Density dependent embedding potentials for piecewise exact densities
Tomasz A. Wesolowski · 2025
Read Density dependent embedding potentials for piecewise exact densities by Tomasz A. Wesolowski (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-08744.txt.
Introduction
Tomasz Wesolowski tests a density embedding theory in a 2025 paper. The theory treats part of a system with full quantum care while the rest is frozen as a density backdrop. Its equations can be read as the outcome of a constrained energy search.
Wesolowski asks what happens when the frozen part matches the true ground density in some region. He shows the solved part then cannot match the true remainder: the joined energy sits strictly above the true ground energy. That sharp gap shapes how exact embedding strengths can be built. The result bears on subsystem methods and on fake-core models used in chemistry.
Catalog record
| CREATOR | Tomasz A. Wesolowski |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-08744
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Density dependent embedding potentials for piecewise exact densities Tomasz A. Wesolowski Abstract Frozen Density Embedding Theory (FDET) [Wesolowski Phys. Rev. A 77 , 012504 (2008)] provides the interpretation of the eigenvalue equations for an embedded N ′ N^{\prime} -electron wavefunction, in which the embedding operator is multiplicative, as the Euler-Lagrange equation corresponding to the constrained minimisation of the Hohenberg-Kohn energy functional. The constraint is given by a non-negative function integrating to an integer N − N ′ > 0 N-N^{\prime}>0 with N N being the total number of electrons in the whole system ( min ρ → ∀ 𝐫 ( ρ ( 𝐫 ) ≥ ρ 2 ( 𝐫 ) CLOSE E v H K [ ρ ] = E v H K [ ρ 1 o + ρ 2 ] ≥ E H K [ ρ v o ] = E v o \min_{\rho\rightarrow\forall_{\mathbf{r}}\big(\rho({\mathbf{r}})\geq\rho_{2}({\mathbf{r}}\big)}E^{HK}_{v}[\rho]=E^{HK}_{v}[\rho_{1}^{o}+\rho_{2}]\geq E^{HK}[\rho_{v}^{o}]=E^{o}_{v} ). The necessary condition that ρ 1 o ( 𝐫 ) = ρ v o ( 𝐫 ) − ρ 2 ( 𝐫 ) \rho_{1}^{o}({\mathbf{r}})=\rho_{v}^{o}({\mathbf{r}})-\rho_{2}({\mathbf{r}}) and consequently that E v H K [ ρ 1 o + ρ 2 ] = E v H K [ ρ v o ] = E v o E^{HK}_{v}[\rho_{1}^{o}+\rho_{2}]=E^{HK}_{v}[\rho_{v}^{o}]=E^{o}_{v} is obviously that ρ 2 ( 𝐫 ) ≤ ρ v o ( 𝐫 ) \rho_{2}({\mathbf{r}})\leq\rho_{v}^{o}({\mathbf{r}}) on any measurable volume element. In this work, we show that the equality ρ 1 o ( 𝐫 ) = ρ v o ( 𝐫 ) − ρ 2 ( 𝐫 ) \rho_{1}^{o}({\mathbf{r}})=\rho_{v}^{o}({\mathbf{r}})-\rho_{2}({\mathbf{r}}) imposes a stronger condition on ρ B \rho_{B} , namely, ρ 2 ( 𝐫 ) < ρ v o ( 𝐫 ) \rho_{2}({\mathbf{r}})<\rho_{v}^{o}({\mathbf{r}}) . Properties of the exact solution of the FDET eigenvalue equations are derived if this condition is not met. The result is discussed in the context of subsystem approach to the Hohenberg-Kohn theorems, pseudopotential theory, and embedding potentials derived from inverted of Kohn-Sham equations. Preliminaries: definitions, notation, statement of the problem This work concerns the following minimisation problem: \displaystyle\min_{\rho\rightarrow\forall_{\mathbf{r}}\big(\rho({\mathbf{r}})\geq\rho_{2}({\mathbf{r}}\big)}E^{HK}_{v}[\rho]=\min_{\rho_{1}\rightarrow N^{\prime}}E^{HK}_{v}[\rho_{1}+\rho_{2}]=E^{HK}_{v}[\rho_{1}^{o}+\rho_{2}]\geq E^{o}_{v} (1) for the Hohenberg-Kohn functional E_{v}^{HK}[\rho] in a system of N -electrons in a Coulombic external potential v , for such \rho_{2} that \forall_{\mathbf{r})}\rho_{2}(\mathbf{r})\geq 0 , \int\rho_{2}(\mathbf{r})d\mathbf{r}=N-N^{\prime} with integer N . Throughout this work, \rho and \rho_{1} denote non-specific functions on which the considered functionals depend. Specific functions are indicated by additional subscripts or superscripts. All considerations are made for a given Coulombic external potential ( v ), total number of electrons ( N ), and the number of embedded electrons ( N^{\prime} ). All formulas are given in atomic units. Eq. 1 provides thus the definition of \rho_{1}^{o}[\rho_{2}] . Square brackets in \rho_{1}^{o}[\rho_{2}] indicate that \rho_{1}^{o} is uniquely determined by \rho_{2} for a given external potential v . Eq. 1 provides also the definition of the functional E^{\rho_{2}}_{v}[\rho_{1}] : \displaystyle E^{\rho_{2}}_{v}[\rho_{1}]\equiv E_{v}^{HK}[\rho_{1}+\rho_{2}], (2) which for a given potential v and the N-N^{\prime} -electron density \rho_{2} . The domain of admissible functions for E^{\rho_{2}}_{v}[\rho_{1}] comprises N^{\prime} -electron densities. \rho_{1}^{o} defined in Eq. 1 is a minimiser of E^{\rho_{2}}_{v}[\rho_{1}] and not of the Hohenberg-Kohn functional. Frozen-Density Embedding Theory (FDET) Wesołowski 2008 provides the formal framework to perform the above search for the minimiser of E^{\rho_{2}}_{v}[\rho_{1}] for given v and \rho_{2} by means of solving an eigenvalue problem for a smaller number of electrons than N . \displaystyle\left(\hat{T}^{N^{\prime}}+\hat{V}^{N^{\prime}}+\hat{V}_{ee}^{N^{\prime}}+\hat{V}_{emb}^{N^{\prime}}\right)\Psi^{N^{\prime}}=\lambda^{N^{\prime}}\cdot\Psi^{N^{\prime}}, (3) where \displaystyle\hat{V}_{ee}^{N^{\prime}} \displaystyle= \displaystyle\sum_{i=1}^{N^{\prime}}\sum_{j>i}^{N^{\prime}}\frac{1}{|\mathbf{r}_{i}-\mathbf{r}_{j}|}, (4) \displaystyle\hat{V}^{N^{\prime}} \displaystyle= \displaystyle\sum_{i=1}^{N^{\prime}}v(\mathbf{r})\delta(\mathbf{r}_{i}-\mathbf{r}), (5) \displaystyle\hat{T}^{N^{\prime}} \displaystyle= \displaystyle-\frac{1}{2}\sum_{i=1}^{N^{\prime}}\nabla^{2}_{\mathbf{r}_{i}}, (6) \displaystyle\hat{V}^{N^{\prime}}_{emb} \displaystyle= \displaystyle\sum_{i=1}^{N^{\prime}}v_{emb}(\mathbf{r})\delta(\mathbf{r}_{i}-\mathbf{r}). (7) In FDET, \hat{V}^{N^{\prime}}_{emb} is multiplicative. This restriction leads to specific mathematical issues concerning the solutions of Eq. 3 and their relations to the exact solution of the full N -electron problem, i.e. E_{v}^{o} for energy and \rho_{v}^{o} for density in the ground state. Some of them are discussed in the present work. FDET provides thus the interpretation of any equation of the form of Eq. 3 , in which the embedding potential depends on election densities, as the Euler-Lagrange equation for E^{\rho_{2}}_{v}[\rho_{1}] , in which \Psi^{N^{\prime}} is an auxiliary quantum descriptor for the N^{\prime} -electron system used to represent the density \rho_{1} as: \rho_{1}(\mathbf{r})=\langle\Psi^{N^{\prime}}|\sum_{i=1}^{N^{\prime}}\delta(\mathbf{r}_{i}-\mathbf{r})|\Psi^{N^{\prime}}\rangle . FDET provides the unique relation (up to an additive constant) between the potential v_{emb}(\mathbf{r}) in Eq. 3 , the embedded wavefunction obtained in this equation, and \rho_{2}(\mathbf{r}) . \displaystyle v_{emb}[\rho_{1}^{{Eq.\ref{eq:emb_SE}}},\rho_{2}](\mathbf{r})=\int\frac{\rho_{2}(\mathbf{r}^{\prime})}{|\mathbf{r}-\mathbf{r}^{\prime}|}d\mathbf{r}^{\prime}+\left.\frac{\delta E_{xct}^{nad}[\rho,\rho_{2}]}{\delta\rho(\mathbf{r})}\right|_{\rho=\rho_{1}^{{Eq.\ref{eq:emb_SE}}}}, (8) in which the second term is the functional derivative of the bi-functional E_{xct}^{nad}[\rho_{1},\rho_{2}] defined as: \displaystyle E_{xct}^{nad}[\rho_{1},\rho_{2}]\equiv E_{xct}[\rho_{1}+\rho_{2}]-E_{xct}[\rho_{1}]-E_{xct}[\rho_{2}], (9) where \displaystyle E_{xct}[\rho]\equiv F^{HK}[\rho]-\frac{1}{2}\int\int\frac{\rho(\mathbf{r})\rho(\mathbf{r}^{\prime})}{\left|\mathbf{r}-\mathbf{r}\right|}\mathrm{d}\mathbf{r}^{\prime}\mathrm{d}\mathbf{r}, (10) and where F^{HK}[\rho] is the universal Hohenberg-Kohn functional defined implicitly in the Levy’s constrained search Levy 1982 : \displaystyle F^{HK}[\rho] \displaystyle\equiv \displaystyle\min_{\Psi\longrightarrow\rho}\left<\Psi|\hat{T}+\hat{V}_{ee}|\Psi\right>=\left<\Psi_{o}[\rho]|\hat{T}+\hat{V}_{ee}|\Psi_{o}[\rho]\right> (11) \displaystyle= \displaystyle\left<\Psi_{o}[\rho]|\hat{T}|\Psi_{o}[\rho]\right>+\left<\Psi_{o}[\rho]|\hat{V}_{ee}|\Psi_{o}[\rho]\right>=T[\rho]+V_{ee}[\rho]. As shown in Ref. 1 , the minimiser defined in Eq. 1 and the density obtained from Eqs. 3 , 8 are the same: \displaystyle\rho_{1}^{Eqs.\;\ref{eq:emb_SE},\ref{eq:embpotwf}}[\rho_{2}]=\rho_{1}^{o}[\rho_{2}], (12) for any N-N^{\prime} -electron density \rho_{2} , provided \rho_{1}^{o}[\rho_{2}] is v -representable Parr and Yang 1989 . Only minima of E^{\rho_{2}}_{v}[\rho_{1}] can be obtained from the FDET eigenvalue equation. It is possible that, there exist such density \rho_{2} for which: \displaystyle\inf_{\rho\rightarrow\forall_{\mathbf{r}}\big(\rho({\mathbf{r}})\geq\rho_{2}({\mathbf{r}}\big)}E^{HK}_{v}[\rho]=E^{HK}_{v}[\rho_{1}^{lowest}[\rho_{2}]+\rho_{2}]\leq E^{HK}_{v}[\rho_{1}^{Eqs.\;\ref{eq:emb_SE},\ref{eq:embpotwf}}[\rho_{2}]+\rho_{2}]. (13) If such density \rho_{2} exists, the corresponding \rho_{1}^{lowest}[\rho_{2}] (defined in Eq. 13 ), is an infimum but not minimum of E^{\rho_{2}}_{v}[\rho_{1}] . It cannot be obtained from FDET therefore. This observation leads to questions addressed in the present work: • Are there any densities \rho_{2} , for which \rho_{1}^{lowest}[\rho_{2}] is not a minimum? • What is \rho_{1}^{Eqs.\;\ref{eq:emb_SE},\ref{eq:embpotwf}}[\rho_{2}] in such a case? Concerning the first issue, we identify a non-trivial class of N-N^{\prime} -electron densities \rho_{2} (denoted with \mathcal{A} below), for which that \rho_{1}^{lowest}[\rho_{2}] cannot be obtained from Eq. 3 . It is shown by proving that the functional derivative of E_{xct}^{nad}[\rho_{1},\rho_{2}] with respect to \rho_{1} does not exist for \rho_{1}=\rho_{v}^{lowest}[\rho_{2}] and \rho_{2}\in\mathcal{A} ( Lemma A ). Theorems I - II provide inequalities for density and energy obtained from FDET eigenvalue equations for \rho_{2}\in\mathcal{A} . The derived inequalities for energy and density are also given for the variant of FDET eigenvalue equation in which a non-interacting reference Hamiltonian is used instead of V_{ee} in Eq. 3 . Definition: class \mathcal{A} The N-N^{\prime} -electron density \rho_{2}(\mathbf{r}) belongs to \mathcal{A} if there exists a measurable volume element V_{z} such that: \displaystyle\forall_{\mathbf{r}\in V_{z}}\rho_{2}(\mathbf{r})=\rho_{v}^{o}(\mathbf{r}), (14) and \displaystyle\forall_{\mathbf{r}}\rho_{2}(\mathbf{r})\leq\rho_{v}^{o}(\mathbf{r}). (15) A density belonging to \mathcal{A} is shown schematically in Fig. 1 . Figure 1: Symbolic representation of the densities considered in this work. \rho_{2} belonging to \mathcal{A} is indicated with blue. The ground state density is indicated with black. Red indicates the sum of \rho_{2} and the stationary density obtained from Eq. 3 . The pair of densities \rho_{2}\in\mathcal{A} and the corresponding \rho_{1}^{lowest}[\rho_{2}] represent a particular partitioning of the exact ground-state density \rho_{v}^{o} as: \displaystyle\inf_{\rho\rightarrow\forall_{\mathbf{r}}\big(\rho({\mathbf{r}})\geq\rho_{2}({\mathbf{r}}\big)}E^{HK}_{v}[\rho]=E^{HK}_{v}[\overbrace{\rho_{1}^{lowest}[\rho_{2}]}^{=\rho_{v}^{o}-\rho_{2}}+\rho_{2}]=E^{HK}_{v}[\rho_{v}^{o}]=E_{v}^{o}\;\mathrm{for}\;\rho_{2}\in\mathcal{A}. (16) We point out that FDET is formulated for a more general class of densities \rho_{2} than the ones corresponding to some partitioning of \rho_{v}^{o} which includes \rho_{2}\in\mathcal{A} as a subset but includes also such N-N^{\prime} -electron densities \rho_{2} that \rho_{2}(\mathbf{r})>\rho_{v}^{o}(\mathbf{r}) on some measurable volume. FDET eigenvalue equation for \rho_{2}\in\mathcal{A} \rho_{1}^{Eqs.\;\ref{eq:emb_SE},\ref{eq:embpotwf}}[\rho_{2}]=\rho_{1}^{o}[\rho_{2}] only if the bi-functional v_{xct}^{nad}[\rho_{1},\rho_{2}] exists for \rho_{1}=\rho_{1}^{o}[\rho_{2}] . The general condition \displaystyle\forall_{\mathbf{r}}\rho_{1}[\rho_{2}]({\mathbf{r}})\geq 0 (17) is necessary that the potential v_{xct}^{nad}[\rho_{1}[\rho_{2}],\rho_{2}] exists. Obviously it is automatically satisfied by densities obtained from Eqs. 3 , 8