REVIEW 2 major objections 3 minor 19 references
Density dependent embedding potentials for piecewise exact densities
T0 review · 2 major / 3 minor · reviewed 2026-08-07 · deepseek-v4-flash
Pith's one-line read Exact frozen densities can never reproduce the true ground-state density.
desk verdict A concise and sound proof that exact FDET cannot reproduce the exact ground-state density when the frozen density is piecewise exact; the gaps are presentational, not substantive. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the FDET embedding potential, a multiplicative one-body potential defined as $v_{\mathrm{emb}}[\rho_1,\rho_2](\mathbf r) = \int \rho_2(\mathbf r')/|\mathbf r-\mathbf r'|\,d\mathbf r' + \delta E_{\mathrm{xct}}^{\mathrm{nad}}[\rho_1,\rho_2]/\delta\rho_1(\mathbf r)$, where $E_{\mathrm{xct}}^{\mathrm{nad}}$ is the non-additive exchange-correlation part of the Hohenberg-Kohn functional. The proof works by showing that the Gateaux differential defining this derivative is direction-dependent at $\rho_1 = \rho_v^o - \rho_2$ when $\rho_2\in\mathcal{A}$: choose a zero-integral perturbation $\Delta$ that is negative on the patch where $\rho_2 = \rho_v^o$, and the perturbed density $\rho_1 + h\Delta$ becomes negative there, leaving the functional's domain. Because one of the two limits in the differential fails to exist, the derivative does not exist, which forces the embedded density away from the exact remainder and yields the strict energy inequality.
What would settle it
Construct a Coulombic system and a frozen density equal to the exact ground-state density on a finite volume, solve the exact FDET eigenvalue equations with exact functionals, and check whether the resulting total density equals the exact ground-state density or whether the total energy equals the ground-state energy; the theorem says neither can happen.
Extended reading notes
Core claim
For a Coulombic $N$-electron system with ground-state density $\rho_v^o$, consider any frozen density $\rho_2$ that is non-negative, integrates to $N-N'$, never exceeds $\rho_v^o$, and equals $\rho_v^o$ on some measurable volume (class $\mathcal{A}$). The paper shows that the exact FDET embedding potential, being the functional derivative of the non-additive exchange-correlation bifunctional, is not defined at $\rho_1 = \rho_v^o - \rho_2$ (Lemma A). Therefore the stationary density $\rho_1^{\mathrm{Eqs. 3,8}}$ obtained from the embedded eigenvalue equation cannot equal $\rho_v^o - \rho_2$; instead it is strictly positive on some subvolume of the region where $\rho_2$ matches the exact density (Theorem I). The assembled total density $\rho_1^{\mathrm{Eqs. 3,8}} + \rho_2$ is not $\rho_v^o$, and by the Hohenberg-Kohn theorems its energy satisfies $E_v^{\mathrm{HK}}[\rho_1^{\mathrm{Eqs. 3,8}} + \rho_2] > E_v^o$ (Theorem II). The same conclusions hold for the non-interacting Kohn-Sham variant of FDET (Lemma B, Eqs. 26-27).
Load-bearing premise
The load-bearing assumption is that a Coulombic ground-state density is strictly positive everywhere except at infinity, so the density perturbation used in the proof stays admissible; if the true density could vanish on a finite region, the nonexistence argument for the embedding potential would no longer go through.
Editorial extensions
If this is right
- In any FDET calculation whose frozen density $\rho_2$ lies in class $\mathcal{A}$, the exact embedding equations cannot yield the exact total ground-state density; the total energy is necessarily above $E_v^o$ even with exact functionals.
- The condition $\rho_2(\mathbf r) \le \rho_v^o(\mathbf r)$ everywhere is necessary but not sufficient; reaching the exact density requires the stricter $\rho_2(\mathbf r) < \rho_v^o(\mathbf r)$.
- Reference embedding potentials constructed by partitioning a total density and inverting Kohn-Sham equations are illegitimate when the partition puts $\rho_2$ in class $\mathcal{A}$: the non-additive kinetic potential does not exist for that pair, not merely is non-unique.
- Embedding operators that are not multiplicative, such as the Phillips-Kleinman pseudopotential, are not blocked in this way, which locates the obstruction in the functional-derivative form of the FDET potential.
Reading between the lines
- Because the strict inequality follows from nondifferentiability at the exact remainder, the energy gap can presumably be made arbitrarily small by letting $\rho_2$ approach $\rho_v^o$ without ever equalling it on a measurable volume; how the gap scales with the size of the matching patch is a natural testable extension.
- The restriction to Coulombic potentials is doing real work: for a hypothetical system whose ground-state density vanishes on a finite region, the Gateaux argument would collapse, so the theorem's boundary is exactly the strict-positivity property of Coulombic densities.
- Practical approximate density-dependent embedding potentials with non-smooth derivatives near zero density may show convergence failures or spurious stationary solutions precisely where the frozen density saturates the total density; this is a testable prediction for numerical codes.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies Frozen Density Embedding Theory (FDET) for a frozen density rho2 that satisfies rho2 <= rho_v^o everywhere and rho2 = rho_v^o on a measurable set V_z (class A). The main result is that the FDET embedding potential v_nad, defined as the functional derivative of the nonadditive exchange-correlation bifunctional, does not exist at the density rho1 = rho_v^o - rho2 for such rho2 (Lemma A). The proof is a boundary-of-domain argument: because rho1 vanishes on V_z, a variation Delta that is negative on V_z makes rho1 + h Delta negative there for h > 0, so the Gateaux quotient for E_xct[rho1] is undefined. From this, the paper derives Theorem I, stating that any solution of the FDET eigenvalue equation (if it exists) must be positive on a subset of V_z, and Theorem II, stating that the corresponding total density differs from rho_v^o and hence has Hohenberg-Kohn energy strictly larger than E_v^o. An analogous lemma (Lemma B) is stated for the noninteracting kinetic-energy bifunctional in the Kohn-Sham variant of FDET, with proof omitted. The discussion places the result in the context of subsystem DFT, pseudopotential theory, and potentials obtained by inversion of Kohn-Sham equations, emphasizing that a 'lucky choice' of rho2 that is piecewise equal to the exact density still cannot lead to the exact total density within FDET.
Significance. The core mathematical observation is sound and clearly presented. The boundary-of-domain argument in Lemma A is simple, self-contained, and does not rely on fitting parameters or numerical approximations. If the stated assumptions are made explicit, the result is a useful rigorous limitation of exact FDET: it shows that a multiplicative density-dependent embedding potential cannot reproduce the exact ground-state density when the frozen density equals the exact density on a finite volume, contrary to a common silent assumption. The distinction from nonmultiplicative pseudopotentials is instructive. The main gaps are the missing nondegeneracy assumption in the energy inequality and the implicit generalization of Lemma A in Theorem I; both are fixable. The paper is concise and suitable for a formal journal after revision.
major comments (2)
- [Theorem II, Eq. (20)] The proof of the strict inequality uses the Hohenberg-Kohn theorem to infer that rho1^{Eqs. 3,8} + rho2 != rho_v^o implies E_HK^v[rho1^{Eqs. 3,8} + rho2] > E_v^o. This inference is valid only for a nondegenerate ground state (or a unique ground-state density). The manuscript does not state such an assumption, and for degenerate Coulombic systems different ground-state densities can have the same energy. Please add the nondegeneracy assumption explicitly or prove that the FDET solution cannot yield another ground-state density. The same issue affects the claimed uniqueness of rho1^o[rho2] in Eq. (1).
- [Theorem I, proof of Eq. (19)] The proof applies Lemma A to an arbitrary density rho1 that vanishes on V_z, but Lemma A is stated only for the specific density rho1 = rho_v^o - rho2 with rho2 in A. The needed generalization is true because the proof of Lemma A uses only rho1 = 0 on V_z, but it is not stated or proved. Please formulate this as an explicit corollary or extend the proof of Lemma A to all densities vanishing on V_z, since Theorem I depends on this step.
minor comments (3)
- [Lemma A, Eq. (18)] The sentence 'the first Gateaux differential exists because rho_v^o is v-representable' is not a theorem in general; differentiability of E_xct at a v-representable density is a subtle issue. The proof does not need this assertion because the failure of the second term already gives the contradiction, so please remove or replace it with a correct justification.
- [Lemma B] The proof of Lemma B is omitted. Since Lemma B is used for the noninteracting variant and in the discussion of Eq. (28), please include a sketch to confirm that the boundary argument applies to T_s^{nad} without additional conditions.
- [Notation and typos] There are several small presentation issues: Eq. (1) contains a misplaced parenthesis ('forall r)' should be 'forall r'); 'bound from below and above' should be 'bounded from below and above'; the Discussion contains the duplicated word 'procedure'; and Eq. (28) uses the notation vs[rho] before it is defined.
Circularity Check
No significant circularity: the central theorem follows from a self-contained boundary-of-domain argument, with the author's prior work cited only for definitions and context.
full rationale
The paper's central result is Lemma A, which shows that the FDET nonadditive potential v_nad_xct[rho1, rho2] does not exist when rho2 is in class A and rho1 = rho_v^o - rho2. The proof constructs a bounded, zero-integral variation Delta that is negative on the volume Vz where rho2 equals rho_v^o, making rho1 + h*Delta negative for h > 0 and hence inadmissible in the Hohenberg-Kohn functional. This is a domain-of-existence argument about functional derivatives, not an assumption of the conclusion. Theorems I and II follow directly from Lemma A together with the standard Hohenberg-Kohn variational and uniqueness principles: any FDET stationary density must differ from rho_v^o - rho2, so the total density differs from rho_v^o and therefore has strictly higher HK energy. No parameter is fitted, no prediction is derived from a fitted input, and no load-bearing step is justified by an unverified self-citation. The citations to Ref. 1 (Wesolowski 2008) are for the definition of FDET, the form of the embedding potential, and the equivalence in Eq. 12, all of which are context rather than evidence for Lemma A or Theorems I-II. The paper's own prior Ref. 19 is cited only to contrast the earlier non-uniqueness observation with the present stronger nonexistence statement, and the external Ref. 5 is cited as prior discussion. Lemma B's proof is omitted but explicitly stated to follow the same steps as Lemma A applied to T_s^nad, so the omission is a presentational choice rather than a circular reduction. The assumptions used, such as Coulombic external potentials and v-representability of rho_v^o, are standard independent DFT facts and are not equivalent to the theorem being proved. The derivation is therefore self-contained against external benchmarks, and the appropriate circularity score is 0.
Assumptions & free parameters
assumptions (4)
- standard math The Hohenberg-Kohn functional E_HK^v[rho] is defined only for nonnegative densities integrating to N, and the ground-state density rho_v^o is its unique minimizer.
- domain assumption For Coulombic external potentials, the ground-state density satisfies rho_v^o(r) > 0 everywhere except at infinity.
- domain assumption The FDET eigenvalue equation solution equals the constrained minimizer only when the latter is v-representable (Eq. 12).
- standard math The functional derivative of E_nad_xct must exist as a two-sided Gateaux derivative for the FDET embedding potential to be defined.
Cite this review
Pith. "Pith review of Density dependent embedding potentials for piecewise exact densities." pith.science (2026). https://pith.science/paper/TTETTIQI
@misc{pith2026250608744,
author = {Pith},
title = {Pith review of: Density dependent embedding potentials for piecewise exact densities},
year = {2026},
howpublished = {\url{https://pith.science/paper/TTETTIQI}},
note = {Machine review of arXiv:2506.08744}
}
abstract
Frozen Density Embedding Theory (FDET) [Wesolowski {\it Phys. Rev. A} {\bf 77}, 012504 (2008)] provides the interpretation of the eigenvalue equations for an embedded $N'$-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 $N-N'>0$ with $N$ being the total number of electrons in the whole system ($\min_{\rho\rightarrow\forall_{\mathbf{r}}\big(\rho({\mathbf r})\ge \rho_2({\mathbf r}\big)} E^{HK}_v[\rho]=E^{HK}_v[\rho_1^{FDET}+\rho_2]\ge E^{HK}[\rho_v^{o}]=E^o_v$). The exact FDET eigenvalue equations are analysed for $\rho_2$ such that it is equal to the exact ground-state density $\rho_v^{o}({\mathbf r})$ in some measurable volume. It is shown that, the stationary ($\rho_1^{FDET}$) obtained from the FDET eigenvalue equations - if it exists - differs from $\rho_1^o=\rho_v^{o}-\rho_2$ leading to the sharp inequality $E^{HK}[\rho_1^{FDET}+\rho_2]> E^o_v$ for such densities $\rho_2$. The result is discussed in the context of subsystem DFT, pseudopotential theory, and exact density-dependent embedding potentials.
Figures
Reference graph
Works this paper leans on
-
[1]
author author T. A. \ Weso owski ,\ title title Embedding a multideterminantal wave function in an orbital-free environment , \ https://doi.org/10.1103/PhysRevA.77.012504 journal journal Physical Review A \ volume 77 ,\ pages 012504 ( year 2008 ) NoStop
-
[2]
author author M. Levy ,\ title title Electron densities in search of Hamiltonians , \ https://doi.org/10.1103/PhysRevA.26.1200 journal journal Physical Review A \ volume 26 ,\ pages 1200--1208 ( year 1982 ) NoStop
-
[3]
author author R. G. \ Parr \ and\ author W. Yang ,\ https://doi.org/10.1093/oso/9780195092769.001.0001 title Density-Functional Theory of Atoms and Molecules \ ( publisher Oxford University Press ,\ address New York ,\ year 1989 ) NoStop
-
[4]
author author W. Kohn \ and\ author L. J. \ Sham ,\ title title Self-Consistent Equations Including Exchange and Correlation Effects , \ https://doi.org/10.1103/PhysRev.140.A1133 journal journal Physical Review \ volume 140 ,\ pages A1133--A1138 ( year 1965 ) NoStop
-
[5]
author author O. V. \ Gritsenko ,\ title title On the Principal Difference Between the Exact and Approximate Frozen-Density Embedding Theory , \ in\ @noop booktitle Recent Progress in Orbital-Free Density Functional Theory ,\ series Recent Progress in Computational Chemistry , Vol. volume 6 ,\ editor edited by\ editor T. A. \ Wesolowski \ and\ editor Y. A...
work page 2013
-
[6]
author author P. Cortona ,\ title title Self-consistently determined properties of solids without band-structure calculations , \ https://doi.org/10.1103/PhysRevB.44.8454 journal journal Physical Review B \ volume 44 ,\ pages 8454--8458 ( year 1991 ) NoStop
-
[7]
author author J. C. \ Phillips \ and\ author L. Kleinman ,\ title title New Method for Calculating Wave Functions in Crystals and Molecules , \ https://doi.org/10.1103/PhysRev.116.287 journal journal Physical Review \ volume 116 ,\ pages 287--294 ( year 1959 ) NoStop
-
[8]
author author T. A. \ Wesolowski \ and\ author A. Savin ,\ title title Non-Additive Kinetic Energy and Potential in Analytically Solvable Systems and Their Approximate Counterparts , \ in\ @noop booktitle Recent Progress in Orbital-Free Density Functional Theory ,\ series Recent Advances in Computational Chemistry , Vol. volume 6 ,\ editor edited by\ edit...
work page 2013
Show all 19 references
-
[9]
Roncero , author M
author author O. Roncero , author M. P. \ de Lara-Castells , author P. Villarreal , author F. Flores , author J. Ortega , author M. Paniagua ,\ and\ author A. Aguado ,\ title title An inversion technique for the calculation of embedding potentials , \ https://doi.org/10.1063/1...
-
[10]
Savin \ and\ author T
author author A. Savin \ and\ author T. A. \ Wesolowski ,\ title title Orbital-Free Embedding Effective Potential in Analytically Solvable Cases , \ in\ https://doi.org/10.1007/978-90-481-2596-8 \_ 15 booktitle Prog. Theor. Chem. Phys. ,\ Vol. volume 19 \ ( year 2009 )\ pp.\ p...
-
[11]
author author J. D. \ Goodpaster , author N. Ananth , author F. R. \ Manby ,\ and\ author T. F. \ Miller ,\ title title Exact nonadditive kinetic potentials for embedded density functional theory , \ https://doi.org/10.1063/1.3474575 journal journal The Journal of Chemical Phy...
-
[12]
Fux , author C
author author S. Fux , author C. R. \ Jacob , author J. Neugebauer , author L. Visscher ,\ and\ author M. Reiher ,\ title title Accurate Frozen-Density Embedding Potentials as a First Step Towards a Subsystem Description of Covalent Bonds , \ https://doi.org/10.1063/1.3376251 ...
-
[13]
de Silva \ and\ author T
author author P. de Silva \ and\ author T. A. \ Wesolowski ,\ title title Exact non-additive kinetic potentials in realistic chemical systems , \ https://doi.org/10.1063/1.4749573 journal journal The Journal of Chemical Physics \ volume 137 ,\ pages 094110 ( year 2012 ) NoStop
-
[14]
author author D. G. \ Artiukhin , author C. R. \ Jacob ,\ and\ author J. Neugebauer ,\ title title Excitation energies from frozen-density embedding with accurate embedding potentials , \ https://doi.org/10.1063/1.4922429 journal journal The Journal of Chemical Physics \ volum...
-
[15]
Nafziger , author K
author author J. Nafziger , author K. Jiang ,\ and\ author A. Wasserman ,\ title title Accurate reference data for the nonadditive, noninteracting kinetic energy in covalent bonds , \ https://doi.org/10.1021/acs.jctc.6b01050 journal journal Journal of Chemical Theory and Compu...
-
[16]
Banafsheh \ and\ author T
author author M. Banafsheh \ and\ author T. A. \ Wesolowski ,\ title title Nonadditive kinetic potentials from inverted Kohn-Sham problem , \ https://doi.org/10.1002/qua.25410 journal journal International Journal of Quantum Chemistry \ volume 118 ,\ pages e25410 ( year 2018 ) NoStop
-
[17]
Schnieders \ and\ author J
author author D. Schnieders \ and\ author J. Neugebauer ,\ title title Accurate embedding through potential reconstruction: A comparison of different strategies , \ https://doi.org/10.1063/1.5037638 journal journal Journal of Chemical Physics \ volume 149 ( year 2018 ),\ 10.10...
-
[18]
Banafsheh , author T
author author M. Banafsheh , author T. A. \ Wesolowski , author T. Gould , author L. Kronik ,\ and\ author D. A. \ Strubbe ,\ title title Nuclear cusps and singularities in the nonadditive kinetic potential bifunctional from analytical inversion , \ https://doi.org/10.1103/Phy...
-
[19]
author author T. A. \ Wesolowski ,\ title title Is the non-additive kinetic potential always equal to the difference of effective potentials from inverting the kohn-sham equation? \ https://doi.org/10.1063/5.0101791 journal journal The Journal of Chemical Physics \ volume 157 ...
Reviewed August 7, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.