete numercal scemes for te approxmaton of te one-dmensonal perodc Vlasov-Posson system. Te metods are based on te couplng of dscontnuous Galerkn approxmaton to te Vlasov equaton and several fnte element (conformng, non-conformng and mxed approxmatons for te Posson problem. We sow optmal error estmates for all te proposed metods n te case of smoot compactly supported ntal data. Te ssue of energy conservaton s also analyzed for some of te metods. Key words. conservaton Vlasov-Posson system; Dscontnuous Galerkn; mxed-fnte elements; energy AMS subject classfcatons. 65N3, 65M12, 65M15, 82D1. 1. ntroducton. Te Vlasov-Posson system s one of te basc and smplest models n te mesoscopc descrpton of large ensembles of nteractng partcles. n one-space dmenson and n dmensonless varables, te Vlasov equaton reads f t + v f x Φ f x v = (x, v, t Ω x R [,, (1.1 were te electrostatc feld, Φ x (x, t, s derved from a potental Φ(x, t tat satsfes: Φ xx = ρ(x, t 1 (x, t Ω x [,, (1.2 wt ρ(x, t beng te carge densty wc s defned by ρ(x, t = f(x, v, t dv for all (x, t Ω x [,. (1.3 R Te above system descrbes te evoluton of a collsonless plasma of carged partcles (electrons and ons n te case were te only nteracton (between partcles consdered relevant s te mean-feld force created troug electrostatc effects, ence neglectng te electromagnetc effects. f(x, v, t s te electron dstrbuton, wc s a non-negatve functon dependng on te poston: x Ω x R; te velocty: v R, and te tme: t R, wt Ω x denotng te spatal doman were te plasma s confned. As ons are muc eaver tan electrons, t s assumed tat ter dstrbuton s unform and snce te plasma sould be neutral, one as ρ(x, t dx = f(x, v, t dv dx = 1 for all t [,. (1.4 Ω x Ω x R We refer to te surveys [42, 12, 34 for good account on te state of te art n te matematcal analyss and propertes of te solutons of te Caucy problem for te Vlasov-Posson system. Departamento de Matemátcas, Unversdad Autónoma de Madrd, Madrd 2849, Span. E-mal: blanca.ayuso@uam.es. CREA and Departament de Matemàtques, Unverstat Autònoma de Barcelona, E-8193 Bellaterra, Span. E-mal: carrllo@mat.uab.es. Dvson of Appled Matematcs, Brown Unversty, Provdence, R 2912, USA. E-mal: su@dam.brown.edu. 1
2 2 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Many efforts ave been dedcated to te numercal approxmaton of te Vlasov- Posson system wt eter probablstc or determnstc solvers. Snce te begnnngs of numercal plasma smulatons n te 6 s, partcle metods [11 ave been often preferred because of ter lower computatonal complexty. For tese metods, te moton of te plasma s approxmated by a fnte number of macro-partcles n te pyscal space tat follow backward te caracterstcs of te Vlasov equaton. Several works ave also analyzed ter convergence n one [3, 59 and ger dmensons [41. However, a well known drawback of tese metods s ter nerent numercal nose wc makes t dffcult to obtan an accurate descrpton of te dstrbuton functon n te pase space for many applcatons. To overcome ts lack of precson, Euleran solvers, metods dscretzng te Vlasov equaton on a mes of te pase space, ave been also consdered. Ter desgn as been explored by many autors and wt many dfferent tecnques: fnte volumes [36, 37; Fourer-Fourer transform [45; fnte elements [6, 61, splttng scemes [33, 17; and sem-lagrangan metods [39, 31, 16. All tese metods present dfferent pros and cons and we refer to [38 and te references teren for a dscusson. Fnte volumes are a smple and nexpensve opton, but n general, are low order. Fourer-Fourer transform scemes suffer from Gbbs penomena f oter tan perodc boundary condtons are mposed. Sem-lagrangan scemes can aceve g order allowng also for tme ntegraton wt larger tme steps. However, tey requre g order nterpolaton to compute te orgn of te caracterstcs, wc n turn destroys te local caracter of te reconstructon. Standard fnte element metods suffer from numercal oscllatons wen approxmatng te Vlasov equaton. n contrast, dscontnuous Galerkn (DG fnte elements are partcularly well suted for yperbolc problems and ter applcaton to non-lnear conservaton laws as already sown ter usefulness [26, 25, 28. Based on a totally dscontnuous fnte element spaces, DG metods are extremely versatle and ave numerous attractve features: local conservaton propertes; can easly andle rregularly refned meses and varable approxmaton degrees (p-adaptvty, weak approxmaton of boundary condtons and bult-n parallelsm wc permts coarse-gran parallelzaton. n addton, DG mass matrces are blockdagonal and can be nverted at a very low computatonal cost, gvng rse to very effcent tme-steppng algortms n te context of tme-dependent problems, as t s te case ere. Poneerng researc on dscontnuous Galerkn metods was pursued n [52, 48, 35, 57, 3. We refer to [24, 4 for a detaled storcal overvew and for more recent developments to [55, 5, 14, 5 and references teren. However, altoug DG metods can deal robustly wt partal dfferental equatons of almost any knd, ter applcaton n te realm of numercal approxmaton of knetc models as been consdered only very recently. n [2 and [9 te autors study, respectvely, te use of DG for te Boltzmann-Posson system n semconductors and for water-bag approxmatons of te Vlasov-Posson system. n [4, an L 1 -analyss s carred out for a smplfed lnear Vlasov-Boltzmann equaton wt a gven confnng force feld. Despte te fact tat te numercal performance of all tese Euleran solvers as been extensvely studed, to our knowledge, te ssue of ter convergence and error analyss for te Vlasov-Posson system, as not been tackled tll very recently, and only for te one-dmensonal perodc case. Te convergence and error analyss for a low order fnte volume sceme s contaned n [37. More recently, sem-lagrangan scemes ave been analyzed; of frst order n [7 and g order s consdered n [8, 1. n tese works te autors ave also proved a-pror error bounds n dfferent norms for bot te dstrbuton functon and te electrostatc feld. We also menton tat for
3 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 3 oter knetc models, fnte dfferences [53 and spectral metods [47, 46 ave been also consdered and analyzed. Te present paper s concerned wt te desgn and analyss of dscontnuous Galerkn approxmaton for te one-dmensonal perodc Vlasov-Posson system. We ntroduce a wole famly of Euleran scemes, based on te combnaton of DG approxmaton to te Vlasov equaton wt varous dfferent fnte element (conformng and nonconformng approxmatons to te electrostatc feld. Te frst one s a drect conformng approxmaton obtaned by takng advantage of te explct ntegraton of te Posson equaton n one dmenson. Suc approxmaton s equvalent to wat most autors, f not all, ave usually consdered for ts system. However, n spte of ts smplcty, t mgt not be te most approprate sceme n vew of te possble extenson/adaptaton of te numercal sceme to ger dmensons and to more complex knetc models. For ts reason, n te present paper we also examne a dfferent approac: snce te couplng n te Vlasov-Posson system s troug te electrostatc feld, te man nterest n te Posson problem s te approxmaton to Φ x rater tan to Φ, and terefore mxed fnte element metods seem to be te rgt coce. We explore Ravart-Tomas and several mxed DG approxmatons for te Posson problem. We also deal wt te convergence and error analyss for te proposed DG metods for te case of smoot compactly supported solutons. We derve optmal error bounds n te L 2 -norm for bot te dstrbuton functon and te electrostatc feld, for g order metods, namely k 1, k beng te polynomal degree of te DG approxmaton for te dstrbuton functon. Te analyss for pecewse constant approxmaton (k = s dfferent and wll be carred out somewere else. Altoug Vlasov equaton mgt be seen as a smple transport equaton, ts couplng wt Posson, brngs nto play n suc equaton, a non-lnear (quadratc and non-local term. Ts generates some dffcultes n te error analyss, precludng a stragtforward extenson of oter works. A key ngredent s te constructon of some projecton operators, nspred n tose ntroduced n [48, 54, 23, 62, but specally desgned for te Vlasov-Posson system. Tese specal projectons allow for avodng te loss of alf order, typcal of te standard error analyses of fnte element metods for yperbolc problems. We ave focused on sem-dscrete scemes; dscusson on sutable tme ntegrators and desgn and analyss of fully dscretzed scemes s outsde te scope of ts paper and wll be te subject of future researc. Fnally, we ws to note tat wle developng te metods, we ave taken specal care n ensurng tat pyscal propertes of te contnuous system are preserved. Te DG approxmaton for te Vlasov equaton ensures n an easy way tat te total carge of te system s preserved (1.4. We also dscuss te conservaton of te total energy for te proposed scemes. n partcular, we propose a full DG metod (DG for te Vlasov equaton and a partcular local dscontnuous Galerkn (LDG for te Posson problem, tat preserves te total dscrete energy of te system. To te best of our knowledge ts s te frst sceme proposed n te lterature for wc energy conservaton can be sown. Our proof owever requres a tecncal assumpton on te polynomal degree for te DG metods, namely k 2. Weter ts restrcton s really necessary or not, wll be te subject of future researc. For many oter full DG scemes presented n te paper, we provde a bound on te energy dsspated by te system. Extenson to ger dmensons, numercal valdaton of te results presented ere and numercal performance of te presented numercal scemes n callengng
4 4 B. AYUSO, J. A. CARRLLO AND C.-W. SHU questons suc as te Landau dampng of Langmur waves [63 or te Raman scatterng nstablty [9 wll be carred out somewere else. Te outlne of te paper s as follows. n secton 2 we descrbe te man propertes of te contnuous problem, we ntroduce te notatons and revse some basc results we need for te descrpton and analyss of te numercal metods. n secton 3 we present te numercal metods proposed to approxmate te one dmensonal perodc Vlasov-Posson system. Te error analyss for te presented metods s detaled n secton 4. We dscuss te energy conservaton propertes of te scemes n secton 5. Te paper s completed wt two appendces, Appendx A and Appendx B, contanng some proofs of tecncal and auxlary lemmas used n te convergence analyss. 2. Prelmnares, notaton and auxlary results. Trougout ts paper, we use te standard notaton for Sobolev spaces [1. For a bounded doman B R 2, we denote by H m (B te L 2 -Sobolev space of order m and by m,b and m,b te usual Sobolev norm and semnorm, respectvely. For m =, we wrte L 2 (B nstead of H (B. We sall denote by H m (B/R te quotent space consstng of equvalence classes of elements of H m (B dfferng by constants; for m = t s denoted by L 2 (B/R. We sall ndcate by L 2 (B te space of L 2 (B functons avng zero average over B. Ts notaton wll also be used for perodc Sobolev spaces wtout any oter explct reference to perodcty to avod cumbersome notatons Contnuous problem: Te 1D perodc Vlasov-Posson system. n te rest of te paper we take Ω x = [, 1 n (1.1-(1.2-(1.3-(1.4. Let f denote a gven ntal dstrbuton f(x, v, = f (x, v n (x, v [, 1 R. We mpose perodc boundary condtons on x for te transport equaton (1.1, f(, v, t = f(1, v, t for all (v, t R [,. and also for te Posson equaton (1.2, Φ(, t = Φ(1, t and Φ x (, t = Φ x (1, t for all t [,. (2.1 Notce tat (1.4 s coerent wt te 1-perodcty of Φ x. Let us also empasze tat te correct way of ncludng perodc boundary condtons s to assume tat f and Φ are te restrcton to [, 1 of perodc functons defned n R n te rgt spaces. To guarantee te unqueness of te soluton Φ (oterwse t s determned only up to a constant, we fx te value of Φ at a pont. We set Φ(, t = for all t [,. (2.2 However, notce tat snce te Posson equaton (1.2 s one-dmensonal t could be drectly ntegrated. More precsely, by usng twce te Fundamental Teorem of Calculus, t follows tat Φ s defned for all t [, as Φ(x, t = D + C E x + x2 2 x s ρ(z, t dz ds x [, 1, (2.3 were D and C E are ntegraton constants determned from (2.2 and (2.1; D =, C E = 1 z ρ(s, t ds dz 1 2 t [, T. (2.4
5 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 5 Denotng ten by E(x, t = Φ x (x, t, and dfferentatng (2.3 wt respect to x, we fnd E(x, t = Φ x (x, t = C E + x x ρ(s, t ds x [, 1, (2.5 wt C E defned n (2.4. Trougout te paper, E wll be referred to as te electrostatc feld. Altoug te pyscal one s ndeed E, we sall use ts abuse n te notaton to follow te standard notaton for te Posson solvers n te dscontnuous Galerkn communty. Observe tat (2.1 mples tat te electrostatc feld as zero average n agreement wt te carge neutralty. n order to perform our error analyss we restrct our attenton to smoot compactly supported solutons f n a fxed tme nterval [, T for all T >. Gven a dstrbuton functon f(x, v, t, we wll denote by Q(t = 1 + sup{ v : x [, 1 and τ [, t suc tat f(x, v, τ }, for all t [, as a measure of te support of te dstrbuton functon. followng result s essentally contaned n [29, 58, 42. Teorem 2.1 (Well-posedness of te contnuous 1DVP. Gven f C 1 (R x R x, 1-perodc n x and compactly supported n v, Q( Q wt Q >. Ten te perodc Vlasov-Posson system (1.1-(1.2 as a unque classcal soluton (f, E, f C 1 (, T ; C 1 (R x R v and E C 1 (, T ; C 1 (R x tat s 1-perodc n x for all tme t n [, T for all T >, suc tat: Regularty: f n addton f C m (R x R x, m 2, ten, te dstrbuton functon f belongs to C m (, T ; C m (R x R v and te force feld E C m (, T ; C m (R x. Control of Support: Tere exsts a constant C dependng on Q and f suc tat Q(T CT for all T >. n te rest of ts work, we wll assume tat te ntal data f satsfes te ypoteses n Teorem 2.1, and tus, te unque classcal soluton to te perodc Vlasov-Posson system (1.1-(1.2 satsfes tat tere exsts L > dependng on f, T and Q suc tat supp( f(t Ω for all t [, T, were we ave defned Ω = J, wt = [, 1 and J = [ L, L. Te Vlasov transport equaton (1.1 s regarded as a transport equaton n Ω T := Ω [, T. Takng nto account te boundary condtons, te weak formulaton of te contnuous problem (1.1 reads: fnd (f, E suc tat f t φ dx dv vfφ x dx dv + Ef φ v dx dv = φ C (Ω. (2.6 Ω Ω Ω t s well known [42, 12, 34 tat te contnuous soluton of (1.1-(1.2 satsfes four mportant propertes: Postvty: f(t, x, v, for all (x, v, t Ω T. Carge conservaton: as gven n (1.4. L p -conservaton: Te f(t Lp (Ω = f Lp (Ω 1 p, t [, T. (2.7 Conservaton of te total Energy: ( d v 2 f(x, v, t dx dv + dt Ω E(x, t 2 dx =. (2.8
6 6 B. AYUSO, J. A. CARRLLO AND C.-W. SHU n dervng numercal metods for (1.1-(1.2, we wll try to ensure tat te resultng scemes wll be able to produce approxmate solutons, enjoyng some of tese propertes. As usual wt g-order scemes for yperbolc problems, we cannot expect to preserve postvty of te sceme. However, we wll be able to conserve te total energy for a partcular metod, see secton Dscontnuous Galerkn approxmaton: Basc notatons. Let {T } be a famly of parttons of our computatonal/pyscal doman Ω = J = [, 1 [ L, L, wc we assume to be regular [21 and made of rectangles. Eac cartesan mes T s defned as T := {T j = J j, 1 N x, 1 j N v } were = [x 1/2, x +1/2 = 1,..., N x ; J j = [v j 1/2, v j+1/2 j = 1,..., N v, and te mes szes x and v relatve to te partton are defned as < x = max 1 N x x := x +1/2 x 1/2, < v = max 1 j N v v j := v +1/2 v 1/2, were x and v j are te cell lengts of and J j, respectvely. Te mes sze of te partton s defned as = max ( x, v. For smplcty n te exposton we also assume tat v = corresponds to a node, v j 1/2 = for some j, of te partton of [ L, L. Te set of all vertcal edges s denoted by Γ x, and respectvely, we wll refer to Γ v as te set of all orzontal edges; Γ x := {x 1/2 } J j, Γ v := {v j 1/2 }, Γ = Γ x Γ v.,j,j By { } we sall denote te famly of parttons of te nterval ; := { : 1 N x } γ x := {x 1/2 }. Next, for k, we defne te dscontnuous fnte element spaces V k and Zk and a conformng fnte element space, W k+1, V k = { ψ L 2 ( : ψ P k (, x = 1,... N x, }, Z k := { z L 2 (Ω : z Q k (T j, (x, v T j = J j,, j }, W k+1 = { χ C ( : χ P k+1 (, x = 1,... N x, } L 2 (/R, were P k ( s te space of polynomals (n one dmenson of degree up to k, and Q k (T j s te space of polynomals of degree at most k n eac varable. Trace Operators: We denote by (ϕ + +1/2,v and (ϕ +1/2,v te values of ϕ at (x +1/2, v from te rgt cell +1 J j and from te left cell J j, respectvely; (ϕ ± +1/2,v = lm ε ϕ (x +1/2 ± ε, v, for all (x, v J or n sort-and notaton (ϕ ± x,j+1/2 = lm ε ϕ (x, v j+1/2 ± ε, (ϕ ± +1/2,v = ϕ (x ± +1/2, v, (ϕ ± x,j+1/2 = ϕ (x, v ± j+1/2, (2.9 for all (x, v J j. Te jump [ and average { } trace operators of ϕ at (x +1/2, v, v J j are defned by [ ϕ +1/2,v := (ϕ + +1/2,v (ϕ +1/2,v ϕ Z k, {ϕ } +1/2,v := 1 [ (ϕ /2,v + (ϕ +1/2,v ϕ Z k. (2.1
7 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM Tecncal tools. We start by defnng te space H m (T := {ϕ L 2 (Ω : ϕ Tj H m (T j T j T } m. n our error analyss, snce we consder a non-conformng approxmaton, we sall employ te followng semnorm and norms, ϕ 2 1, =,j ϕ 2 1,T j ϕ 2 m,t :=,j ϕ 2 m,t j ϕ H m (T, m ϕ,,t = sup T j T ϕ,,tj ϕ p L p (T :=,j ϕ p L p (T j ϕ L p (T, for all 1 p <. We also ntroduce te followng norms over te skeleton of te fnte element partton, ϕ 2,Γ x := (ϕ +1/2,v 2 dv, ϕ 2,Γ v = (ϕ x,j+1/2 2 dx ϕ H 1 (T.,j J j,j Ten, we defne ϕ 2,Γ = ϕ 2,Γ x + ϕ 2,Γ v. We notce tat all te above defntons apply also for te partton wt te obvous canges n te notaton. Projecton operators: For k, we denote by P k : L 2 ( V k te standard L2 - projecton onto te fnte element space V k defned locally,.e., for eac 1 N x, ( P k (w w q dx = q P k (. (2.11 Ts projecton s stable n L p ( for all p [32,.e., P k (w L p ( C w L p ( w L p (, 1 p. (2.12 We next ntroduce two more refned projectons (see [54, wc we denote by π ±, tat can be defned only for more regular functons, say w H 1/2+ɛ ( for all. Te projectons π + (w and π (w are te unque polynomals of degree at most k 1, tat satsfy for eac 1 N x ( π ± (w w q dx = q P k 1 (, (2.13 togeter wt te matcng condtons; π + (w(x + 1/2 = w(x+ 1/2 ; π (w(x +1/2 = w(x +1/2. (2.14 Provded w enjoys enoug regularty, say w H k+1 (, te followng error estmates can be easly sown for all tese projectons: } w P k (w, C k+1 w w π ± k+1, w H k+1 (. (2.15 (w, were C s a constant dependng only on te sape-regularty of te mes and te polynomal degree [21, 54. For te standard L 2 -projecton we wll also need estmates n te L -norm [56, w P k (w,, C k+1 w k+1,,. (2.16
8 8 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Let k and let P : L 2 (Ω Z k be te standard L2 -projecton (n te twodmensonal case defned by P (w = (Px k Pv k (w;.e., for all and j, (P (w(x, v w(x, v ϕ (x, v dv dx = ϕ P k ( P k (J j. (2.17 J j From ts defnton, ts L 2 -stablty follows mmedately, but t can be sown to be stable n L p for all p [32, P (w L p (T C w L p (Ω w L p (Ω, 1 p. ( Te suggested numercal metods. n ts secton we formulate te numercal scemes we propose to approxmate te Vlasov-Posson system. Te frst one s a sceme were te DG approxmaton for te transport equaton s coupled wt a smple conformng approxmaton of ger degree for te electrostatc feld. Te second sceme results by combnng mxed fnte element approxmaton for te Posson problem togeter wt DG approxmaton to te transport equaton. Te last approac s based on fully DG approxmaton for bot varables, te electron dstrbuton f and te electrostatc feld. Due to te specal structure of te transport equaton: v s ndependent of x and E s ndependent of v; for all metods te DG approxmaton for te electron dstrbuton functon s done exactly n te same way. Terefore we start by ntroducng te DG metod for te transport equaton (1.1, and n wat follows, we denote by E te restrcton to of te fnte element approxmaton E to be defned later on. Let f ( = P (f be te approxmaton to te ntal data. Te numercal metod reads: fnd (E, f : [, T (W, Z k suc tat N x N v Bj (E ; f, ϕ = ϕ Z k, (3.1 =1 j=1 were te blnear form B j (E ; f, ϕ s defned for eac, j and ϕ Z k as: B j (E ; f, ϕ = T j + f t ϕ ϕ dv dx vf dv dx + T j x [ ((vf ϕ +1/2,v ((vf ϕ + 1/2,v J j [ ( ( ( E f ϕ ( E f ϕ + x,j+1/2 E ϕ f dv dx T j v dv (3.2 x,j 1/2 dx, were we ave used te sort and notaton gven n (2.9. Notce tat te expresson B j (E ; f, ϕ s n fact a blnear form. E s used only to empasse te nonlnear dependence on t. Here, te boundary terms are te so-called numercal fluxes, wc are notng but te approxmaton of te functons vf and Ef at te vertcal and orzontal boundares Γ x and Γ v, respectvely. By specfyng tem, te DG metod s completely determned. Te desgn of tese numercal fluxes s te key ssue to ensure te stablty of te numercal sceme. We consder te followng upwnd coce: vf = { v f f v, v f + f v <, Ê f = { E f + f E, E f f E <. (3.3
9 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 9 We defne te numercal fluxes at te boundary Ω by ( vf 1/2,v = ( vf Nx+1/2,v, (Ê f x,1/2 = (Ê f x,nv+1/2 =, (x, v J, so tat te perodcty n x and te compactness n v are reflected. Te dscrete densty, denoted by ρ (x, t, s gven by ρ (x, t = f (x, v, t dv j J j x, t [, T. (3.4 Note tat from te defntons (1.3 and (3.4 of ρ and ρ, respectvely, and usng Caucy-Scwartz s nequalty t s stragtforward to see tat ρ(t ρ (t 2, 2L f (t f(t 2,T t [, T. (3.5 One of te nce features of te DG approxmaton for te transport s tat carge conservaton s ensured by constructon, as te followng result sows: Lemma 3.1. Partcle or Mass Conservaton: Let k and let f C 1 ([, T ; Z k be te DG approxmaton to f, satsfyng (3.1-(3.2. Ten, f (t dv dx = f ( dv dx = f dv dx = 1 t [, T. (3.6,j T j,j T j,j T j Proof. Note tat snce f ( = P (f, from te defnton of te L 2 -projecton (2.17 (wt ϕ = 1 togeter wt (1.4 we ave f ( dv dx = P (f dv dx = f dv dx = 1. (3.7,j T j,j T j,j T j We now fx an arbtrary T j and take n (3.2 te test functon ϕ = 1 n T j ; ϕ = elsewere. Notng tat suc a test functon verfes (ϕ +1/2,v = (ϕ + 1/2,v = 1, we ave B j (E : f, 1 = d f dv dx + [(vf dt +1/2,v (vf 1/2,v dv T j J j [(E f x,j+1/2 (E f x,j 1/2 dx. Moreover, note tat snce te coce of T j was done arbtrarly, te dentty above olds true for all, j. By summng t over all and j, te flux terms telescope and tere s no boundary term left because of te perodc (for and compactly supported (for j boundary condtons. Hence, takng nto account (3.1 we ave, = B j (E ; f, 1 = d f dv dx =, dt,j,j T j and so ntegraton n tme togeter wt (3.7 lead to (3.6. We next deal wt te approxmaton to te electrostatc feld E(x, t = Φ x (x, t. Te dscrete Posson problem reads, (Φ xx = 1 ρ x [, 1, Φ (1, t = Φ (, t. (3.8
10 1 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Te well posedness of te above dscrete problem s guaranteed by (3.6 from Lemma 3.1 wc n partcular mples (Φ x (1, t = (Φ x (, t. (3.9 To ensure te unqueness of te soluton we set Φ (, t =. To get te soluton of te dscrete Posson problem at least two possble approaces arse: Drect ntegraton of te dscrete Posson problem (3.8, approxmaton of (1.2 wt some mxed fnte element metod; possbly dscontnuous. We next consder n detal tese approaces Conformng approxmaton to te electrostatc potental. Reasonng as n secton 2.1, drect ntegraton of te dscrete Posson problem (3.8 togeter wt Φ (, t = gves Φ (x, t = C E x + x2 2 x s ρ (z, tdzds x [, 1, (3.1 were C E s determned from te boundary condtons n (3.8, C E = 1 z ρ (s, t ds dz 1 2 Ten, dfferentaton w.r.t x n (3.1 leads to t [, T. (3.11 x E (x, t = CE + x ρ (s, t ds x [, 1. (3.12 Observe tat snce ρ V k, E turns out to be a contnuous polynomal of degree k + 1; so E s conformng. ts restrcton to s gven by x (x, t = E 1(x 1/2, t + (x x 1/2 f (s, χ, tdχds x, (3.13 E x 1/2 and E (x, t = for all x [x 1/2, x +1/2. Te boundary condton (3.9 reads J E (x 1/2, t = E Nx (x N x+1/2, t t [, T. (3.14 To sow tat E ndeed belongs to W k+1 we ave to verfy tat t as zero average. From (3.11 t follows stragtforwardly E (x dx = E (x 1/2, t + x 2 +1/2 x2 1/2 2 x x 1/2 ρ (x dx =. Fnally, we state a Lemma tat relates te error commtted n te approxmaton to E, wt te error n accumulated n te approxmaton to f. Ts result wll be used n our subsequent analyss and ts proof s gven n Appendx A. Lemma 3.2. Let k and let (E, f C ([, T ; W k+1 C 1 ([, T ; Z k be te conformng-dg approxmaton to te soluton of Vlasov-Posson system (E, f, soluton of (3.1 (3.2 (3.12. Ten, E(t E (t, C 1 f(t f (t,t t [, T, (3.15
11 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 11 were 2L =meas(j and C 1 = (4L(1 + x 1/2. Furtermore, f te force feld E C ([, T ; H 1 (, te followng estmates also old for all t [, T, and E(t E (t,, C 2 f(t f (t,t wt C 2 = ((2L 1/2 + C 1, (3.16 E (t,, C 2 f(t f (t,t + E(t 1,. ( Mxed fnte element approxmaton for te Posson problem. We rewrte problem (3.8 as a frst order system: E = Φ x x [, 1; E x = ρ 1 x [, 1 (3.18 wt boundary condton Φ (, t = Φ (1, t =. n ts secton, we consder a mxed approxmaton to (3.18, wt te one-dmensonal verson of Ravart-Tomas elements, RT k k [51, 15. n 1D te mxed fnte element spaces turn out to be te (W k+1, V k-fnte element spaces. Note tat n partcular, d k+1 dx (W = V k. For k te sceme reads: fnd (E, Φ W k+1 V k suc tat E z dx + Φ z x dx = z W k+1, (3.19 (E x p dx = (ρ 1p dx p V k. (3.2 We refer to [51, 15 for te stablty and error analyss of te metod for lnear second order problems (see also [6 for te 1D-verson of te sceme n te lowest order case k =. However, n our case, te Posson problem s non lnear snce te source term n (1.2 depends on te soluton troug ρ. Terefore n te error analyss a consstency error appears. We ave te followng result, wose proof can be found n Appendx A. Lemma 3.3. Let k and let (E, Φ C ([, T ; W k+1 V k be te RT k approxmaton to te Posson problem (3.18 and let E C ([, T ; H k+1 ( Ten, te followng estmates old for all t [, T : E(t E (t, + E(t E (t 1, C k+1 E(t k+1, + 2L f(t f (t,t, E(t E (t,, C k+1 E(t k+1, + (2L 1/2 f(t f (t,t, (3.21 E (t,, E(t 1, + (2L 1/2 f(t f (t,t + C E(t 1, DG approxmaton for te Posson problem. Consder te DG approxmaton to te frst order system (3.18: fnd (E, Φ V r Vr suc tat for all : E z dx = Φ z x dx + [( Φ z +1/2 ( Φ z + 1/2 z V, r (3.22 E p x dx [(Êp +1/2 (Êp + 1/2 = (ρ 1p dx p V r, (3.23
12 12 B. AYUSO, J. A. CARRLLO AND C.-W. SHU were ( Φ 1/2 and (Ê 1/2 are te numercal fluxes. n ts work we focus on te followng famly of DG-scemes (see owever remark 3.5: { ( Φ 1/2 = {Φ } 1/2 c 12 [ Φ 1/2 + c 22 [ E 1/2, (3.24 (Ê 1/2 = {E } 1/2 + c 12 [ E 1/2 + c 11 [ Φ 1/2, were te parameters c 11, c 12 and c 22 depend solely on x 1/2, and are stll at our dsposal. At te boundary nodes due to perodcty n x we mpose ( Φ 1/2 = ( Φ Nx+1/2, (Ê 1/2 = (Ê Nx+1/2. Followng [18 we defne a(e, z := E zdx + c 22 [ E 1/2 [ z 1/2, b(φ, z := Φ z x dx + ({Φ } c 12 [ Φ [ z 1/2, c(φ, p := c 11 [ Φ 1/2 [ p 1/2, and A((E, Φ ; (z, p = a(e, z + b(φ, z b(p, E + c(φ, p. Tus, problem (3.22-(3.23 can be rewrtten as: fnd (E, Φ V r V r suc tat A((E, Φ ; (z, p = (ρ 1p dx (z, p V r V r. (3.25 Note tat A(, nduces te followng sem-norm (z, p H 1 ( H 1 ( : (z, p 2 A := A((z, p; (z, p = z 2, + c 1/2 22 [ z 2,γ x + c 1/2 11 [ p 2,γ x. (3.26 We also defne te norm for all r (E, Φ 2 r+1, := E 2 r+1, + Φ 2 r+2, (E, Φ H r+1 ( H r+2 (. (3.27 We next descrbe te specfc coces of te metods we consder (by specfyng te parameters n (3.24. We restrct ourselves to k 1, k beng te order of approxmaton used for f. ( Local dscontnuous Galerkn (LDG metod: we take r = k + 1 so te spaces are V r = V k+1 and we set c 22 = and c 11 = c 1 wt c a strctly postve constant. Ts metod was frst ntroduced n [27 for a tme dependent convecton dffuson problem (wt c 11 = O(1. n ts paper we take c 11 = c 1 wt c a postve constant, and c 12 = 1/2; tat s: { (Ê 1/2 = {E } 1/2 c 12 [ E 1/2 + c 1 [ Φ 1/2, ( Φ 1/2 = {Φ } 1/2 + c 12 [ Φ 1/2 c 12 = 1 2. (3.28 For te approxmaton of lnear problems, t as been proved (see [27, 18 convergence of order r + 1 and r for Φ and E, respectvely.
13 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 13 ( Mnmal dsspaton LDG and DG metods (MD-LDG and MD-DG: we set r = k and te spaces are taken as V r = V k. For te MD-LDG metod, te numercal fluxes are defned by takng n (3.24 c 22 =, c 12 = 1/2 and c 11 = except at a boundary node, tat s, { (Φ 1/2 = (Φ 1/2 (E 1/2 = (E + 1/2 + c 11 [ Φ 1/2, c 11 = { Nx 1, cr 1 = N x. (3.29 For te MD-DG metod te same coce apples except for (Φ 1/2 = (Φ 1/2 + c 22 [ E 1/2 wt c 22 = c/r. For te approxmaton of lnear problems, te MD- LDG metod was frst ntroduced for te 2D case n [23 but wt c 11 = O(1 rater tan O( 1 at te boundary. Te analyss n te one-dmensonal case for bot te MD-LDG and te MD-DG can be found n [19, were te autors sow tat te approxmaton to E, wt bot metods, superconverges wt order r + 1. ( General DG & ybrdzed LDG metod: we set r = k so tat te spaces are taken as V r = V k, and we take te numercal fluxes as n (3.24 wt: c 11, c 22, > c 12 bounded c 11 1 c 22. Superconvergence results are proved n [22 (for dmenson d 2 for te approxmaton of lnear problems. Anoter opton wc also provdes superconvergence and could be effcently mplemented, s te ybrdzed LDG metod (see [22 n wc te numercal fluxes can be recast n te form (3.24 by settng: ( ( ( τ (Ê 1/2 = τ + +τ (E + 1/2 + τ + τ + +τ (E 1/2 + τ τ + τ + +τ [ Φ 1/2, ( ( ( τ ( Φ 1/2 = + (Φ + 1/2 + τ (Φ 1/2 + 1 [ E 1/2, τ + +τ τ + +τ τ + +τ were τ ± are non-negatve constants. To aceve superconvergence, t s enoug to take n eac nterval one τ at one end and at te oter end we set τ =. Superconvergence can be sown by followng te analyss n [22 but usng te specal projectons defned troug (2.13-(2.14. As t appened wt RT k approxmaton, our Posson problem s nonlnear and terefore te estmates sown n [18, [19 and [22 are not drectly applcable. However, we ave te followng result, wose proof can be found n Appendx A. Lemma 3.4. Let k 1 and let (E, Φ C ([, T ; V r V r be te DG approxmaton (3.22-(3.23-(3.24 to te Posson problem (3.18, wt any of te tree coces (, ( or (. Let (E, Φ C ([, T ; H r+1 ( H r+2 ( Ten, te followng estmates old for all t [, T, E(t E (t 2, C 2(k+1 (E(t, Φ(t 2 r+1, + 2L f(t f (t 2,T, (3.3 were r s te order of polynomals of V r also olds as gven n (, (, (. Furtermore, t (E(t E (t, Φ(t Φ (t 2 A C2(k+1 (E(t, Φ(t 2 r+1, +2L f(t f (t 2,T. were r = k + 1 for ( and r = k for ( and (.
14 14 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Remark 3.5. Snce k 1, one mgt consder any of te (consstent and stable DG metods tat ft n te framework gven n [4 for approxmatng te Posson problem (1.2. Most of te results sown n ts paper for te general LDG dscretzaton (wt general c 12, old (wt mnor canges n te proofs for any of te resultng metods. For te sake of concseness, te detals are omtted. 4. Error Analyss. We start by sowng a cell-entropy nequalty [44 for te proposed DG scemes (3.1, wc guarantees ter L 2 -stablty. We ten derve te error equaton and gve some auxlary results tat are used n te proofs of te man results, wc are gven at te end of te secton Stablty. Next proposton sows tat te above selecton of te numercal fluxes s enoug to preserve te L 2 -stablty of numercal soluton of (3.1-(3.2, for all k. Proposton 4.1 (L 2 -stablty. Let k and let f Z k be te approxmaton (3.1-(3.2 of problem (1.1, wt te numercal fluxes as n (3.3. Ten f (t,t f (,T t [, T. (4.1 Proof. By settng ϕ = f n (3.2 we ave B,j(E ;f, f = J j E [ (Ê f f (f 2 dv dx 1 J j t 2 J j (f 2 dv dx + v J j x,j+1/2 (Ê f f + v (f 2 x dv dx [ ( vf f +1/2,v ( vf f + 1/2,v x,j 1/2 dx. Takng nto account tat E depends only on x (troug f wle v s ndependent of x, ntegraton of te second and trd volume terms leads to dv B,j (E ; f, f = 1 d [ 2 dt f 2,T j + F+1/2,j F 1/2,j + Θ F 1/2,j + [Ĝ,j+1/2 Ĝ,j 1/2 + Θ G,j 1/2, (4.2 were F +1/2,j, Ĝ,j+1/2 are defned for all, j, as and Θ F 1/2,j = Θ G,j 1/2 = J j F +1/2,j = Ĝ,j+1/2 = J j [ v 2 (f 2 vf f [ E 2 (f 2 Ê f f [ v 2 (f 2 vf f [ E 2 (f 2 Ê f f 1/2,v dv + dx x,j 1/2 J j +1/2,v dv dx, x,j+1/2 [ v 2 (f 2 + vf f + [ E 2 (f 2 + Ê f f + 1/2,v dv, dx. x,j 1/2
15 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 15 We next sow tat te coce (3.3 ensures tat bot Θ F 1/2,j and ΘG,j 1/2, for all and j, are non-negatve. By rewrtng our coce of te numercal fluxes (3.3 as: (vf = v{f } v 2 [ f, [E f = E {f } + E 2 [ f, (4.3 and usng tat [ f 2 = 2{f }[ f, t can be easly seen tat Θ F 1/2,j and ΘG,j 1/2 become [ v Θ F 1/2,j = 2 [ f 2 vf v [ f dv = 2 [ f 2 1/2,vdv, (4.4 Θ G,j 1/2 = J j 1/2,v J j [ Ê f [ f E 2 [ f 2 E dx = x,j 1/2 2 [ f 2 x,j 1/2dx. (4.5 Terefore, Θ F 1/2,j and ΘG,j 1/2 for all and j and so substtuton n (4.2 leads to 1 d [ f 2 dv dx + F+1/2,j 2 dt F 1/2,j + [Ĝ,j+1/2 Ĝ,j 1/2, T,j By summng n te above nequalty over and j, te flux terms telescope and tere s no boundary term left because of te perodc (for and compactly supported (for j boundary condtons. Hence, 1 d f 2 2 dt dv dx = 1 d T,j 2 dt f 2.T, (4.6,j and terefore, ntegraton n tme of te above nequalty yelds to (4.1. Remark 4.2. By carefully revsng te proof one realse tat n fact nequalty (4.6 s replaced by te dentty ( 1 d 2 dt f 2,T + v 1/2 [ f 2,Γ x + E 1/2 [ f 2,Γ v =. (4.7 Terefore, by defnng te norm f (t 2 := f (t 2,T + t v 1/2 [ f (s 2,Γ x ds+ t te concluson of Proposton 4.1 can be reformulated as: E 1/2 [ f (s 2,Γ v ds, (4.8 f (t 2 = f ( 2,T f ( 2 for all t [, T. Fnally, we note tat for te convergence and error analyss of numercal scemes for non-lnear problems, one usually needs to assume/prove tat some a-pror estmate on te approxmate soluton olds for all tme. n fact, wat s generally done s to assume tat tere exsts some C κ > suc tat, f f,t C κ, t [, T,
16 16 B. AYUSO, J. A. CARRLLO AND C.-W. SHU were,t usually refer to a stronger norm tan te one for wc te error analyss s carred out. For nstance,t =,,T f te error analyss s carred out n te L 2 or energy norm, see [49. We ws to stress tat n te present work, due to te structure of te contnuous problem, suc type of assumpton s not requred. Te man reason s tat altoug our L 2 -error analyss requres a bound on E,,, suc an estmate would depend ultmately on ρ (zero order moment of f, wc n general s more regular tan f tself. n te end, ts fact allows for gettng a bound for E,, dependng on te L 2 -error f f,t, for wc we can easly guarantee tat tere exsts c κ > suc tat, f f,t c κ, t [, T. (4.9 Estmate (4.9 follows from te L 2 conservaton property of te contnuous soluton (2.7 and te L 2 -stablty of ts approxmaton f gven n Proposton 4.1, togeter wt trangle nequalty and te L 2 -stablty of te standard L 2 projecton, (2.18 wt p = 2, f(t f (t 2,T 2( f(t 2,T + f (t 2,T 2 f 2,Ω + 2 P k (f 2,T 2(1 + C f 2,T = c κ. Let us pont out tat ts result allows us to obtan error estmates tat old for all and not only n te asymptotc regme Error equaton and specal projecton. To derve te error equaton te weak formulaton (2.6 s of lttle use, snce we sould take te test functon n Z. Hence, by allowng te test functon to be dscontnuous we fnd tat te true soluton satsfes te varatonal formulaton: were B,j (E; f, ϕ = N x N v B,j (E; f, ϕ = ϕ Z k, (4.1 =1 j=1 f T,j t ϕ dv dx [ + (vfϕ +1/2,v (vfϕ + 1/2,v J j dv dx + vf ϕ T,j x [ (E dv fϕ E f ϕ T,j v dv dx (4.11 x,j+1/2 ( E fϕ + x,j 1/2 dx, E beng te restrcton of te electrostatc feld E to ;.e., E = E. Subtractng (3.1 from (4.1 we obtan te error equaton, =,j B,j (E; f, ϕ B,j (E ; f, ϕ (4.12 =,j a,j (f f, ϕ +,j N,j (E; f, ϕ N,j (E ; f, ϕ ϕ Z. were te blnear form a(, =,j a,j(, gaters all lnear terms: [ f a,j (f, ϕ = t ϕ ϕ vf dv dx x J j + J j [ ( vf ϕ +1/2,v ( vf ϕ + 1/2,v dv
17 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 17 and N,j (E ;, (resp. N,j (E;, carres te nonlnear part; N,j (E ; f, ϕ = E f ϕ dv dx J j v [ (Ê f ϕ (Ê f ϕ + x,j+1/2 N,j (E ; f, ϕ = E f ϕ dv dx J j v [ (E fϕ x,j+1/2 ( E + fϕ x,j 1/2 x,j 1/2 dx. dx, Notce tat due to te nonlnearty, te true soluton f does not satsfy te equatons defnng te numercal sceme (3.1 (3.2. n fact we ave a consstency error: N (E ; f, ϕ N (E; f, ϕ for all ϕ Z, wc s dden n te nonlnear error N (E; f, ϕ N (E ; f, ϕ. Specal Projecton: We next ntroduce te 2-dmensonal projecton operator Π : C (Ω Z k wc s defned n te followng way. Let T,j = J j be an arbtrary element of T and let w C (T,j. Te restrcton of Π (w to T,j s defned by { πx π Π (w = v (w, f sgn(e = constant, P k π v (w, f sgn(e constant, (4.13 were Px k denotes te standard L 2 -projecton onto P k ( defned n (2.11 and π x, π v are defned by π x (w = { π + x (w f E >, πx (w f E <, π v (w = { π v (w f v >, π v + (w f v <, (4.14 wt π x ± : C ( V k and π± v : C (J j V k beng te specal projecton operators n te x and v drecton respectvely, defned as n (2.13-(2.14. Te defnton of projecton Π s nspred n tose consdered n [48, 23 and tat ntroduced n [62 for te analyss of Runge-Kutta DG metods for conservaton laws, see Remark 4.4. Note tat takng nto account (4.13-(4.14 togeter wt (2.13-(2.14, t s stragtforward to see tat Π (w s unquely defned. Te next lemma, altoug elementary, provdes several approxmaton results needed for our analyss. Lemma 4.3. Let w H s+2 (T,j, s and let Π be te projecton operator defned troug (4.13-(4.14. Ten, w Π (w,tj C mn (s+2,k+1 w s+1,tj, w Π (w,e C mn (s+ 3 2,k+ 1 2 w s+1,tj, e =, J j T j. (4.15 Proof. From te defnton (4.13 we dstngus two cases. f T j s an element suc tat sgn(e (x s constant x T j, te proof s te same as [18, Lemma 3.2. f on te contrary, T j s suc tat x T j for wc E (x =, we ave Π (w = P k π v (w. But stll, snce Π s a polynomal preservng and lnear operator, estmates (4.15 follow also n ts case from Bramble-Hlbert lemma, trace Teorem and standard scalng arguments. Detals are omtted for te sake of concseness.
18 18 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Summng estmates (4.15 from Lemma 4.3, over elements of te partton T, w Π (w,ω + 1/2 w Π (w,γ C k+1 w k+1,ω w H k+1 (Ω. (4.16 Now, denotng by we can wrte ω = Π (f f, ω e = Π (f f, (4.17 f f = [Π (f f [Π (f f = ω ω e. (4.18 Ten, by takng as test functon ϕ = ω Z k, te error equaton (4.12 becomes [ a(ω ω e, ω + N,j (E ; f, ω N,j (E ; f, ω =. (4.19,j We next defne K 1 (v, f, ω =,j K 1,j (v, ωe, ω, K 2 (E, f, ω =,j K 2,j (E, f, ω, (4.2 were K,j 1 (v, f, ω = vω e ωx dv dx T,j [ ( vω e (ω +1/2,v ( vω e (ω + 1/2,v dv, (4.21 J j K,j(E 2, f, ω = E ω e ωv dv dx T,j [(Ê ω e (ω x,j+1/2 (Ê ω e (ω + x,j 1/2 dx. (4.22 Te next two lemmas provde estmates for te terms defned n (4.2. Bot lemmas extend and generalze [23, Lemma 3.6 to te case of varable coeffcents and nonlnear problems, respectvely. To keep te readablty flow of te paper, te proofs of tese tecncal lemmas are postponed tll Appendx B. Remark 4.4. We ws to note tat te defnton (4.13 of Π s done n terms of E (and v, wle te defnton of te numercal fluxes s done n terms of E (and v. Ts s due to te non-lnearty of te problem and t s nspred n te deas used n [62. By defnng Π n terms of E rater tan E and usng te regularty of te soluton, we wll be able to estmate optmally te expresson K 2 wtout any furter assumpton on te mes partton T. Lemma 4.5. Let T be a cartesan mes of Ω, k 1 and let f Z k be te approxmate dstrbuton functon satsfyng (3.1-(3.2. Let f C ([, T ; H k+2 (Ω and let K 1 be defned as n (4.2. Assume tat te partton T s constructed so tat v = corresponds to a node of te partton. Ten, te followng estmate olds true K 1 (v, f, ω C k+1 ( f k+1,ω + CL f k+2,ω ω,t. (4.23
19 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 19 Lemma 4.6. Let T be a cartesan mes of Ω, k 1 and let (E, f W Z k be te soluton to (3.1-(3.2 wt W a fnte element space, conformng or non-conformng, of at least frst order (W = W k+1 or W = V r. Let (E, f C ([, T ; W 1, ( H k+2 (Ω and let K 2 be defned as n (4.2. Ten, te followng estmate olds K 2 (E, f, ω C k E E,, f k+1,ω ω,t ( C k+1 ( f k+2,ω E,, + f k+1,ω E 1,, ω,t Auxlary results. We next prove two lemmas tat are needed for te proofs of te man Teorems 4.9, 4.13, and Te frst one reduces te expresson for te lnear part of te error equaton (4.19: Lemma 4.7. Let f C (Ω and let f Z k equalty olds a(f f, ω = ( ω t ω e t ω dxdv +,j T,j,j wt k 1. Ten, te followng J j v 2 [ ω 2 1/2,v dv + K1 (v, f, ω. Proof. From (4.18 we get a(f f, ω = a(ω, ω a(ω e, ω. Argung as for (4.4 n te proof of Proposton 4.1 (note tat ω Z, we ave for te frst term a(ω, ω = ωt ω dxdv + v,j J j,j J j 2 [ ω 2 1/2,vdv. (4.25 Te defnton (4.21 of K 1, te contnuty of f and te numercal fluxes (3.3 mply a(ω e, ω = ωt e ω dxdv vω e ωx dxdv [ vω e [ ω 1/2,v dv,j J j J j,j J j = ωt e ω dxdv K 1 (v, f, ω.,j J j wc togeter wt (4.25 completes te proof. Te oter auxlary Lemma deals wt te error comng from te nonlnear term: Lemma 4.8. Let E C (, f C (Ω and f Z k wt k 1. Ten, te followng dentty olds [N,j (E; f; ω N,j (E ; f, ω = (4.26,j =,j E 2 [ ω 2 x,j 1/2 dx,j T,j [E E f v ω dv dx K 2 (E, f, ω. Proof. Subtractng te nonlnear terms n (4.11 and (3.2 we ave N,j (E; f;ω N,j(E ; f, ω = [E Jj f E f ω dv dx v [ ([E f Ê f ω x,j 1/2 ([E f Ê f ω + x,j 1/2 dx. (4.27
20 2 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Notce tat te ntegrand of te volume part above, can be decomposed as [ E f E f ± E f = [E E f + E (f f, (4.28 and so substtutng nto (4.27 we fnd were T 1 = T 3 = N,j (E; f; ω N,j(E ; f, ω = T 1 + T 2 + T 3, (4.29 [E E fω v dv dx, T 2 = E [f f ωv dv dx, J j J j [([(E f + Ê f (ω + x,j 1/2 ([(E f Ê f (ω x,j+1/2 dx. Snce neter E nor E depend on v, ntegraton by parts of T 1 gves T 1 = T 1a + T 1b : T 1 = [E E f J j v ω dv dx + (E E [(fω x,j+1/2 (fω + x,j 1/2 dx. Summng now over j and takng nto account te contnuty of f we fnd for T 1b, T 1b = (E E (f [ ω x,j 1/2 dx. (4.3 j j We next deal wt T 2. From te splttng (4.18 we ave T 2 = E ω ωv dv dx E ω e ωv dv dx = T 2a + T 2b, J j J j and so, ntegratng te frst term and summng over j we easly get T 2a = 1 E (ω 2 j j Jj 2 dv dx = E v j 2 [ (ω 2 x,j 1/2 dx. (4.31 We fnally deal wt te boundary terms collected n T 3. Summaton over j and te contnuty of E and f gves T 3 = [E f Ê f x,j 1/2 [ ω x,j 1/2 dx. j j Ten, reasonng as n (4.28, we deduce for all tat ( ( E f Ê f ±Ef = (E Ef + Ef Ê f = (E Ef + E (ω E (ωe, were n te last step we ave used te contnuty of f togeter wt te consstency of te numercal flux Ê f. Tus, substtutng back nto T 3, we nfer T 3 = ( (E E f [ ω + Ê ω [ ω Ê ωe [ ω dx j j x,j 1/2 = j T 3a + j T 3b + j T 3c.
21 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 21 Ten, for te frst term, T 3a, recallng te expresson (4.3, we get [T 1b + T 3a =. (4.32 j Next, summng T 3b and T 2a from (4.31 and argung as for (4.5 n te proof of Proposton 4.1, we fnd [T 2a + T 3b = j j E 2 [ ω 2 x,j 1/2dx. (4.33 Fnally, recallng te defnton (4.22 of K 2 and addng up T 3c wt T 2b we get [T 2b + T 3c = K 2 (E, f, ω. j Tus, substtutng te above dentty togeter (4.33 and te expresson for T 1a nto te equaton (4.29 we reac (4.26 and so te proof s complete Approxmaton. We next sow te man convergence results of ts work provng a-pror error estmates for te electron dstrbuton f, for all te proposed metods. n eac case, as a byproduct result, we also get te correspondng convergence results for te electrostatc feld E. Te secton s closed wt some remarks about te comparson wt te convergence of oter metods. We start wt te result for te conformng-dg metod: Teorem 4.9 (Conformng-DG metod. Let k 1 and consder te unque compactly supported soluton of te Vlasov-Posson system (1.1-(1.2 gven by Teorem 2.1 wt f C 1 ([, T ; H k+2 (Ω and E C ([, T ; W 1, (. Let (E, f C ([, T ; W k+1 C 1 ([, T ; Z k be te conformng-dg approxmaton,.e. soluton of (3.1, (3.2 and (3.12. Ten, f(t f (t,t C k+1 t [, T, were C depends on te tme t, te polynomal degree k, te sape regularty of te partton and depends also on f and on E troug te norms C = C ( f(t k+2,ω, f t (t k+1,ω, L, E(t 1,,. Proof. Recallng te error equaton (4.19 a(ω ω e, ω + N (E ; f, ω N (E ; f, ω =, and usng Lemmas 4.7 and 4.8, we ave ωt ω dv dx + v,j T,j,j J j 2 [ ω 2 +1/2,v dv +,j = ωt e ω dv dx +,j T,j,j E 2 [ ω 2 x,j+1/2 dx T,j [E E f v ω dv dx K 1 (v, f, ω + K 2 (E, f, ω = T 1 + T 2 K 1 + K 2. (4.34
22 22 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Notce tat te left and sde of te above equaton, s exactly wat results after summaton over and j n (4.2 from Proposton (4.1, see also (4.7. Ten, t s enoug to estmate te terms on te rgt and sde of te above equaton. Te frst term s drectly estmated by usng Caucy-Scwartz and te artmetc-geometrc nequaltes togeter wt te nterpolaton property (4.16 T 1 C 2 ( ωe t 2,T + ω 2,T C 2k+2 f t H k+1 (Ω + C ω 2,T. (4.35 Te second term on te rs of (4.34, s readly estmated by usng Hölder nequalty togeter wt estmate (3.16 from Lemma 3.2, te splttng (4.18, te artmetcgeometrc nequalty and te nterpolaton estmate (4.16, T 2 C E E,, f v,ω ω,t CC 2 f f,t f v,ω ω,t CC 2 ( ω e,t + ω,t f v,ω ω,t CC 2 2k+2 f 2 k+1,ω f v,ω + C 2 f v,ω ω 2,Ω, (4.36 were C 2 L 1/2 s te constant n Lemma 3.2. Estmate (4.23 from Lemma 4.5 and te artmetc-geometrc nequalty gve for te trd term, K 1 C 2k+2 L 2 f 2 k+2,ω + C ω 2,T. (4.37 Last term s bounded by usng estmate (4.24 from Lemma 4.6 and argung smlarly as for T 2 ; usng estmate (3.16 from Lemma 3.2, te splttng (4.18, te artmetcgeometrc nequalty and te nterpolaton estmate (4.16, K 2 C k f k+1,ω ( ω e,t + ω,t ω,t + C k+1 f k+2,ω E 1,, ω,t C 2k+2 ( f 2 k+2,ω E 2 1,, + C 2 k f 3 k+1,ω + C(1 + k f k+1,ω ω 2,T. Ten, by substtutng te above estmate togeter wt (4.35, (4.36 and (4.37 nto te error equaton (4.34, we conclude d dt ω (t 2,T A(t ω (t 2,T + 2k+2 B(t wt A(t = (C + L 1/2 f v,ω + CL 1/2 k f k+1,ω and B(t= C f 2 k+2,ω(l 2 + E 2 1,,+ f t 2 k+1,ω+cl 1/2 f 2 k+1,ω( f v 2,Ω+ k f k+1,ω. Terefore, ntegraton n tme of te above nequalty and a standard applcaton of Gronwall s nequalty gves te error estmate, ω (t 2,T C 2 2k+2, (4.38 were C s as stated n te clam. Hence, Teorem 4.9 follows from te trangle nequalty and te nterpolaton property (4.16. As a drect consequence of Teorem 4.9 togeter wt estmates (3.15 and (3.16 of Lemma 3.2, we obtan te followng result on te error of te electrostatc feld. Corollary 4.1. estmates old Under te ypotess of Teorem 4.9, te followng error E(t E (t, C C 1 k+1 t [, T, E(t E (t, C C 2 k+1 t [, T,
23 DG METHODS FOR ONE-DMENSONAL VLASOV-POSSON SYSTEM 23 were C 1 and C 2 are gven n (3.15 and (3.16, respectvely and C n Teorem 4.9. Next result establses te convergence for te RT k -DG metod: Teorem 4.11 (RT k -DG metod. Let k 1 and consder te unque compactly supported soluton of te Vlasov-Posson system (1.1-(1.2 gven by Teorem 2.1 wt f C 1 ([, T ; H k+2 (Ω and E C ([, T ; H k+1 (. Let ((E, Φ, f C ([, T ; (W k+1 V k C1 ([, T ; Z k be te RT k-dg approxmaton soluton of (3.1, (3.2, (3.19, and (3.2. Ten, f(t f (t,t C 4 k+1 t [, T, were C 4 depends on te tme t, te polynomal degree k, te sape regularty of te partton and depends also on f and on E troug te norms C 4 = C 4 ( f(t k+2,ω, f t (t k+1,ω, L, E(t k+1,, Proof. Te proof follows exactly te same lnes as te proof of Teorem 4.9. n ts case, to bound te error E E,, tat appears n te estmates for T 2 and K 2 one as to use estmate (3.21 from Lemma 3.3. We omt te detals for te sake of concseness. Corollary estmates old Under te ypotess of Teorem 4.11, te followng error E(t E (t, + E(t E (t 1, 2C 4 L 1/2 k+1 + C k+1 E k+1, E(t E (t,, C 4 L 1/2 k+1 + C k+1 E k+1, for all t [, T, were C 4 s te constant of Teorem Fnally, we sow te convergence for te full DG approxmaton: Teorem 4.13 (DG-DG metod. Let r k 1 and consder te unque compactly supported soluton of te Vlasov-Posson system (1.1-(1.2 gven by Teorem 2.1 wt f C 1 ([, T ; H k+2 (Ω and E C ([, T ; H r+1 (. Let ((E, Φ, f C ([, T ; V r Vr C1 ([, T ; Z k be te DG-DG approxmaton tat satsfes (3.1, (3.2, (3.22, and (3.23 wt any of te tree coces (, ( or (. Ten, f(t f (t,t C 4 k+1 t [, T, were C 5 depends on tme t, te polynomals degrees k and r, te sape regularty of te partton and depends also on f and on (E, Φ troug te norms C 5 = C 5 ( f(t k+2,ω, f t (t k+1,ω, L, (E, Φ r+1,. Proof. Te proof follows essentally te same lnes as te proof of Teorems 4.9 and 4.11, but dealng wt T 2 we use estmate (3.3 from Lemma 3.4; T 2 C E E, f v,,ω ω,t [C k+1 (E, Φ r+1, + (2L 1/2 f f,t f v,,ω ω,t C 2k+2 ( (E, Φ 2 r+1, + 2L f 2 k+1,ω f v,,ω + (C + (2L 1/2 f v,,ω ω 2,Ω. (4.39
24 24 B. AYUSO, J. A. CARRLLO AND C.-W. SHU Also, to bound for K 2 we frst note tat E = P k+1 (E snce E V r (and r = k +1 or r = k, so tat nverse nequalty, estmate (2.16 and te L 2 -stablty of te L 2 - projecton gve E E,, E P k+1 (E,, + C 1/2 P k+1 (E E, C k+1 E k+1,, + C 1/2 E E,. (4.4 Ten, usng estmate (4.24 from Lemma 4.6 togeter wt te above estmate and te L 2 -bound for te error E E gven n Lemma 3.4, we get K 2 C(1 + L 1/2 k 1/2 f k+1,ω ω 2,T + C 2k+2 ( E 2 1,, f 2 k+2,ω + (E, Φ 2 r+1, f 2 k+1,ω + k 1/2 f 3 k+1,ω, were we ave neglected g order terms of order O( 4k 1/4. Notng tat k 1, te proof can now be completed by argung as n te proof of Teorem 4.9. We omt te detals for te sake of brevty. Remark Takng nto account te defnton (4.8 of te norm (see Remark 4.2, observe tat n te proof of Teorems 4.9, 4.11 and 4.13, smlarly as ow t s obtaned te error estmate (4.38, we also get ω (t 2 C 2 s 2k+2 s =, 4, 5. (4.41 As a drect consequence of Teorem 4.13 and Lemma 3.4 we ave te followng corollary wose proof s omtted. Corollary Under te ypotess of Teorem 4.13, te followng error estmates old for all t [, T E(t E (t 2, C 2k+2 (E(t, Φ(t 2 r+1, + C2 5 L2k+2 were C 5 s te constant of Teorem 4.13, and E(t E (t 2, + c 1/2 11 [ Φ (t 2,γ x + c 1/2 22 [ E (t 2,γ x C 6 2k+2, wt C 6 = C5 2L + C (E(t, Φ(t 2 r+1, were r = k + 1 for ( and r = k for ( and (. Remark 4.16 (Order of convergence attaned by oter metods. As noted n te ntroducton, tere are very few works dealng wt te convergence and error analyss of Euleran solvers for te (perodc Vlasov-Posson system. Hg order scemes ave been only analyzed n te context of sem-lagrangan metods [7, 8, 1. Altoug, t s dffcult to compare ter results wt ours, snce tese analyss deal wt fully dscrete scemes, we just menton brefly wat one can expect to aceve wt tese metods n te case of a constant Courant-Fredrcs-Levy CFL (ν = dt/ =constant and n te case were te tme step dt were taken te largest possble. n [7, error estmates n L of frst order (for CFL=constant and slgtly better tan frst order (at most of order 4/3 for te largest possble tme step, are sown assumng te ntal data s of class C 2. Hg order scemes, by usng polynomals of degree k n te reconstructon, are consdered n [8, 1. Tere, te autors prove error bounds for te dstrbuton functon and te electrostatc feld n L 2 and L, respectvely, of at most order k (f CFL=constant and of order 2(k + 1/3 f te largest possble tme step wants to be used. Tese works typcally requre te tecncal assumpton f W k+1, (Ω.
The Finite Element Method: A Short Introduction
Te Fnte Element Metod: A Sort ntroducton Wat s FEM? Te Fnte Element Metod (FEM) ntroduced by engneers n late 50 s and 60 s s a numercal tecnque for solvng problems wc are descrbed by Ordnary Dfferental
More informationStanford University CS254: Computational Complexity Notes 7 Luca Trevisan January 29, Notes for Lecture 7
Stanford Unversty CS54: Computatonal Complexty Notes 7 Luca Trevsan January 9, 014 Notes for Lecture 7 1 Approxmate Countng wt an N oracle We complete te proof of te followng result: Teorem 1 For every
More informationAppendix B. The Finite Difference Scheme
140 APPENDIXES Appendx B. The Fnte Dfference Scheme In ths appendx we present numercal technques whch are used to approxmate solutons of system 3.1 3.3. A comprehensve treatment of theoretcal and mplementaton
More information5 The Laplace Equation in a convex polygon
5 Te Laplace Equaton n a convex polygon Te most mportant ellptc PDEs are te Laplace, te modfed Helmoltz and te Helmoltz equatons. Te Laplace equaton s u xx + u yy =. (5.) Te real and magnary parts of an
More informationTR/95 February Splines G. H. BEHFOROOZ* & N. PAPAMICHAEL
TR/9 February 980 End Condtons for Interpolatory Quntc Splnes by G. H. BEHFOROOZ* & N. PAPAMICHAEL *Present address: Dept of Matematcs Unversty of Tabrz Tabrz Iran. W9609 A B S T R A C T Accurate end condtons
More informationNumerical Simulation of One-Dimensional Wave Equation by Non-Polynomial Quintic Spline
IOSR Journal of Matematcs (IOSR-JM) e-issn: 78-578, p-issn: 319-765X. Volume 14, Issue 6 Ver. I (Nov - Dec 018), PP 6-30 www.osrournals.org Numercal Smulaton of One-Dmensonal Wave Equaton by Non-Polynomal
More informationLecture 12: Discrete Laplacian
Lecture 12: Dscrete Laplacan Scrbe: Tanye Lu Our goal s to come up wth a dscrete verson of Laplacan operator for trangulated surfaces, so that we can use t n practce to solve related problems We are mostly
More information3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X
Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number
More informationMore metrics on cartesian products
More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of
More informationOn Pfaff s solution of the Pfaff problem
Zur Pfaff scen Lösung des Pfaff scen Probles Mat. Ann. 7 (880) 53-530. On Pfaff s soluton of te Pfaff proble By A. MAYER n Lepzg Translated by D. H. Delpenc Te way tat Pfaff adopted for te ntegraton of
More informationNumerical Heat and Mass Transfer
Master degree n Mechancal Engneerng Numercal Heat and Mass Transfer 06-Fnte-Dfference Method (One-dmensonal, steady state heat conducton) Fausto Arpno f.arpno@uncas.t Introducton Why we use models and
More informationAPPENDIX A Some Linear Algebra
APPENDIX A Some Lnear Algebra The collecton of m, n matrces A.1 Matrces a 1,1,..., a 1,n A = a m,1,..., a m,n wth real elements a,j s denoted by R m,n. If n = 1 then A s called a column vector. Smlarly,
More informationNUMERICAL DIFFERENTIATION
NUMERICAL DIFFERENTIATION 1 Introducton Dfferentaton s a method to compute the rate at whch a dependent output y changes wth respect to the change n the ndependent nput x. Ths rate of change s called the
More informationLectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix
Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could
More informationCOMP4630: λ-calculus
COMP4630: λ-calculus 4. Standardsaton Mcael Norrs Mcael.Norrs@ncta.com.au Canberra Researc Lab., NICTA Semester 2, 2015 Last Tme Confluence Te property tat dvergent evaluatons can rejon one anoter Proof
More informationSolution for singularly perturbed problems via cubic spline in tension
ISSN 76-769 England UK Journal of Informaton and Computng Scence Vol. No. 06 pp.6-69 Soluton for sngularly perturbed problems va cubc splne n tenson K. Aruna A. S. V. Rav Kant Flud Dynamcs Dvson Scool
More informationMultigrid Methods and Applications in CFD
Multgrd Metods and Applcatons n CFD Mcael Wurst 0 May 009 Contents Introducton Typcal desgn of CFD solvers 3 Basc metods and ter propertes for solvng lnear systems of equatons 4 Geometrc Multgrd 3 5 Algebrac
More informationMultivariate Ratio Estimator of the Population Total under Stratified Random Sampling
Open Journal of Statstcs, 0,, 300-304 ttp://dx.do.org/0.436/ojs.0.3036 Publsed Onlne July 0 (ttp://www.scrp.org/journal/ojs) Multvarate Rato Estmator of te Populaton Total under Stratfed Random Samplng
More informationCHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE
CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng
More informationLecture 21: Numerical methods for pricing American type derivatives
Lecture 21: Numercal methods for prcng Amercan type dervatves Xaoguang Wang STAT 598W Aprl 10th, 2014 (STAT 598W) Lecture 21 1 / 26 Outlne 1 Fnte Dfference Method Explct Method Penalty Method (STAT 598W)
More informationDifference Equations
Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1
More informationExercise Solutions to Real Analysis
xercse Solutons to Real Analyss Note: References refer to H. L. Royden, Real Analyss xersze 1. Gven any set A any ɛ > 0, there s an open set O such that A O m O m A + ɛ. Soluton 1. If m A =, then there
More informationInner Product. Euclidean Space. Orthonormal Basis. Orthogonal
Inner Product Defnton 1 () A Eucldean space s a fnte-dmensonal vector space over the reals R, wth an nner product,. Defnton 2 (Inner Product) An nner product, on a real vector space X s a symmetrc, blnear,
More information338 A^VÇÚO 1n ò Lke n Mancn (211), we make te followng assumpton to control te beavour of small jumps. Assumpton 1.1 L s symmetrc α-stable, were α (,
A^VÇÚO 1n ò 1oÏ 215c8 Cnese Journal of Appled Probablty and Statstcs Vol.31 No.4 Aug. 215 Te Speed of Convergence of te Tresold Verson of Bpower Varaton for Semmartngales Xao Xaoyong Yn Hongwe (Department
More informationNON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS
IJRRAS 8 (3 September 011 www.arpapress.com/volumes/vol8issue3/ijrras_8_3_08.pdf NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS H.O. Bakodah Dept. of Mathematc
More informationModule 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur
Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:
More informationOn a nonlinear compactness lemma in L p (0, T ; B).
On a nonlnear compactness lemma n L p (, T ; B). Emmanuel Matre Laboratore de Matématques et Applcatons Unversté de Haute-Alsace 4, rue des Frères Lumère 6893 Mulouse E.Matre@ua.fr 3t February 22 Abstract
More informationA Discrete Approach to Continuous Second-Order Boundary Value Problems via Monotone Iterative Techniques
Internatonal Journal of Dfference Equatons ISSN 0973-6069, Volume 12, Number 1, pp. 145 160 2017) ttp://campus.mst.edu/jde A Dscrete Approac to Contnuous Second-Order Boundary Value Problems va Monotone
More informationThe Order Relation and Trace Inequalities for. Hermitian Operators
Internatonal Mathematcal Forum, Vol 3, 08, no, 507-57 HIKARI Ltd, wwwm-hkarcom https://doorg/0988/mf088055 The Order Relaton and Trace Inequaltes for Hermtan Operators Y Huang School of Informaton Scence
More informationThe finite element method explicit scheme for a solution of one problem of surface and ground water combined movement
IOP Conference Seres: Materals Scence and Engneerng PAPER OPEN ACCESS e fnte element metod explct sceme for a soluton of one problem of surface and ground water combned movement o cte ts artcle: L L Glazyrna
More informationThe Feynman path integral
The Feynman path ntegral Aprl 3, 205 Hesenberg and Schrödnger pctures The Schrödnger wave functon places the tme dependence of a physcal system n the state, ψ, t, where the state s a vector n Hlbert space
More informationHigh resolution entropy stable scheme for shallow water equations
Internatonal Symposum on Computers & Informatcs (ISCI 05) Hgh resoluton entropy stable scheme for shallow water equatons Xaohan Cheng,a, Yufeng Ne,b, Department of Appled Mathematcs, Northwestern Polytechncal
More informationShuai Dong. Isaac Newton. Gottfried Leibniz
Computatonal pyscs Sua Dong Isaac Newton Gottred Lebnz Numercal calculus poston dervatve ntegral v velocty dervatve ntegral a acceleraton Numercal calculus Numercal derentaton Numercal ntegraton Roots
More informationPhysics 5153 Classical Mechanics. D Alembert s Principle and The Lagrangian-1
P. Guterrez Physcs 5153 Classcal Mechancs D Alembert s Prncple and The Lagrangan 1 Introducton The prncple of vrtual work provdes a method of solvng problems of statc equlbrum wthout havng to consder the
More informationAppendix B. Criterion of Riemann-Stieltjes Integrability
Appendx B. Crteron of Remann-Steltes Integrablty Ths note s complementary to [R, Ch. 6] and [T, Sec. 3.5]. The man result of ths note s Theorem B.3, whch provdes the necessary and suffcent condtons for
More informationStanford University CS359G: Graph Partitioning and Expanders Handout 4 Luca Trevisan January 13, 2011
Stanford Unversty CS359G: Graph Parttonng and Expanders Handout 4 Luca Trevsan January 3, 0 Lecture 4 In whch we prove the dffcult drecton of Cheeger s nequalty. As n the past lectures, consder an undrected
More informationCompetitive Experimentation and Private Information
Compettve Expermentaton an Prvate Informaton Guseppe Moscarn an Francesco Squntan Omtte Analyss not Submtte for Publcaton Dervatons for te Gamma-Exponental Moel Dervaton of expecte azar rates. By Bayes
More informationON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION
Advanced Mathematcal Models & Applcatons Vol.3, No.3, 2018, pp.215-222 ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EUATION
More informationOn a direct solver for linear least squares problems
ISSN 2066-6594 Ann. Acad. Rom. Sc. Ser. Math. Appl. Vol. 8, No. 2/2016 On a drect solver for lnear least squares problems Constantn Popa Abstract The Null Space (NS) algorthm s a drect solver for lnear
More informationLecture 10 Support Vector Machines II
Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed
More informationDirect Methods for Solving Macromolecular Structures Ed. by S. Fortier Kluwer Academic Publishes, The Netherlands, 1998, pp
Drect Metods for Solvng Macromolecular Structures Ed. by S. Forter Kluwer Academc Publses, Te Neterlands, 998, pp. 79-85. SAYRE EQUATION, TANGENT FORMULA AND SAYTAN FAN HAI-FU Insttute of Pyscs, Cnese
More informationConvexity preserving interpolation by splines of arbitrary degree
Computer Scence Journal of Moldova, vol.18, no.1(52), 2010 Convexty preservng nterpolaton by splnes of arbtrary degree Igor Verlan Abstract In the present paper an algorthm of C 2 nterpolaton of dscrete
More informationSolving Singularly Perturbed Differential Difference Equations via Fitted Method
Avalable at ttp://pvamu.edu/aam Appl. Appl. Mat. ISSN: 193-9466 Vol. 8, Issue 1 (June 013), pp. 318-33 Applcatons and Appled Matematcs: An Internatonal Journal (AAM) Solvng Sngularly Perturbed Dfferental
More informationBOUNDEDNESS OF THE RIESZ TRANSFORM WITH MATRIX A 2 WEIGHTS
BOUNDEDNESS OF THE IESZ TANSFOM WITH MATIX A WEIGHTS Introducton Let L = L ( n, be the functon space wth norm (ˆ f L = f(x C dx d < For a d d matrx valued functon W : wth W (x postve sem-defnte for all
More informationTHE WEIGHTED WEAK TYPE INEQUALITY FOR THE STRONG MAXIMAL FUNCTION
THE WEIGHTED WEAK TYPE INEQUALITY FO THE STONG MAXIMAL FUNCTION THEMIS MITSIS Abstract. We prove the natural Fefferman-Sten weak type nequalty for the strong maxmal functon n the plane, under the assumpton
More informationGeneral viscosity iterative method for a sequence of quasi-nonexpansive mappings
Avalable onlne at www.tjnsa.com J. Nonlnear Sc. Appl. 9 (2016), 5672 5682 Research Artcle General vscosty teratve method for a sequence of quas-nonexpansve mappngs Cuje Zhang, Ynan Wang College of Scence,
More informationn α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0
MODULE 2 Topcs: Lnear ndependence, bass and dmenson We have seen that f n a set of vectors one vector s a lnear combnaton of the remanng vectors n the set then the span of the set s unchanged f that vector
More informationUniqueness of Weak Solutions to the 3D Ginzburg- Landau Model for Superconductivity
Int. Journal of Math. Analyss, Vol. 6, 212, no. 22, 195-114 Unqueness of Weak Solutons to the 3D Gnzburg- Landau Model for Superconductvty Jshan Fan Department of Appled Mathematcs Nanjng Forestry Unversty
More informationLecture 5.8 Flux Vector Splitting
Lecture 5.8 Flux Vector Splttng 1 Flux Vector Splttng The vector E n (5.7.) can be rewrtten as E = AU (5.8.1) (wth A as gven n (5.7.4) or (5.7.6) ) whenever, the equaton of state s of the separable form
More informationRepresentation theory and quantum mechanics tutorial Representation theory and quantum conservation laws
Representaton theory and quantum mechancs tutoral Representaton theory and quantum conservaton laws Justn Campbell August 1, 2017 1 Generaltes on representaton theory 1.1 Let G GL m (R) be a real algebrac
More informationCME 302: NUMERICAL LINEAR ALGEBRA FALL 2005/06 LECTURE 13
CME 30: NUMERICAL LINEAR ALGEBRA FALL 005/06 LECTURE 13 GENE H GOLUB 1 Iteratve Methods Very large problems (naturally sparse, from applcatons): teratve methods Structured matrces (even sometmes dense,
More informationREAL ANALYSIS I HOMEWORK 1
REAL ANALYSIS I HOMEWORK CİHAN BAHRAN The questons are from Tao s text. Exercse 0.0.. If (x α ) α A s a collecton of numbers x α [0, + ] such that x α
More informationCurvature and isoperimetric inequality
urvature and sopermetrc nequalty Julà ufí, Agustí Reventós, arlos J Rodríguez Abstract We prove an nequalty nvolvng the length of a plane curve and the ntegral of ts radus of curvature, that has as a consequence
More informationprinceton univ. F 17 cos 521: Advanced Algorithm Design Lecture 7: LP Duality Lecturer: Matt Weinberg
prnceton unv. F 17 cos 521: Advanced Algorthm Desgn Lecture 7: LP Dualty Lecturer: Matt Wenberg Scrbe: LP Dualty s an extremely useful tool for analyzng structural propertes of lnear programs. Whle there
More informationOne can coose te bass n te 'bg' space V n te form of symmetrzed products of sngle partcle wavefunctons ' p(x) drawn from an ortonormal complete set of
8.54: Many-body penomena n condensed matter and atomc pyscs Last moded: September 4, 3 Lecture 3. Second Quantzaton, Bosons In ts lecture we dscuss second quantzaton, a formalsm tat s commonly used to
More informationMATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS
MATH 241B FUNCTIONAL ANALYSIS - NOTES EXAMPLES OF C ALGEBRAS These are nformal notes whch cover some of the materal whch s not n the course book. The man purpose s to gve a number of nontrval examples
More informationELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM
ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM An elastc wave s a deformaton of the body that travels throughout the body n all drectons. We can examne the deformaton over a perod of tme by fxng our look
More information5 The Rational Canonical Form
5 The Ratonal Canoncal Form Here p s a monc rreducble factor of the mnmum polynomal m T and s not necessarly of degree one Let F p denote the feld constructed earler n the course, consstng of all matrces
More informationNP-Completeness : Proofs
NP-Completeness : Proofs Proof Methods A method to show a decson problem Π NP-complete s as follows. (1) Show Π NP. (2) Choose an NP-complete problem Π. (3) Show Π Π. A method to show an optmzaton problem
More informationDIFFERENTIAL FORMS BRIAN OSSERMAN
DIFFERENTIAL FORMS BRIAN OSSERMAN Dfferentals are an mportant topc n algebrac geometry, allowng the use of some classcal geometrc arguments n the context of varetes over any feld. We wll use them to defne
More informationThe equation of motion of a dynamical system is given by a set of differential equations. That is (1)
Dynamcal Systems Many engneerng and natural systems are dynamcal systems. For example a pendulum s a dynamcal system. State l The state of the dynamcal system specfes t condtons. For a pendulum n the absence
More informationAdditional Codes using Finite Difference Method. 1 HJB Equation for Consumption-Saving Problem Without Uncertainty
Addtonal Codes usng Fnte Dfference Method Benamn Moll 1 HJB Equaton for Consumpton-Savng Problem Wthout Uncertanty Before consderng the case wth stochastc ncome n http://www.prnceton.edu/~moll/ HACTproect/HACT_Numercal_Appendx.pdf,
More informationRobust Norm Equivalencies and Preconditioning
Robust Norm Equvalences and Precondtonng Karl Scherer Insttut für Angewandte Mathematk, Unversty of Bonn, Wegelerstr. 6, 53115 Bonn, Germany Summary. In ths contrbuton we report on work done n contnuaton
More informationInductance Calculation for Conductors of Arbitrary Shape
CRYO/02/028 Aprl 5, 2002 Inductance Calculaton for Conductors of Arbtrary Shape L. Bottura Dstrbuton: Internal Summary In ths note we descrbe a method for the numercal calculaton of nductances among conductors
More informationCanonical transformations
Canoncal transformatons November 23, 2014 Recall that we have defned a symplectc transformaton to be any lnear transformaton M A B leavng the symplectc form nvarant, Ω AB M A CM B DΩ CD Coordnate transformatons,
More informationTHE CARTIER ISOMORPHISM. These are the detailed notes for a talk I gave at the Kleine AG 1 in April Frobenius
THE CARTIER ISOMORPHISM LARS KINDLER Tese are te detaled notes for a talk I gave at te Klene AG 1 n Aprl 2010. 1. Frobenus Defnton 1.1. Let f : S be a morpsm of scemes and p a prme. We say tat S s of caracterstc
More informationPHYS 705: Classical Mechanics. Newtonian Mechanics
1 PHYS 705: Classcal Mechancs Newtonan Mechancs Quck Revew of Newtonan Mechancs Basc Descrpton: -An dealzed pont partcle or a system of pont partcles n an nertal reference frame [Rgd bodes (ch. 5 later)]
More informationLab session: numerical simulations of sponateous polarization
Lab sesson: numercal smulatons of sponateous polarzaton Emerc Boun & Vncent Calvez CNRS, ENS Lyon, France CIMPA, Hammamet, March 2012 Spontaneous cell polarzaton: the 1D case The Hawkns-Voturez model for
More informationarxiv: v1 [math.na] 26 Sep 2007
EQUIVALENCE THEOREMS IN NUMERICAL ANALYSIS : INTEGRATION, DIFFERENTIATION AND INTERPOLATION arxv:0709.4046v1 [mat.na] 26 Sep 2007 JOHN JOSSEY AND ANIL N. HIRANI Abstract. We sow tat f a numercal metod
More informationPerron Vectors of an Irreducible Nonnegative Interval Matrix
Perron Vectors of an Irreducble Nonnegatve Interval Matrx Jr Rohn August 4 2005 Abstract As s well known an rreducble nonnegatve matrx possesses a unquely determned Perron vector. As the man result of
More informationAffine transformations and convexity
Affne transformatons and convexty The purpose of ths document s to prove some basc propertes of affne transformatons nvolvng convex sets. Here are a few onlne references for background nformaton: http://math.ucr.edu/
More informationMaximizing the number of nonnegative subsets
Maxmzng the number of nonnegatve subsets Noga Alon Hao Huang December 1, 213 Abstract Gven a set of n real numbers, f the sum of elements of every subset of sze larger than k s negatve, what s the maxmum
More informationSalmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2
Salmon: Lectures on partal dfferental equatons 5. Classfcaton of second-order equatons There are general methods for classfyng hgher-order partal dfferental equatons. One s very general (applyng even to
More informationTR/28. OCTOBER CUBIC SPLINE INTERPOLATION OF HARMONIC FUNCTIONS BY N. PAPAMICHAEL and J.R. WHITEMAN.
TR/8. OCTOBER 97. CUBIC SPLINE INTERPOLATION OF HARMONIC FUNCTIONS BY N. PAPAMICHAEL and J.R. WHITEMAN. W960748 ABSTRACT It s sown tat for te two dmensonal Laplace equaton a unvarate cubc splne approxmaton
More informationTHE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens
THE CHINESE REMAINDER THEOREM KEITH CONRAD We should thank the Chnese for ther wonderful remander theorem. Glenn Stevens 1. Introducton The Chnese remander theorem says we can unquely solve any par of
More informationTR/01/89 February An O(h 6 ) cubic spline interpolating procedure for harmonic functions. N. Papamichael and Maria Joana Soares*
TR/0/89 February 989 An O( 6 cubc splne nterpolatng procedure for armonc functons N. Papamcael Mara Joana Soares* *Área de Matematca, Unversdade do Mno, 4700 Braga, Portugal. z 6393 ABSTRACT An O( 6 metod
More informationMA 323 Geometric Modelling Course Notes: Day 13 Bezier Curves & Bernstein Polynomials
MA 323 Geometrc Modellng Course Notes: Day 13 Bezer Curves & Bernsten Polynomals Davd L. Fnn Over the past few days, we have looked at de Casteljau s algorthm for generatng a polynomal curve, and we have
More informationModelli Clamfim Equazione del Calore Lezione ottobre 2014
CLAMFIM Bologna Modell 1 @ Clamfm Equazone del Calore Lezone 17 15 ottobre 2014 professor Danele Rtell danele.rtell@unbo.t 1/24? Convoluton The convoluton of two functons g(t) and f(t) s the functon (g
More informationDeriving the X-Z Identity from Auxiliary Space Method
Dervng the X-Z Identty from Auxlary Space Method Long Chen Department of Mathematcs, Unversty of Calforna at Irvne, Irvne, CA 92697 chenlong@math.uc.edu 1 Iteratve Methods In ths paper we dscuss teratve
More informationA PROCEDURE FOR SIMULATING THE NONLINEAR CONDUCTION HEAT TRANSFER IN A BODY WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY.
Proceedngs of the th Brazlan Congress of Thermal Scences and Engneerng -- ENCIT 006 Braz. Soc. of Mechancal Scences and Engneerng -- ABCM, Curtba, Brazl,- Dec. 5-8, 006 A PROCEDURE FOR SIMULATING THE NONLINEAR
More informationMath 217 Fall 2013 Homework 2 Solutions
Math 17 Fall 013 Homework Solutons Due Thursday Sept. 6, 013 5pm Ths homework conssts of 6 problems of 5 ponts each. The total s 30. You need to fully justfy your answer prove that your functon ndeed has
More informationFoundations of Arithmetic
Foundatons of Arthmetc Notaton We shall denote the sum and product of numbers n the usual notaton as a 2 + a 2 + a 3 + + a = a, a 1 a 2 a 3 a = a The notaton a b means a dvdes b,.e. ac = b where c s an
More informationU.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016
U.C. Berkeley CS94: Spectral Methods and Expanders Handout 8 Luca Trevsan February 7, 06 Lecture 8: Spectral Algorthms Wrap-up In whch we talk about even more generalzatons of Cheeger s nequaltes, and
More informationbounds, but Mao [{4] only dscussed te mean square (te case of p = ) almost sure exponental stablty. Due to te new tecnques developed n ts paper, te re
Asymptotc Propertes of Neutral Stocastc Derental Delay Equatons Xuerong Mao Department of Statstcs Modellng Scence Unversty of Stratclyde Glasgow G XH, Scotl, U.K. Abstract: Ts paper dscusses asymptotc
More informationSTAT 309: MATHEMATICAL COMPUTATIONS I FALL 2018 LECTURE 16
STAT 39: MATHEMATICAL COMPUTATIONS I FALL 218 LECTURE 16 1 why teratve methods f we have a lnear system Ax = b where A s very, very large but s ether sparse or structured (eg, banded, Toepltz, banded plus
More informationMAT 578 Functional Analysis
MAT 578 Functonal Analyss John Qugg Fall 2008 Locally convex spaces revsed September 6, 2008 Ths secton establshes the fundamental propertes of locally convex spaces. Acknowledgment: although I wrote these
More informationAnother converse of Jensen s inequality
Another converse of Jensen s nequalty Slavko Smc Abstract. We gve the best possble global bounds for a form of dscrete Jensen s nequalty. By some examples ts frutfulness s shown. 1. Introducton Throughout
More informationLecture 17 : Stochastic Processes II
: Stochastc Processes II 1 Contnuous-tme stochastc process So far we have studed dscrete-tme stochastc processes. We studed the concept of Makov chans and martngales, tme seres analyss, and regresson analyss
More informationProblem Set 4: Sketch of Solutions
Problem Set 4: Sketc of Solutons Informaton Economcs (Ec 55) George Georgads Due n class or by e-mal to quel@bu.edu at :30, Monday, December 8 Problem. Screenng A monopolst can produce a good n dfferent
More informationDECOUPLING THEORY HW2
8.8 DECOUPLIG THEORY HW2 DOGHAO WAG DATE:OCT. 3 207 Problem We shall start by reformulatng the problem. Denote by δ S n the delta functon that s evenly dstrbuted at the n ) dmensonal unt sphere. As a temporal
More informationLecture 4. Instructor: Haipeng Luo
Lecture 4 Instructor: Hapeng Luo In the followng lectures, we focus on the expert problem and study more adaptve algorthms. Although Hedge s proven to be worst-case optmal, one may wonder how well t would
More informationA discontinuous Galerkin method for nonlinear parabolic equations and. gradient flow problems with interaction potentials.
A dscontnuous Galerkn metod for nonlnear parabolc equatons and gradent flow problems wt nteracton potentals Zeng Sun, José A. Carrllo and C-Wang Su 3 Abstract We consder a class of tme dependent second
More information, rst we solve te PDE's L ad L ad n g g (x) = ; = ; ; ; n () (x) = () Ten, we nd te uncton (x), te lnearzng eedbac and coordnates transormaton are gve
Freedom n Coordnates Transormaton or Exact Lnearzaton and ts Applcaton to Transent Beavor Improvement Kenj Fujmoto and Tosaru Suge Dvson o Appled Systems Scence, Kyoto Unversty, Uj, Kyoto, Japan suge@robotuassyoto-uacjp
More informationModule 3: Element Properties Lecture 1: Natural Coordinates
Module 3: Element Propertes Lecture : Natural Coordnates Natural coordnate system s bascally a local coordnate system whch allows the specfcaton of a pont wthn the element by a set of dmensonless numbers
More information2.3 Nilpotent endomorphisms
s a block dagonal matrx, wth A Mat dm U (C) In fact, we can assume that B = B 1 B k, wth B an ordered bass of U, and that A = [f U ] B, where f U : U U s the restrcton of f to U 40 23 Nlpotent endomorphsms
More information2.29 Numerical Fluid Mechanics Fall 2011 Lecture 12
REVIEW Lecture 11: 2.29 Numercal Flud Mechancs Fall 2011 Lecture 12 End of (Lnear) Algebrac Systems Gradent Methods Krylov Subspace Methods Precondtonng of Ax=b FINITE DIFFERENCES Classfcaton of Partal
More informationUNIVERSITY OF BOLTON RAK ACADEMIC CENTRE BENG(HONS) MECHANICAL ENGINEERING SEMESTER TWO EXAMINATION 2017/2018 FINITE ELEMENT AND DIFFERENCE SOLUTIONS
OCD0 UNIVERSITY OF BOLTON RAK ACADEMIC CENTRE BENG(HONS) MECHANICAL ENGINEERING SEMESTER TWO EXAMINATION 07/08 FINITE ELEMENT AND DIFFERENCE SOLUTIONS MODULE NO. AME6006 Date: Wednesda 0 Ma 08 Tme: 0:00
More informationChapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems
Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons
More informationWaveguides and resonant cavities
Wavegudes and resonant cavtes February 8, 014 Essentally, a wavegude s a conductng tube of unform cross-secton and a cavty s a wavegude wth end caps. The dmensons of the gude or cavty are chosen to transmt,
More informationMAE140 - Linear Circuits - Winter 16 Final, March 16, 2016
ME140 - Lnear rcuts - Wnter 16 Fnal, March 16, 2016 Instructons () The exam s open book. You may use your class notes and textbook. You may use a hand calculator wth no communcaton capabltes. () You have
More information