HAL Id: hal-00471938
https://hal.archives-ouvertes.fr/hal-00471938
Submitted on 9 Apr 2010
HAL
is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or
L’archive ouverte pluridisciplinaire
HAL, estdestinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires
Numerical analysis of the planewave discretization of some orbital-free and Kohn-Sham models
Eric Cancès, Rachida Chakir, Yvon Maday
To cite this version:
Eric Cancès, Rachida Chakir, Yvon Maday. Numerical analysis of the planewave discretization of
some orbital-free and Kohn-Sham models. ESAIM: Mathematical Modelling and Numerical Analysis,
EDP Sciences, 2012, 46 (02), pp.341-388. �10.1051/m2an/2011038�. �hal-00471938�
Numerical analysis of the planewave discretization of some orbital-free and Kohn-Sham models
Eric Canc` es
∗, Rachida Chakir
†and Yvon Maday
†‡April 8, 2010
Abstract
We providea priorierror estimates for the spectral and pseudospectral Fourier (also called planewave) discretizations of the periodic Thomas- Fermi-von Weizs¨acker (TFW) model and for the spectral discretization of the Kohn-Sham model, within the local density approximation (LDA).
These models allow to compute approximations of the ground state energy and density of molecular systems in the condensed phase. The TFW model is stricly convex with respect to the electronic density, and allows for a comprehensive analysis. This is not the case for the Kohn-Sham LDA model, for which the uniqueness of the ground state electronic density is not guaranteed. Under a coercivity assumption on the second order optimality condition, we prove that for large enough energy cut-offs, the discretized Kohn-Sham LDA problem has a minimizer in the vicinity of any Kohn-Sham ground state, and that this minimizer is unique up to unitary transform. We then derive optimal a priori error estimates for the spectral discretization method.
1 Introduction
Density Functional Theory (DFT) is a powerful method for computing ground state electronic energies and densities in quantum chemistry, materials science, molecular biology and nanosciences. The models originating from DFT can be classified into two categories: the orbital-free models and the Kohn-Sham models. The Thomas-Fermi-von Weizs¨acker (TFW) model falls into the first
∗Universit´e Paris-Est, CERMICS, Project-team Micmac, INRIA-Ecole des Ponts, 6 & 8 avenue Blaise Pascal, 77455 Marne-la-Vall´ee Cedex 2, France.
†UPMC Univ Paris 06, UMR 7598 LJLL, Paris, F-75005 France ; CNRS, UMR 7598 LJLL, Paris, F-75005 France
‡Division of Applied Mathematics, Brown University, Providence, RI, USA
category. It is not very much used in practice, but is interesting from a math- ematical viewpoint [1, 7, 12]. It indeed serves as a toy model for the analysis of the more complex electronic structure models routinely used by Physicists and Chemists. At the other extremity of the spectrum, the Kohn-Sham models [8, 11] are among the most widely used models in Physics and Chemistry, but are much more difficult to deal with. We focus here on the numerical analysis of the TFW model on the one hand, and of the Kohn-Sham model, within the local density approximation (LDA), on the other hand. More precisely, we are interested in the spectral and pseudospectral Fourier, more commonly called planewave, discretizations of the periodic versions of these two models. In this context, the simulation domain, sometimes referred to as the supercell, is the unit cell of some periodic lattice ofR3. In the TFW model, periodic boundary conditions (PBC) are imposed to the density; in the Kohn-Sham framework, they are imposed to the Kohn-Sham orbitals (Born-von Karman PBC). Impos- ing PBC at the boundary of the simulation cell is a standard method to compute condensed phase properties with a limited number of atoms in the simulation cell, hence at a moderate computational cost.
This article is organized as follows. In Section 2, we briefly introduce the func- tional setting used in the formulation and the analysis of the planewave dis- cretization of orbital-free and Kohn-Sham models. In Section 3, we provide a priori error estimates for the planewave discretization of the TFW model, in- cluding numerical integration. In Section 4, we deal with the Kohn-Sham LDA model.
2 Basic Fourier analysis for planewave discretiza- tion methods
Throughout this article, we denote by Γ the simulation cell, byRthe periodic lattice, and by R∗ the dual lattice. For simplicity, we assume that Γ = [0, L)3 (L > 0), in which case R is the cubic lattice LZ3, and R∗ = 2πLZ3. Our arguments can be easily extended to the general case. For k∈ R∗, we denote byek(x) =|Γ|−1/2eik·x the planewave with wavevectork. The family (ek)k∈R∗ forms an orthonormal basis of
L2#(Γ,C) :=
u∈L2loc(R3,C)|uR-periodic , and for allu∈L2#(Γ,C),
u(x) = X
k∈R∗
b
ukek(x) with buk = (ek, u)L2
# =|Γ|−1/2 Z
Γ
u(x)e−ik·xdx.
In our analysis, we will mainly consider real valued functions. We therefore introduce the Sobolev spaces of real valued R-periodic functions
H#s(Γ) :=
(
u(x) = X
k∈R∗
b
ukek(x)| X
k∈R∗
(1 +|k|2)s|buk|2<∞and∀k, bu−k=ub∗k )
, s∈R(here and in the sequela∗ denotes the complex conjugate of the complex number a), endowed with the inner products
(u, v)Hs#= X
k∈R∗
(1 +|k|2)sbu∗kbvk. ForNc∈N, we denote by
VNc =
X
k∈R∗| |k|≤2πLNc
ckek| ∀k, c−k=c∗k
(1) (the constraintsc−k=c∗k imply that the functions ofVNc are real valued). For all s ∈ R, and each v ∈ H#s(Γ), the best approximation of v in VNc for any H#r-norm,r≤s, is
ΠNcv= X
k∈R∗| |k|≤2πLNc
b vkek.
The more regular v (the regularity being measured in terms of the Sobolev normsHr), the faster the convergence of this truncated series tov: for all real numbers randswithr≤s, we have for eachv∈H#s(Γ),
kv−ΠNcvkH#r = min
vNc∈VNckv−vNckH#r ≤ L
2π s−r
Nc−(s−r)kv−ΠNcvkHs#
≤ L
2π s−r
Nc−(s−r)kvkHs#. (2) ForNg ∈N\ {0}, we denote byφbFFT,Ng the discrete Fourier transform on the carterisan gridGNg :=NLgZ3 of the functionφ∈C#0(Γ,C), where
C#0(Γ,C) :=
u∈C0(R3,C)|uR-periodic . Recall that ifφ=P
k∈R∗φbkek∈C#0(Γ,C), the discrete Fourier transform ofφ is theNgR∗-periodic sequenceφbFFT,Ng = (φbFFT,Nk g)k∈R∗ where
φbFFT,Nk g = 1 Ng3
X
x∈GNg∩Γ
φ(x)e−ik·x=|Γ|−1/2 X
K∈R∗
φbk+NgK.
We now introduce the subspaces
WN1Dg =
Span
eily|l∈ 2π
LZ, |l| ≤ 2π L
Ng−1 2
(Ng odd), Span
eily|l∈ 2π
LZ, |l| ≤ 2π L
Ng
2
⊕C(eiπNgy/L+e−iπNgy/L) (Ng even),
(WN1Dg ∈C#∞([0, L),C) and dim(WN1Dg) =Ng), andWN3Dg =WN1Dg ⊗WN1Dg ⊗WN1Dg. Note thatWN3Dg is a subspace ofH#s(Γ,C) of dimensionNg3, for alls∈R, and that ifNg is odd,
WN3Dg = Span
ek|k∈ R∗=2π
LZ3, |k|∞≤2π L
Ng−1 2
(Ng odd).
It is then possible to define the interpolation projectorINg fromC#0(Γ,C) onto WN3Dg by [INg(φ)](x) =φ(x) for all x∈ GNg. It holds
∀φ∈C#0(Γ,C), Z
ΓINg(φ) = X
x∈GNg∩Γ
L Ng
3
φ(x). (3)
The coefficients of the expansion of INg(φ) in the canonical basis of WN3Dg is given by the discrete Fourier transform ofφ. In particular, whenNg is odd, we have the simple relation
INg(φ) =|Γ|1/2 X
k∈R∗| |k|∞≤2πL“Ng−1
2
”
φbFFT,Nk gek (Ng odd).
It is easy to check that ifφis real-valued, then so is INg(φ).
We will assume in the sequel that Ng ≥ 4Nc+ 1. We will then have for all v4Nc∈V4Nc,
Z
Γ
v4Nc = X
x∈GNg∩Γ
L Ng
3
v4Nc(x) = Z
ΓINg(v4Nc). (4) The following lemma gathers some technical results which will be useful for the numerical analysis of the planewave discretization of orbital-free and Kohn- Sham models.
Lemma 2.1 Let Nc∈N∗ andNg ∈N∗ such thatNg≥4Nc+ 1.
1. LetV be a function ofC#0(Γ,C)andvNcandwNcbe two functions ofVNc.
Then Z
ΓINg(V vNcwNc) = Z
ΓINg(V)vNcwNc; (5)
Z
ΓINg(V|vNc|2)
≤ kVkL∞kvNck2L2#. (6) 2. Lets >3/2,0≤r≤s, andV a function ofH#s(Γ). Then,
(1− INg)(V)Hr
# ≤ Cr,sNg−(s−r)kVkH#s; (7) Π2Nc(INg(V))
L2# ≤ Z
ΓINg(|V|2) 1/2
; (8)
Π2Nc(INg(V))
H#s ≤ (1 +Cs,s)kVkHs#, (9)
for constantsCr,s independent ofV. Besides if there existsm >3andC∈ R+ such that|Vbk| ≤C|k|−m, then there exists a constantCV independent ofNc andNg such that
Π2Nc(1− INg)(V)Hr
# ≤ CVNcr+3/2Ng−m. (10) 3. Let φ be a Borel function from R+ toR such that there exists Cφ ∈R+
for which|φ(t)| ≤Cφ(1 +t2)for allt∈R+. Then, for allvNc ∈VNc,
Z
ΓINg(φ(|vNc|2))
≤ Cφ
|Γ|+kvNck4L4#
. (11)
Proof Forz2Nc ∈V2Nc, it holds Z
ΓINg(V z2Nc) = X
x∈GNg∩Γ
L Ng
3
V(x)z2Nc(x)
= X
x∈GNg∩Γ
L Ng
3
(INg(V))(x)z2Nc(x)
= Z
ΓINg(V)z2Nc (12)
sinceINg(V)z2Nc∈VNg+2Nc⊂V2Ng is exactly integrated. The functionvNcwNc
being inV2Nc, (5) is proved. Moreover, as|vNc|2∈V4Nc, it follows from (4) that
Z
ΓINg(V|vNc|2) =
X
x∈GNg∩Γ
L Ng
3
V(x)|vNc(x)|2
≤ kVkL∞
X
x∈GNg∩Γ
L Ng
3
|vNc(x)|2
= kVkL∞ Z
Γ|vNc|2.
Hence (6). The estimate (7) is proved in [6]. To prove (8), we notice that kΠ2Nc(INg(V))k2L2# ≤ kINg(V)k2L2#
= Z
Γ
(INg(V))∗(INg(V))
= X
x∈GNg∩Γ
L Ng
3
(INg(V))(x)∗(INg(V))(x)
= X
x∈GNg∩Γ
L Ng
3
|V(x)|2
= Z
Γ
INg(|V|2).
The bound (9) is a straightforward consequence of (7):
kΠ2Nc(INg(V))kH#s ≤ kINg(V)kHs#≤ kVkH#s +k(1−INg)(V)kH#s ≤(1 +Cs,s)kVkH#s. Now, we notice that
Π2Nc(INg(V)) = |Γ|1/2 X
k∈R∗| |k|≤4πLNc
VbkFFT,Ngek
= X
k∈R∗| |k|≤4πLNc
X
K∈R∗
Vbk+NgK
!
ek. (13) From (13), we obtain
Π2Nc(1− INg)(V)2Hs
# = X
k∈R∗| |k|≤4πLNc
(1 +|k|2)s
X
K∈R∗\{0}
Vbk+NgK
2
≤
X
k∈R∗| |k|≤4πLNc
(1 +|k|2)s
max
k∈R∗| |k|≤4πLNc
X
K∈R∗\{0}
Vbk+NgK
2
.
On the one hand, X
k∈R∗| |k|≤4πLNc
(1 +|k|2)s ∼
Nc→∞
32π 2s+ 3
4π L
2s
Nc2s+3, and on the other hand, we have for eachk∈ R∗ such that|k| ≤ 4πLNc,
X
K∈R∗\{0}
Vbk+NgK
≤ C X
K∈R∗\{0}
1
|k+NgK|m
≤ C C0
L 2π
m
Ng−m where
C0= max
y∈R3| |y|≤1/2
X
K∈Z3\{0}
1
|y−K|m.
The estimate (10) then easily follows. Let us finally prove (11). Using (3) and (4), we have
Z
ΓINg(φ(|vNc|2)) =
X
x∈GNg∩Γ
L Ng
3
φ(|vNc(x)|2)
≤ Cφ
X
x∈GNg∩Γ
L Ng
3
(1 +|vNc(x)|4)
= Cφ
Z
Γ
(1 +|vNc|4) =Cφ
|Γ|+kvNck4L4#
.
This completes the proof of Lemma 2.1.
3 Planewave approximation of the TFW model
In the TFW model, as well as in any orbital-free model, the ground state elec- tronic density of the system is obtained by minimizing an explicit functional of the density. Denoting by N the number of electrons in the simulation cell and by
RN =
ρ≥0|√ρ∈H#1(Γ), Z
Γ
ρ=N
the set of admissible densities, the TFW problem reads ITFW= inf
ETFW(ρ), ρ∈RN , (14)
where
ETFW(ρ) = CW
2 Z
Γ|∇√ρ|2+CTF
Z
Γ
ρ5/3+ Z
Γ
ρVion+1
2DΓ(ρ, ρ).
CWis a positive real number (CW= 1, 1/5 or 1/9 depending on the context [8]), and CTF is the Thomas-Fermi constant: CTF = 103(3π2)2/3. The last term of the TFW energy models the periodic Coulomb energy: forρandρ′ inH#−1(Γ),
DΓ(ρ, ρ′) := 4π X
k∈R∗\{0}
|k|−2ρb∗kρb′k.
We finally make the assumption thatVion is aR-periodic potential such that
∃m >3, C ≥0 s.t. ∀k∈ R∗, |Vbkion| ≤C|k|−m. (15) Note that this implies thatVionis inHm−3/2−ǫ(Γ) for allǫ >0, hence inC#0(Γ) sincem−3/2−ǫ >3/2 forǫsmall enough. It is convenient to reformulate the TFW model in terms of v=√ρ. It can be easily seen that
ITFW = inf
ETFW(v), v∈H#1(Γ), Z
Γ|v|2=N
, (16)
where
ETFW(v) = CW
2 Z
Γ|∇v|2+CTF
Z
Γ|v|10/3+ Z
Γ
Vion|v|2+1
2DΓ(|v|2,|v|2).
Let F(t) = CTFt5/3 and f(t) = F′(t) = 53CTFt2/3. The function F is in C1([0,+∞))∩C∞((0,+∞)), is strictly convex on [0,+∞), and for all (t1, t2)∈ R+×R+,
|f(t22)t2−f(t21)t2−2f′(t21)t21(t2−t1)| ≤ 70
27CTFmax(t1/31 , t1/32 )|t2−t1|2. (17)
The first and second derivatives ofETFW are respectively given by hETFW′(v), wiH#−1,H#1 = 2hHTFW|v|2 v, wi;
hETFW′′(v)w1, w2iH#−1,H#1 = 2hH|v|TFW2 w1, w2i+ 4DΓ(vw1, vw2) + 4 Z
Γ
f′(|v|2)|v|2w1w2, where we have denoted by HTFWρ the TFW Hamiltonian associated with the
densityρ
HTFWρ =−CW
2 ∆ +f(ρ) +Vion+VρCoulomb, where
VρCoulomb(x) = 4π X
k∈R∗\{0}
|k|−2bρkek(x)
is theR-periodic Coulomb potential generated by theR-periodic charge distri- bution ρ. Recall that VρCoulomb can also be defined as the unique solution in H#1(Γ) to
−∆VρCoulomb= 4π
ρ− |Γ|−1 Z
Γ
ρ Z
Γ
VρCoulomb= 0.
Let us recall (see [12] and the proof of Lemma 2 in [3]) that
• (14) has a unique minimizerρ0, and that the minimizers of (16) areuand
−u, whereu=p ρ0;
• uis inH#m+1/2−ǫ(Γ) for eachǫ >0 (hence inC#2(Γ) sincem+1/2−ǫ >7/2 forǫsmall enough);
• u >0 onR3;
• usatisfies the Euler equation HTFW|u|2 (u) =−CW
2 ∆u+ 5
3CTFu4/3+Vion+VuCoulomb2
u=λu for some λ ∈ R, (the ground state eigenvalue of HTFWρ0 , that is non- degenerate).
The planewave discretization of the TFW model is obtained by choosing 1. an energy cut-off Ec > 0 or, equivalently, a finite dimensional Fourier
spaceVNc, the integerNc being related toEc through the relationNc :=
[√
2EcL/2π];
2. a cartesian grid GNg with step size L/Ng where Ng ∈ N∗ is such that Ng≥4Nc+ 1,
and by considering the finite dimensional minimization problem INTFWc,Ng = inf
ETFWNg (vNc), vNc∈VNc, Z
Γ|vNc|2=N
, (18) where
ENTFWg (vNc) = CW
2 Z
Γ|∇vNc|2+CTF
Z
ΓINg(|vNc|10/3) + Z
ΓINg(Vion)|vNc|2 +1
2DΓ(|vNc|2,|vNc|2),
INg denoting the interpolation operator introduced in the previous section. The Euler equation associated with (18) can be written as a nonlinear eigenvalue problem
∀vNc∈VNc, h(HeTFW,N|uNc,Ngg|2−λNc,Ng)uNc,Ng, vNciH#−1,H#1 = 0, where we have denoted by
HeTFW,Nρ g =−CW
2 ∆ +INg
5
3CTFρ2/3+Vion
+VρCoulomb
the pseudospectral TFW Hamiltonian associated with the density ρ, and by λNc,Ng the Lagrange multiplier of the constraintR
Γ|vNc|2 =N. We therefore have
−CW
2 ∆uNc,Ng+ΠNc
INg
5
3CTF|uNc,Ng|4/3+Vion
+V|uCoulombNc,Ng|2
uNc,Ng
=λNc,NguNc,Ng. Under the condition thatNg≥4Nc+ 1, we have for allφ∈C#0(Γ),
∀(k, l)∈ R∗× R∗ s.t. |k|,|l| ≤ 2π LNc,
Z
ΓINg(φ)e∗kel=φbFFTk−l, so that,HeTFWuNc,Ng is defined onVNc by the Fourier matrix
[HbTFW,N|uNc,Ngg|2]kl = CW
2 |k|2δkl+5
3CTF \
(|uNc,Ng|4/3)
FFT,Ng
k−l +(V\ion)FFT,Nk−l g +4π
(|u\Nc,Ng|2)FFT,Nk−l g
|k−l|2 (1−δkl),
where, by convention, the last term of the right hand side is equal to zero for k=l.
We also introduce the variational approximation of (16) INTFWc = inf
ETFW(vNc), vNc∈VNc, Z
Γ|vNc|2=N
. (19)
Any minimizeruNc to (19) satisfies the elliptic equation
−CW
2 ∆uNc+ ΠNc
5
3CTF|uNc|4/3uNc+VionuNc+V|uCoulomb
Nc|2 uNc
=λNcuNc, (20) for someλNc ∈R.
The main result of this section is an extension of results previously obtained by A. Zhou [16].
Theorem 3.1 For each Nc ∈ N, we denote byuNc a minimizer to (19) such that (uNc, u)L2
# ≥ 0 and, for each Nc ∈ N and Ng ≥ 4Nc + 1, we denote by uNc,Ng a minimizer to (18) such that (uNc,Ng, u)L2
# ≥ 0. Then for Nc large enough,uNc anduNc,Ng are unique, and the following estimates hold true
kuNc−ukH#s ≤ Cs,ǫN−(m−s+1/2−ǫ)
c ; (21)
|λNc−λ| ≤ CǫNc−(2m−1−ǫ); (22) γkuNc−uk2H#1 ≤INTFWc −ITFW ≤ CkuNc−uk2H#1; (23) kuNc,Ng−uNckH#s ≤ CsNc3/2+(s−1)+Ng−m; (24)
|λNc,Ng−λNc| ≤ CNc3/2Ng−m; (25)
|INTFWc,Ng −INTFWc | ≤ CNc3/2Ng−m, (26) for all −m+ 3/2 < s < m+ 1/2 and ǫ > 0, and for some constants γ > 0, Cs,ǫ≥0,Cǫ≥0,C≥0 andCs≥0 independent ofNc andNg.
Remark 1 More complex orbital-free models have been proposed in the recent years [15], which are used to perform multimillion atom DFT calculations. Some of these models however are not well posed (the energy functional is not bounded from below [2]), and the others are not well understood from a mathematical point of view. For these reasons, we will not deal with those models in this article.
3.1 A priori estimates for the variational approximation.
In this section, we prove the first part of Theorem 3.1, related to the variational approximation (19). The estimates (21), (22) and (23) originate from arguments already introduced in [3]. For brevity, we only recall the main steps of the proof and leave the details to the reader.
The difference between (16) and the problem dealt with in [3] is the presence of the Coulomb termDΓ(|v|2,|v|2), for which the following estimates are available:
0≤DΓ(ρ, ρ) ≤ Ckρk2L2#, for allρ∈L2#(Γ), (27)
|DΓ(uv, uw)| ≤ CkvkL2#kwkL2#, for all (v, w)∈(L2#(Γ))2, (28)
|DΓ(ρ, vw)| ≤ CkρkL2#kvkL2#kwkL2#, for all (ρ, v, w)∈(L2#(Γ))3, (29) kVρCoulombkL∞ ≤ CkρkL2#, for allρ∈L2#(Γ), (30) kVρCoulombkH#s+2 ≤ CkρkH#s, for allρ∈H#s(Γ). (31) Here and in the sequel, C denotes a non-negative constant which may depend on Γ, VionandN, but not on the discretization parameters.
Using (27), (28) and the fact thatf′>0 on (0,+∞), we can then show (see the proof of Lemma 1 in [3]) that there existβ >0,γ >0 andM ≥0 such that for allv∈H#1(Γ),
0≤ h(HTFWρ0 −λ)v, viH−#1,H1#≤Mkvk2H#1, (32) βkvk2H#1 ≤ h(ETFW′′(u)−2λ)v, viH#−1,H#1 ≤Mkvk2H#1, (33) and for allv∈H#1(Γ) such thatkvkL2# =N1/2 and (v, u)L2#≥0,
γkv−uk2H#1 ≤ h(HTFWρ0 −λ)(v−u),(v−u)iH#−1,H#1. (34) Remarking that
ETFW(uNc)−ETFW(u) = h(HρTFW0 −λ)(uNc−u),(uNc−u)iH#−1,H1#
+1
2DΓ(|uNc|2− |u|2,|uNc|2− |u|2) +
Z
Γ
F(|uNc|2)−F(|u|2)−f(|u|2)(|uNc|2− |u|2) (35) and using (34), the positivity of the bilinear formDΓ, and the convexity of the functionF, we obtain that
INTFWc −ITFW=ETFW(uNc)−ETFW(u)≥γkuNc−uk2H1#.
For each Nc ∈ N, euNc = N1/2ΠNcu/kΠNcukL2# satisfies (euNc, u)L2# ≥ 0 and keuNckL2# =N1/2, and the sequence (ueNc)Nc∈Nconverges to uinH#m+1/2−ǫ(Γ) for each ǫ >0. As the functionalETFW is continuous onH#1(Γ), we have
kuNc−uk2H#1 ≤γ−1 INTFWc −ITFW
≤γ−1 ETFW(euNc)−ETFW(u)
N−→c→∞0.
Hence, (uNc)Nc∈Nconverges touinH#1(Γ), and we also have λNc = N−1
CW
2 Z
Γ|∇uNc|2+ Z
Γ
f(|uNc|2)|uNc|2+ Z
Γ
Vion|uNc|2+DΓ(|uNc|2,|uNc|2)
N−→c→∞ N−1 CW
2 Z
Γ|∇u|2+ Z
Γ
f(|u|2)|u|2+ Z
Γ
Vion|u|2+DΓ(|u|2,|u|2)
= λ.
As f(|uNc|2)uNc+VionuNc+V|uCoulomb
Nc|2 uNc is bounded inL2#(Γ), uniformly in Nc, we deduce from (20) that the sequence (uNc)Nc∈N is bounded in H#2(Γ), hence inL∞(Γ). Now
∆(uNc−u) = 2CW−1
ΠNc
f(|uNc|2)uNc−f(|u|2)u+Vion(uNc−u) + V|uCoulomb
Nc|2 uNc−V|u|Coulomb2 u
+ (1−ΠNc)
f(|u|2)u+Vionu+V|u|Coulomb2 u
−λNc(uNc−u)−(λNc−λ)u
.
Observing that the right-hand side goes to zero in L2#(Γ) when Nc goes to infinity, we obtain that (uNc)Nc∈N converges to u in H#2(Γ), and therefore in C#0,1/2(Γ). In addition, we know from Harnack inequality [10] thatu >0 inR3. Consequently, for Nc large enough, the function uNc (which is continuous and R-periodic) is bounded away from 0, uniformly inNc. Asf ∈C∞(0,+∞), one can see by a simple bootstrap argument that the convergence of (uNc)Nc∈Ntou also holds inH#m+1/2−ǫ(Γ) for eachǫ >0. The upper bound in (23) is obtained from (35), remarking that
0 ≤
Z
Γ
F(|uNc|2)−F(|u|2)−f(|u|2)(|uNc|2− |u|2)
≤ 35 9 CTF
Z
Γ
max(|uNc|4/3,|u|4/3)|uNc−u|2
≤ 35 9 CTF
Nmaxc∈NkuNckL∞ 4/3
kuNc−uk2L2#, and that
0 ≤ DΓ(|uNc|2− |u|2,|uNc|2− |u|2)≤Ck|uNc|2− |u|2k2L2#
≤ 4C
Nmaxc∈NkuNckL∞ 2
kuNc−uk2L2#.
The uniqueness ofuNc forNclarge enough can then be checked as follows. First,
(uNc, λNc) satisfies the variational equation
∀vNc∈VNc, h(HTFW|uNc|2−λNc)uNc, vNciH#−1,H#1 = 0.
Therefore λNc is the variational approximation in VNc of some eigenvalue of HTFW|uNc|2. As (uNc)Nc∈N converges to u in L∞(Γ), HTFW|uNc|2 − HTFWρ0 converges to 0 in operator norm. Consequently, the nth eigenvalue of HTFW|uNc|2 converges to thenth eigenvalue ofHTFWρ0 whenNc goes to infinity, the convergence being uniform inn. Together with the fact that the sequence (λNc)Nc∈Nconverges to λ, the non-degenerate ground state eigenvalue of HTFWρ0 , this implies that for Nc large enough,λNc is the ground state eigenvalue of HTFW|uNc|2 in VNc and for allvNc∈VNc such thatkvNckL2# =N1/2 and (vNc, uNc)L2# ≥0,
ETFW(vNc)−ETFW(uNc) = h(HTFW|uNc|2−λNc)(vNc−uNc),(vNc−uNc)iH#−1,H#1
+1
2DΓ(|vNc|2− |uNc|2,|vNc|2− |uNc|2) +
Z
Γ
F(|vNc|2)−F(|uNc|2)−f(|uNc|2)(|vNc|2− |uNc|2)
≥ h(HTFW|uNc|2−λNc)(vNc−uNc),(vNc−uNc)iH#−1,H#1
≥ γ
2kvNc−uNck2H#1. (36) It easily follows that forNclarge enough, (19) has a unique minimizeruNc such
that (uNc, u)L2
# ≥0.
Let us now establish the rates of convergence of |λNc−λ| and kuNc −ukH#s. First,
λNc−λ = N−1
h(HTFW|u|2 −λ)(uNc−u),(uNc−u)iH#−1,H#1
+ Z
Γ
wNc(uNc−u)
(37) with
wNc =f(|uNc|2)−f(|u|2)
uNc−u |uNc|2+V|uCoulomb
Nc|2 (uNc+u).
As uNc is bounded away from 0 and f ∈ C∞((0,+∞)), the function wNc is uniformly bounded inH#m−3/2−ǫ(Γ) (at least forNclarge enough). We therefore obtain that for all 0≤r < m−3/2, there exists a constantCr∈R+ such that for allNc large enough,
|λNc−λ| ≤Cr
kuNc−uk2H#1 +kuNc−ukH−#r
. (38)
In order to evaluate theH#1-norm of the error (uNc−u), we first notice that
∀vNc ∈VNc, kuNc−ukH1# ≤ kuNc−vNckH#1 +kvNc−ukH#1, (39)
and that
kuNc−vNck2H#1 ≤ β−1h(ETFW′′(u)−2λ)(uNc−vNc),(uNc−vNc)iH#−1,H#1
= β−1
h(ETFW′′(u)−2λ)(uNc−u),(uNc−vNc)iH#−1,H#1
+h(ETFW′′(u)−2λ)(u−vNc),(uNc−vNc)iH#−1,H#1
.(40) For allzNc∈VNc,
h(ETFW′′(u)−2λ)(uNc−u), zNciH#−1,H#1
= −2 Z
Γ
[f(|uNc|2)uNc−f(|u|2)uNc−2f′(|u|2)|u|2(uNc−u)]zNc
−2DΓ((uNc−u)(uNc+u),(uNc−u)zNc)−2DΓ((uNc−u)2, uzNc) +2(λNc−λ)
Z
Γ
uNczNc. (41)
On the other hand, we have for allvNc∈VNc such thatkvNckL2# =N1/2, Z
Γ
uNc(uNc−vNc) =N − Z
Γ
uNcvNc =1
2kuNc−vNck2L2#.
Using (17), (29), (38) with r= 0 and the above equality, we therefore obtain for allvNc ∈VNc such thatkvNckL2#=N1/2,
h(ETFW′′(u)−2λ)(uNc−u),(uNc−vNc)iH#−1,H#1
≤C
kuNc−uk2H#1 kuNc−vNckH#1
+
kuNc−uk2H#1 +kuNc−ukL2#
kuNc−vNck2L2#
. (42) Therefore, forNc large enough, we have for allvNc ∈VNc such thatkvNckL2# = N1/2,
kuNc−vNckH#1 ≤C
kuNc−uk2H#1 +kvNc−ukH#1
.
Together with (39), this shows that there exists N ∈NandC ∈R+ such that for allNc≥N,
∀vNc∈VNc s.t. kvNckL2#=N1/2, kuNc−ukH#1 ≤CkvNc−ukH#1. By a classical argument (see e.g. the proof of Theorem 1 in [3]), we deduce from (2) and the above inequality that
kuNc−ukH#1 ≤C min
vNc∈VNckvNc−ukH1#≤C1,ǫNc−(m−1/2−ǫ), (43)
for some constant C1,ǫ independent of Nc. This completes the proof of the estimate in theH#1–norm. We proceed with the analysis of theL2#–norm.
Forw∈L2#(Γ), we denote byψw the unique solution to the adjoint problem
( findψw∈u⊥ such that
∀v∈u⊥, h(ETFW′′(u)−2λ)ψw, viH#−1,H#1 =hw, viH−1# ,H#1, (44)
where
u⊥=
v∈H#1(Γ)| Z
Γ
uv= 0
.
The functionψw is solution to the elliptic equation
−CW
2 ∆ψw+ Vion+VuCoulomb2 +f(u2) + 2f′(u2)u2−λ
ψw+ 2VuψCoulombw u
= 2 Z
Γ
f′(u2)u3ψw+DΓ(u2, uψw)
u+1 2
w−(w, u)L2
#u ,
from which we deduce that if w ∈ H#r(Γ) for some 0 ≤ r < m−3/2, then ψw∈H#r+2(Γ) and
kψwkHr+2# ≤CrkwkH#r, (45)
for some constantCr independent ofw. Letu∗Nc be the orthogonal projection, for the L2#inner product, ofuNc on the affine spacen
v∈L2#(Γ)|R
Γuv=No . One has
u∗Nc∈H#1(Γ), u∗Nc−u∈u⊥, u∗Nc−uNc = 1
2NkuNc−uk2L2#u,