Entropic Density Functional Theory

Preprint

Article

Entropic Density Functional Theory

Altmetrics

Downloads

Views

Comments

A peer-reviewed article of this preprint also exists.

This version is not peer-reviewed

Submitted:

22 November 2023

Posted:

04 December 2023

You are already at the latest version

Alerts

Abstract

A formulation of the Density Functional Theory (DFT) is constructed as an application of the method of maximum entropy for an inhomogeneous fluid in thermal equilibrium. The use of entropy as a systematic method to generate optimal approximations is extended from the classical to the quantum domain. This process introduces a family of trial density operators that are parametrized by the particle density. The optimal density operator is that which maximizes the quantum entropy relative to the exact canonical density operator. This approach reproduces the variational principle of DFT and allows a simple proof of the Hohenberg-Kohn theorem at finite temperature. Finally, as an illustration, we discuss the Kohn-Sham approximation scheme at finite temperature

Keywords:

Subject: Physical Sciences - Condensed Matter Physics

1. Introduction

The Density Functional Theory (DFT) is one of the most widely used methods for calculations of the structure of inhomogeneous many-body systems including atoms, molecules, liquids, solids, and surfaces [1,2] (for a pedagogic introduction see [3]). The theory, which finds its earliest roots in the Thomas-Fermi-Dirac model, was first introduced in its modern form by Hohenberg and Kohn who showed that the ground state of an electron gas in an external potential can be uniquely characterized by the electron density [4] and by Kohn and Sham who showed how to include the effects of exchange and correlations [5]. The implications of these ideas were soon extended to finite temperatures in the context of the grand canonical framework [6] (see also [7,8,9] and references therein) and to the statistical mechanics of non-uniform classical fluids, such as the liquid-vapor interface [10,11,12,13] (see also [14] for more references).

In previous work we derived the classical DFT as an application of the method of maximum entropy [15]. A central concept is the use of entropy itself as a tool to generate optimal approximations to probability distributions [17] in terms of those variables that capture the relevant physical information namely, the particle density

n (x)

. We showed that the entropic DFT (eDFT) approach directly leads to Evans’ variational principle of the classical DFT [11].

In this paper we are concerned with the very foundations of the DFT framework and our main goal is to extend the entropic DFT (eDFT) formalism to the quantum domain. We emphasize that our goal is neither to derive an alternative to DFT nor to develop improvements to the approximations that are inevitably necessary to the successful implementation of DFT in practical applications.

In Section 2 we review the use of relative entropy as a tool to update density operators in response to new information and we extend the use of entropy as a tool to derive optimal approximations from the classical context [17] to the quantum domain. In Section 3 we construct the entropic DFT formalism and prove a form of the Hohenberg-Kohn theorem at finite temperature within the canonical (fixed number of particles) framework. In Section 4, as an illustration of the eDFT formalism, we discuss the Kohn-Sham model in the local density approximation. Finally in Section 5 we summarize our conclusions.

2. Preliminaries

The realization that a fundamental theory such as thermodynamics should be interpreted as an application of a general scheme for inference on the basis of information codified into constraints can be traced to Brillouin and Jaynes [18,19,20,21,22]. According to Jaynes – as motivated by the Shannon’s axioms [23] – entropy is interpreted as the amount of information that is missing in a probability distribution. The preferred probability distribution is that which agrees with what we know — the information codified into the constraints — but is maximally ignorant about everything else. Thus, one is led to maximize the entropy subject to constraints, a procedure dubbed the MaxEnt method.

A drawback of this approach is that the interpretation of entropy as an amount of missing information is not completely satisfactory. To address this problem Shore and Johnson [24] proposed that one could avoid invoking questionable measures of information by directly axiomatizing the method for updating probabilities through a variational principle that involved maximizing an entropy functional satisfying certain desirable properties. The question of why should one adopt a variational principle was later clarified by Skilling [25] who proposed a simple ranking strategy: in order to select an optimal distribution (he had in mind the more general case of positive additive distributions which include e.g. intensities in an image) one proceeds by ranking the distributions according to some preference criteria and then choosing the one which ranks the highest. The ranking scheme is naturally implemented by associating a real number – the entropy – to each distribution with the preference criteria fixed through the axioms of Shore and Johnson. In later work the nature of the method of maximum entropy was further streamlined as a scheme designed to update probabilities when confronted with new information. In this approach the question “what is information?” receives a very simple answer. Information is just the constraints we decide to impose on our beliefs, and there is no need to define “amounts” of information. The motivation behind the design criteria was clarified and their number reduced from five to two [26,27,28,29] (reviewed in [31] and [39]).

2.1. The Quantum MaxEnt Method

The task of extending the method of maximum entropy to the quantum domain as a method to update density operators was carried out by Vanslette [29]. The goal is to update a prior density operator

\hat{σ}

when provided with new information in the form of the expected value of some self-adjoint operators

〈 {\hat{A}}_{i} 〉 = A_{i}

. Vanslette showed that the Umegaki relative entropy [32],

S_{r} [\hat{ρ} | \hat{σ}] \overset{def}{=} - Tr (\hat{ρ} log \hat{ρ} - \hat{ρ} log \hat{σ}),

(1)

provides the unique criterion to rank density operators

\hat{ρ}

relative to the prior

\hat{σ}

The maximization of

S_{r} [\hat{ρ} | \hat{σ}]

subject to the constraints

〈 {\hat{A}}_{i} 〉 = A_{i}

and normalization,

δ [S_{r} [\hat{ρ} | \hat{σ}] + α_{0} (1 - Tr \hat{ρ}) + \sum_{i} α_{i} (A_{i} - Tr \hat{ρ} {\hat{A}}_{i})] = 0,

(2)

leads to the posterior density operator

{\hat{ρ}}^{*} = \frac{1}{Z} exp (log \hat{σ} - \sum_{i} α_{i} {\hat{A}}_{i}),

(3)

where

Z (α_{i}) = e^{α_{0}} = Tr exp (log \hat{σ} - \sum_{i} α_{i} {\hat{A}}_{i}) .

(4)

Substituting

{\hat{ρ}}^{*}

back into eq.(1) gives the value of the maximized entropy,

S (A_{i}) \overset{def}{=} S_{r} [{\hat{ρ}}^{*} | \hat{σ}] = \sum_{i} α_{i} A_{i} + log Z .

(5)

It is widely known that the classical MaxEnt method leads to a mathematical formalism characterized by a contact structure (see e.g., [35,36]). In a parallel development the use of Legendre transforms in the context of DFT has also been widely explored [7,33,34]. These results can be extended to the quantum domain leading to a similar contact structure (see e.g., [16]). The significance of these results is that the physical content of the formalism is preserved under Legendre transformations quite independently of restrictions to thermal equilibrium and of the physical significance of the so-called “free energies” or Massieu functions.

2.2. Optimal Approximations of Density Operators

The last prerequisite for the construction of the DFT formalism is a systematic method of approximation for density operators. The method we adopt is an extension of the technique developed by Tseng and Caticha in the classical context [17]. The problem is that the exact probability distributions Q obtained using the MaxEnt method are often too intractable to be useful in actual calculations. The solution is to consider a family of more tractable trial distributions

P_{θ}

dependent on some parameters

θ

. The goal is to select the trial distribution

P_{θ^{*}}

that best approximates the exact distribution Q. In [17] it was argued that the criterion to select the optimal parameters

θ^{*}

is again provided by the method of maximum entropy: The optimal

P_{θ^{*}}

is that which is “closest” to the exact Q in the sense that it maximizes the relative entropy

S [P_{θ} | Q]

Next, we extend this approximation technique to the quantum domain. We consider a family of tractable density operators

{\hat{ρ}}_{θ}

parametrized by parameters

θ

. The member of the trial family

{\hat{ρ}}_{θ}

that best approximates the exact density operator

{\hat{ρ}}^{*}

is the one which maximizes the entropy of

{\hat{ρ}}_{θ}

relative to

{\hat{ρ}}^{*}

{\frac{\partial}{\partial θ} S_{r} [{\hat{ρ}}_{θ} | {\hat{ρ}}^{*}]|}_{θ = θ^{*}} = 0 .

(6)

As an example, consider the special case where

{\hat{ρ}}^{*}

and

{\hat{ρ}}_{θ}

take the exponential form,

{\hat{ρ}}^{*} = \frac{1}{Z} e^{- β \hat{H}} and {\hat{ρ}}_{θ} = \frac{1}{Z_{θ}} e^{- β {\hat{H}}_{θ}},

(7)

the Gibbs inequality,

S_{r} [{\hat{ρ}}_{θ} | {\hat{ρ}}^{*}] \leq 0,

(8)

reduces to the Bogolyubov inequality,

F \leq F_{θ} + {〈 \hat{H} - {\hat{H}}_{θ} 〉}_{θ},

(9)

where

F = - \frac{1}{β} log Z, F_{θ} = - \frac{1}{β} log Z_{θ}, and {〈 \cdot 〉}_{θ} = Tr [{\hat{ρ}}_{θ} (\cdot)] .

(10)

Thus, the argument above shows the popular approximation method based on the Bogolyubov inequality (see e.g., [37]) is a special case of the more general approximation method based on entropy maximization.

3. Density functional formalism

The goal of the DFT formalism is to find tractable approximations to study the structure of matter. The first crucial step is to recognize that the quantity that captures the desired structural information is the electron density

n (x)

. We wish to design a formalism in which the central role played by the electron density is explicitly displayed.

In the absence of magnetic fields the time independent Schrödinger equation for an electron gas of N particles is

\hat{H} | ψ 〉 = E | ψ 〉,

(11)

where

{\hat{H}}_{v} = {\hat{H}}^{(0)} + \hat{V} = \hat{K} + \hat{U} + \hat{V} = \sum_{i = 1}^{N} \frac{{\hat{p}}_{i}^{2}}{2 m} + \frac{e^{2}}{2} \sum_{j \neq k}^{N} \frac{1}{| {\hat{x}}_{j} - {\hat{x}}_{k} |} + \sum_{l = 1}^{N} v ({\hat{x}}_{l}),

(12)

and

| ψ 〉

is an antisymmetrized product of N two-spinor orbitals. The potential

\hat{U}

describes interparticle interactions and the potential

\hat{V}

describes interactions with nuclei and other external potentials.

3.1. Introducing density as the relevant variable

We are interested in the thermal properties of an inhomogeneous electron fluid and therefore we need trial states that describe both thermal equilibrium and inhomogeneity. The former is imposed by a constraint on the expected value of energy and the latter is incorporated by constraints on the expected value

n (x)

of the electron density

\hat{n} (x)

. The continuous density function

n (x)

plays a role analogous to the discrete parameters

θ

in equations (6-10).

Adopting a uniform prior, the relevant trial states are obtained by maximizing the entropy

S_{r} [\hat{ρ} | \hat{1}] = - Tr \hat{ρ} log \hat{ρ},

(13)

subject to the constraints

\begin{matrix} Tr \hat{ρ} & = & 1, \end{matrix}

(14)

\begin{matrix} Tr \hat{ρ} {\hat{H}}_{v} & = & E, \end{matrix}

(15)

\begin{matrix} and Tr \hat{ρ} \hat{n} (x) & = & n (x), \end{matrix}

(16)

where

\hat{n} (x) = \sum_{i = 1}^{N} δ ({\hat{x}}_{i} - x) and \int d^{3} x n (x) = N .

(17)

To be clear, throughout this work the trace is taken over the Hilbert space of a fixed number N of particles and in this respect our formalism resembles the canonical ensemble approach. Indeed, all states

| ψ 〉

in the Hilbert space are eigenstates of the number operator,

\hat{N} | ψ 〉 = \int d^{3} x \hat{n} (x) | ψ 〉 = N | ψ 〉 so that 〈 ψ | \hat{N} | ψ 〉 = N,

(18)

but they need not be eigenstates of the density operators

\hat{n} (x)

. Our formalism differs from the canonical formalism in that eq.(16) represents an additional infinite number of constraints — one constraint on the expected density function

n (x)

at each point in space. Due to (18) the expected density function

n (x)

is not arbitrary; it is constrained to obey (18).

Proceeding to the MaxEnt analog of eq.(3) we find the trial density operator

{\hat{ρ}}_{n} = \frac{1}{Z_{v}} exp (- β {\hat{H}}_{v} - \int d^{3} x α (x) \hat{n} (x)),

(19)

where

Z_{v} (β; α] = Tr exp (- β {\hat{H}}_{v} - \int d^{3} x α (x) \hat{n} (x)),

(20)

and where

β

and the infinite number of Lagrange multipliers

α (x)

are implicitly determined by

\frac{\partial log Z_{v} (β; α]}{\partial β} = - E and \frac{δ log Z_{v} (β; α]}{δ α (x)} = - n (x),

(21)

with the additional constrain (17),

\int d^{3} x n (x) = - \int d^{3} x \frac{δ Z_{v} (β; α]}{δ α (x)} = N .

(22)

The notation

Z_{v} (β; α]

indicates that Z is a function of

β

and a functional of

α (x)

and depends on

v (x)

through the Hamiltonian

{\hat{H}}_{v}

. At this point in the argument there is no implication that the trial states

{\hat{ρ}}_{n}

are in any way more computationally tractable than the exact state

{\hat{ρ}}^{*}

obtained from (19) by setting

α (x)

to zero.

Next we calculate the entropy of

{\hat{ρ}}_{n}

relative to the uniform prior to define the trial entropy,

S_{r} [{\hat{ρ}}_{n} | \hat{1}] = β E + \int d^{3} x α (x) n (x) + log Z_{v} (β; α] \overset{def}{=} S_{v} (E; n] .

(23)

An important symmetry of the DFT formalism, which is what makes the whole DFT formalism work, arises from the fact that the dependence of

{\hat{ρ}}_{n}

and

Z_{v} (β; α]

v (x)

and

α (x)

occurs only through the particular combination

α_{int} (x) \overset{def}{=} α (x) + β v (x) .

(24)

The reason for the subscript ‘int’, which denotes ‘intrinsic’, will become clear later in eq.(56). This DFT symmetry implies that a change in the potential

v (x)

can be compensated by a suitable change in the multiplier

α (x)

in such a way that

α_{int} (x)

and the expected density

n (x)

remain unaffected. From (12) and (24) we find that (20) can be written as

Z_{v} (β; α] = Tr exp (- β {\hat{H}}^{(0)} - \int d^{3} x α_{int} (x) \hat{n} (x)) \overset{def}{=} Z (β; α_{int}],

(25)

so that eqs.(19) and (21) become

{\hat{ρ}}_{n} = \frac{1}{Z (β; α_{int}]} exp (- β {\hat{H}}^{(0)} - \int d^{3} x α_{int} (x) \hat{n} (x))

(26)

and

n (x) = - \frac{δ log Z (β; α_{int}]}{δ α_{int} (x)} .

(27)

3.2. The entropic DFT variational principle

The exact canonical density operator

{\hat{ρ}}^{*}

is found by maximizing (13) subject to (14) and (15). The result can be read off eq.(19) by setting

α (x) = 0

{\hat{ρ}}^{*} = \frac{1}{Z_{v} (β)} exp (- β {\hat{H}}_{v}) and Z_{v} (β) = Tr exp (- β {\hat{H}}_{v}) .

(28)

(We use a star * to denote exact canonical quantities.) The goal is to approximate

{\hat{ρ}}^{*}

by the best matching member of the family

{{\hat{ρ}}_{n}}

with all density operators referring to the same

β

and N. This involves maximizing the entropy of

{\hat{ρ}}_{n}

relative to

{\hat{ρ}}^{*}

{\frac{δ S_{r} [{\hat{ρ}}_{n} | {\hat{ρ}}^{*}]}{δ n (x)}|}_{β, N} = 0 .

(29)

From (19) and (28) we find

S_{r} [{\hat{ρ}}_{n} | {\hat{ρ}}^{*}] = \int d^{3} x α (x) n (x) + log Z_{v} (β; α] - log Z_{v} (β) .

(30)

Introducing a Lagrange multiplier

α^{*}

to enforce the constraint on N we have,

\frac{δ}{δ n (x)} {[S_{r} [{\hat{ρ}}_{n} | {\hat{ρ}}^{*}] + α^{*} (N - \int d^{3} x^{'} n (x^{'}))]}_{β} = 0 .

(31)

From the construction above one might expect that the optimal

{\hat{ρ}}_{n}

coincides with the exact

{\hat{ρ}}^{*}

. We can check that this is indeed the case. Substituting eq.(30) into (31) we find

\int d^{3} x^{'} [n (x^{'}) + \frac{δ log Z_{v} (β; α]}{δ α (x^{'})}] \frac{δ α (x^{'})}{δ n (x)} = α^{*} - α (x)

(32)

The LHS vanishes by eq.(21). Therefore, the optimal

{\hat{ρ}}_{n}

is achieved for

α (x) = α^{*}

. From (19), (28) and (30) we see that

α^{*} = 0

which means that imposing the N constraint was unnecessary: the optimal density reproduces the exact density

n^{*} (x)

whether the variations

δ n (x)

preserve the total N or not.

We conclude that the entropic DFT variational principle,

{\frac{δ S_{r} [{\hat{ρ}}_{n} | {\hat{ρ}}^{*}]}{δ n (x)}|}_{n^{*} (x)} = 0,

(33)

leads to an optimal

{\hat{ρ}}_{n}

which coincides with the exact canonical

{\hat{ρ}}^{*}

in eq.(28),

{\hat{ρ}}_{n}^{opt} = {\hat{ρ}}^{*}, where α^{opt} (x) = α^{*} = 0,

(34)

Thus, at this point our “approximation” scheme is (trivially) exact: by explicit construction we have demonstrated the existence of a functional of the density

n (x)

β

and N — the relative entropy

S_{r} [{\hat{ρ}}_{n} | {\hat{ρ}}^{*}]

— that assumes its maximum value at the exact density

n^{*} (x)

. At this point, however, we have not yet shown that this variational principle is equivalent to the thermal DFT principle derived by Mermin [6]. This, we show next.

3.3. The DFT Theorem

Equations (23) and (30) allows us to write

S_{r} [{\hat{ρ}}_{n} | {\hat{ρ}}^{*}] = - β Ω_{v} (β; n] - log Z_{v} (β)

(35)

where we have introduced the “free energy” functional

Ω_{v} (β; n] \overset{def}{=} E - \frac{1}{β} S_{v} (E; n] .

(36)

The new functional

Ω_{v}

Ω_{v} (β; n] = - \frac{1}{β} \int d^{3} x α (x) n (x) - \frac{1}{β} log Z_{v} (β; α],

(37)

allows us to rewrite the entropic variational principle (31) as

{\frac{δ Ω_{v} (β; n]}{δ n (x)}|}_{n^{*} (x)} = 0 .

(38)

The optimal density

n^{*} (x)

is found by minimizing

Ω_{v} (β; n]

at fixed

β

and N. Furthermore, from (37) the multipliers

α (x)

are obtained from

α (x) = - β \frac{δ Ω_{v} (β; n]}{δ n (x)} .

(39)

From eq.(34),

α^{opt} (x) = α^{*} = const

, we obtain

{\nabla \frac{δ Ω_{v} (β; n]}{δ n (x)}|}_{n^{*} (x)} = 0,

(40)

which has been called the “core integro-differential equation of DFT” [11].

To proceed further, substitute (12), (15), into (36) to find

Ω_{v} (β; n] = {〈 \hat{K} + \hat{U} 〉}_{{\hat{ρ}}_{n}} + \int d^{3} x v (x) n (x) - \frac{1}{β} S_{v} (E; n],

(41)

so that

Ω_{v} (β; n] = F_{v} (β; n] + \int d^{3} x v (x) n (x),

(42)

where we have introduced

F_{v} (β; n] \overset{def}{=} {〈 \hat{K} + \hat{U} 〉}_{{\hat{ρ}}_{n}} - \frac{1}{β} S_{v} (E; n] .

(43)

We are now ready to state the finite temperature DFT theorem.

The Density Functional Theorem:

The density functional

F_{v} [n]

is independent of the external potential

v (x)

{\frac{δ F_{v} (β; n]}{δ v (x)}|}_{β, n (x)} = 0 .

(44)

This result justifies dropping the index v,

F (β; n] \overset{def}{=} F_{v} (β; n],

(45)

and referring to

F (β; n]

as the intrinsic density functional. (The term ‘intrinsic’ indicates that

F (β; n]

is independent of the external potential

v (x)

Proof:

The crucial observation behind the entropic DFT formalism is that

{\hat{ρ}}_{n}

and

Z_{v} (β; α]

depend on the external potential

v (x)

and the Lagrange multiplier function

α (x)

only through the particular combination

α_{int} (x)

defined in (24). Substitute (23), (24) and (25) into (43) to get

F_{v} (β; n] = - \frac{1}{β} \int d^{3} x α_{int} (x) n (x) - \frac{1}{β} log Z (β; α_{int}] .

(46)

Then the derivative

δ / δ v (x^{'})

at fixed

β

and

n (x)

\frac{δ F_{v} (β; n]}{δ v (x^{'})} = \int d^{3} x^{″} \frac{δ F_{v} (β; n]}{δ α_{int} (x^{″})} {\frac{δ α_{int} (x^{″})}{δ v (x^{'})}|}_{β, n (x)} .

(47)

Eq.(26) shows that keeping

n (x)

fixed is achieved by keeping

α_{int} (x)

fixed and vice versa, therefore

{\frac{δ α_{int} (x^{″})}{δ v (x^{'})}|}_{β, n (x)} = {\frac{δ α_{int} (x^{″})}{δ v (x^{'})}|}_{β, α_{int} (x)} = 0,

(48)

which implies (44) and concludes the proof.

Equations (19) and (40) suggest that (up to an additive constant) the multiplier

α (x)

plays a role analogous to that of a chemical potential. Let us then use eq.(39) to introduce

γ (x) \overset{def}{=} - \frac{α (x)}{β} = \frac{δ Ω_{v} (β; n]}{δ n (x)},

(49)

which we shall call the “local chemical potential.” The core equation (40) has a natural interpretation: the condition for neighboring volume elements to be in equilibrium is that the local chemical potential be uniform,

{\nabla γ (x)|}_{n^{*}} = 0 .

(50)

The optimal value of

γ (x)

γ^{*} = - \frac{α^{*}}{β} = 0 so that \nabla γ^{*} = 0 .

(51)

From eq.(42) we have

δ Ω_{v} (β; n] = δ F (β; n] + \int d^{3} x [n (x) δ v (x) + v (x) δ n (x)],

(52)

while eq.(49) gives

δ Ω_{v} (β; n] = \int d^{3} x (\frac{δ Ω_{v}}{δ v (x)} δ v (x) + \frac{δ Ω_{v}}{δ n (x)} δ n (x)) = \int d^{3} x [n (x) δ v (x) + γ (x) δ n (x)] .

(53)

Subtracting these two equations gives

δ F (β; n] = \int d^{3} x [γ (x) - v (x)] δ n (x),

(54)

which shows that

δ F / δ n

can be interpreted as the local intrinsic chemical potential,

\frac{δ F (β; n]}{δ n (x)} \overset{def}{=} γ_{int} (x),

(55)

with

γ (x) = γ_{int} (x) + v (x) and γ_{int} (x) = - \frac{α_{int} (x)}{β} .

(56)

Evaluating at

n^{*}

gives the equilibrium intrinsic chemical potential,

{\frac{δ F (β; n]}{δ n (x)}|}_{n^{*}} = γ_{int}^{*} (x) .

(57)

(The term ‘intrinsic’ reminds us that both

γ_{int} (x)

and

γ_{int}^{*} (x)

are independent of the external potential

v (x)

We mentioned earlier that the multiplier

α (x)

plays a role analogous to that of a chemical potential. We can now be more explicit. Let

μ (x) = μ (x; n] and μ_{int} (x) = μ_{int} (x; n]

(58)

be the actual chemical potential and the intrinsic chemical potential at location x,

μ (x; n] = μ_{int} (x; n] + v (x) .

(59)

Equilibrium among different volume elements is achieved when

{\nabla μ (x; n]|}_{n^{*}} = 0 ⟹ μ (x; n^{*}] = μ^{*} = const .

(60)

Then eqs.(49)-(51) lead us to identify

γ (x) = μ (x) - μ^{*} so that γ^{*} = 0,

(61)

and

γ_{int} (x) = μ_{int} (x; n] - μ (x; n^{*}] .

(62)

We can express the eDFT variational principle in terms of F. Using eqs.(38) and (42) we find

{\frac{δ}{δ n (x)} (F (β; n] + \int d^{3} x^{'} v (x^{'}) n (x^{'}))|}_{n^{*} (x)} = 0 .

(63)

To summarize, we have reproduced the foundational theorem behind the thermal DFT formalism as an application of maximum entropy methods. This is the main result of this paper. The treatment, so far, has been exact. In the next section, as an illustration of the method, we adapt the well-known Kohn-Sham model to the entropic DFT approach.

4. The Kohn-Sham approximation scheme

The exact calculation of

F (β; n]

requires calculating

Z (β; α_{int}]

. Unfortunately, this is just as difficult as calculating the original canonical partition function

Z_{v} (β)

which was precisely what we wanted to avoid. An analogous problem arises in the standard many-body theory: even for relatively small particle numbers the calculation of the N-particle wave function becomes impractically difficult because the wave function

Ψ ({\vec{r}}_{1} \dots {\vec{r}}_{N})

lives in a

3 N

-dimensional configuration space. The DFT framework attempts to evade this problem by focusing attention on the hopefully easier problem of calculating the density

n (x)

which is a function that lives in a mere 3 dimensions. Unfortunately, the problem is not solved, but merely transferred to the calculation of the functional

F (β; n]

. Not all is lost, however, because the reformulation in terms of the density

n (x)

suggests new useful approximations.

The discussion below parallels closely the ground state formulation of Kohn and Sham [5]. It differs from the grand canonical thermal DFT of Mermin [6] in that here we remain within the canonical framework of fixed particle number. In common with the Hartree-Fock approximation the Kohn-Sham model reduces an interacting many-particle Schrödinger equation to that of a single particle in the presence of an effective potential that includes exchange and correlation effects. An important advantage is that, unlike Hartree-Fock, the Kohn-Sham framework can in principle be exact. In practice, however, the success of the model hinges on whether the approximations for exchange and correlations are sufficiently simple and accurate. Fortunately, the “local density approximation,” which is exact for a uniform electron gas, and should remain valid for slowly varying potentials, has turned out to be quite successful for the prediction of bond lengths and molecular structures even when these involve inhomogeneities at the atomic scale.

Referring to eq.(43) the idea is that

F (β; n]

can be split into three terms,

F (β; n] = F_{0} (β; n] + U_{C} [n] + F_{xc} (β; n] .

(64)

The first term

F_{0} (β; n]

represents the intrinsic free energy of a gas of non-interacting and uncorrelated particles at the same temperature and density. The second term

U_{C} [n]

is the classical Coulomb interaction,

U_{C} [n] = \frac{e^{2}}{2} \int d^{3} x d^{3} x^{'} \frac{n (x) n (x^{'})}{| x - x^{'} |},

(65)

that represents the dominant contribution from the interparticle potential term

{〈 \hat{U} 〉}_{{\hat{ρ}}_{n}}

in (43). The third

F_{xc} (β; n]

is a correction that accounts for all additional exchange and correlations effects. To the extent that we can define

F_{xc} (β; n]

to be the difference

F_{xc} (β; n] \overset{def}{=} F (β; n] - F_{0} (β; n] - U_{C} [n],

(66)

equation (64) is trivially exact.

We are now ready to substitute (64) into the eDFT variational principle (63). The result is

{[\frac{δ F_{0}}{δ n (x)} + v (x) + \int d^{3} x^{'} \frac{e^{2} n (x^{'})}{| x - x^{'} |} + v_{xc} (x; n]]}_{n^{*} (x)} = 0,

(67)

where we introduced

v_{xc} (x; n] \overset{def}{=} \frac{δ F_{xc}}{δ n (x)} .

(68)

So far this is exact. However, to make further progress we note that although exchange correlations are intrinsically non-local, for a thermal system we can assume that entanglement effects are appreciable only over short distances. Therefore it might not be unreasonable to approximate

F_{xc}

by a sum over independent volume elements. Accordingly, we adopt the so-called local density approximation,

F_{xc} (β; n] \approx F_{xc}^{L D A} (β; n] = \int d^{3} x f_{xc} (n (x)) n (x),

(69)

where the function

f_{xc} (n)

is assumed known: it is the exchange correlation free energy per particle for a uniform electron gas with density n. The corresponding potential

v_{xc} (x) = {\frac{d}{d u} (f_{xc} (u) u)|}_{u = n (x)}

(70)

is therefore also known.

To find the optimal density

n^{*} (x)

that solves the variational equation (67) we can use the same trick introduced by Kohn and Sham. They noticed that their variational equation for the ground state — the analogue of our eq.(67) — is exactly of the form one obtains for a gas of non-interacting and uncorrelated particles moving in an effective single-particle potential. This leads us to rewrite (67) as

{[\frac{δ F_{0}}{δ n (x)} + v_{eff} (x)]}_{n^{*} (x)} = 0,

(71)

where

v_{eff} (x) = v (x) + \int d^{3} x^{'} \frac{e^{2} n (x^{'})}{| x - x^{'} |} + v_{xc} (x) .

(72)

Thus, the problem of N interacting particles has been translated into the problem of a single particle moving in an density-dependent effective potential created by all the other particles. This shows that we can adopt the same iterative procedure followed with the Hartree self-consistent potential. If

n^{(j)} (x)

is the density at the

j^{th}

iteration, use (72) to construct the potential

v_{eff}^{(j)} (x)

, and solve the single-particle equation,

[- \frac{1}{2} \nabla^{2} + v_{eff}^{(j)} (x)] ψ_{k}^{(j)} (x) = ε_{k}^{(j)} ψ_{k}^{(j)} (x) .

(73)

Then construct the density

n^{(j + 1)} (x)

for the next iteration as the thermal average,

n^{(j + 1)} (x) = \sum_{k = 1}^{k_{\max}} \frac{| ψ_{k}^{(j)} {(x) |}^{2}}{1 + exp [β (ε_{k}^{(j)} - μ)]}

(74)

where the cutoff

k_{\max}

is such the occupation of orbitals with

k > k_{\max}

can be neglected and

μ

is found by imposing

\int d^{3} x n (x) = N

. The process is repeated until convergence to the optimal

n^{*}

is achieved.

Just as in the standard Kohn-Sham model neither the single particle potential

v_{eff} (x)

, nor the wave functions

ψ_{k}

and energies

ε_{k}

are to be given any real physical interpretation. They are auxiliary quantities whose only purpose is the calculation of the physical density

n^{*} (x)

5. Conclusion

To summarize our conclusions:

We have produced a reconstruction of DFT that makes explicit how DFT fits within an ongoing research program that places the concepts of entropy and information at the very foundation for all of physics (see e.g., [31]). This includes statistical mechanics [18,19,20,21,22], quantum mechanics [38,39], and as we have shown in this work, also the main techniques to study structure — variational principles including mean field methods and DFT.

We extended the use of entropy as a systematic method to generate optimal approximations from the classical to the quantum domain. This allowed an entropic reconstruction of quantum DFT. This process involves a family of trial density operators parametrized by the particle density. The optimal density operator is found by maximizing the quantum entropy relative to the exact canonical density operator. This approach reproduces the variational principle of DFT and allows a proof of the Hohenberg-Kohn theorem at finite temperature that is simpler in that it evades some of the subtleties of the ground state formalism. Our formalism differs from previous approaches in that (i) the central role of entropy is explicit, and (ii) we remain with the canonical ensemble formalism.

Acknowledgments

We are thankful to Oleg Lunin, Carlo Cafaro, Herbert Fotso, and Daniel Robins, for their reading and insightful comments as the committee members for Yousefi’s thesis; from which this paper is extracted.

Conflicts of Interest

The authors declare no conflict of interest.

References

Kohn, W. Noble Lecture: Electronic structure of matter—wave functions and density functionals. Rev. Mod. Phys. 1999, 71, 1253. [Google Scholar] [CrossRef]
Jones, R.O. Density functional theory: Its origins, rise to prominence, and future. Rev. Mod. Phys. 2015, 87, 897. [Google Scholar] [CrossRef]
Argaman, M.; Makov, G. Density functional theory: an introduction. Am. J. Phys. 2000, 68, 69. [Google Scholar]
Hohenberg, P.; Kohn, W. Inhomogeneous electron gas. Phys. Rev. 1964, 136, B864. [Google Scholar] [CrossRef]
Kohn, W.; Sham, L.J. Self-consistent equations including exchange and correlation effects. Phys. Rev. 1965, 140, A1133. [Google Scholar] [CrossRef]
Mermin, D. Thermal properties of inhomogeneous electron gas. Phys. Rev. 1965, 137, A1441. [Google Scholar] [CrossRef]
Eschrig, H. T>0 ensemble-state density functional theory via Legendre transform. Phys. Rev. B 2010, 205120. [Google Scholar]
Pribram-Jones, A.; Pittalis, S.; Gross, E.; Burke, K. Thermal Density Functional Theory in Context. In Frontiers and Challenges in Warm Dense Matter, Lecture Notes in Computational Science and Engineering; Graziani, F., et al., Eds.; Springer International Publishing, 2014; Volume 96, pp. 25–60. [Google Scholar]
Burke, K.; Smith, J.C.; Grabowski, P.E.; Pribram-Jones, A. Exact conditions on the temperature dependence of density functionals. Phys. Rev. B 2016, 93, 195132. [Google Scholar] [CrossRef]
Ebner, C.; Saam, W.F.; Stroud, D. Density-functional theory of simple classical fluids. I. Surfaces. Phys. Rev. A 1976, 14, 2264. [Google Scholar] [CrossRef]
Evans, R. The nature of the liquid-vapor interface and other topics in the statistical mechanics of non-uniform classical fluids. Advances in Physics 1979, 28, 143–200. [Google Scholar] [CrossRef]
Tarazona, P. Free-energy density functional for hard spheres. Phys. Rev. A 1985, 31, 2672. [Google Scholar] [CrossRef] [PubMed]
Rosenfeld, Y. Free-energy model for the inhomogeneous hard-sphere fluid mixture and density-functional theory of freezing. Phys. Rev. Lett. 1989, 63, 980. [Google Scholar] [CrossRef] [PubMed]
Evans, R.; Oettel, M.; Roth, R.; Kahl, G. New developments in classical density functional theory. J. Phys.: Condens. Matter 2016, 28, 240401. [Google Scholar] [CrossRef] [PubMed]
Yousefi, A.; Caticha, A. An entropic approach to classical density functional theory. Phys. Sci. Forum 2021, arXiv:2108.015943, 13. [Google Scholar]
Yousefi, A. Entropic density functional theory: entropic inference and the equilibrium state of inhomogeneous fluids; State University of New York at Albany ProQuest Dissertation Publishing, 2021; p. 28863510. [Google Scholar]
Tseng, C.-Y.; Caticha, A. Using relative entropy to find optimal approximations: an application to simple fluids. Physica A 2008, arXiv:0808.4160v1387, 6759. [Google Scholar] [CrossRef]
Brillouin, L. Science and Information Theory; Academic Press: New York, 1952. [Google Scholar]
Brillouin, L. The negentropy principle of information. J. Appl. Phys. 1953, 24, 1152. [Google Scholar] [CrossRef]
Jaynes, E.T. Information theory and statistical mechanics. Phys. Rev. 1957, 106, 620. [Google Scholar] [CrossRef]
Jaynes, E.T. Information theory and statistical mechanics II. Phys. Rev. 1957, 108, 171. [Google Scholar] [CrossRef]
Jaynes, E.T. Where do we stand on maximum entropy? In The Maximum Entropy Principle; Levine, R.D., Tribus, M., Eds.; MIT Press, 1979. [Google Scholar]
Shannon, C.E. A mathematical theory of communication. The Bell System Technical Journal 1948, 27, 3. [Google Scholar] [CrossRef]
Shore, J.; Johnson, R. Axiomatic derivation of the principle of maximum entropy and the principle of minimum cross-entropy. IEEE Transactions on Information Theory 1980, 26, 26–37. [Google Scholar] [CrossRef]
Skilling, J. The Axioms of Maximum Entropy. Maximum-Entropy and Bayesian Methods in Science and Engineering 1988, 173–187. [Google Scholar]
Caticha, A. Relative Entropy and Inductive Inference. AIP Conf. Proc. 2004, arXiv:physics/0311093707, 75. [Google Scholar]
Caticha, A.; Giffin, A. Updating Probabilities. AIP Conf. Proc 2006, arXiv:physics/0608185872, 31. [Google Scholar]
Caticha, A. Information and Entropy. AIP Conf. Proc. 2007, arXiv:0710.1068954, 11. [Google Scholar]
Vanslette, K. Entropic Updating of Probabilities and Density Matrices. Entropy 2017, arXiv:1710.0937319, 664. [Google Scholar] [CrossRef]
Caticha, A. Entropy, Information, and the Updating of Probabilities. Entropy 2021, arXiv:2107.0452923, 895. [Google Scholar] [CrossRef] [PubMed]
Caticha, A. Entropic Physics: Probability, Entropy and the Foundations of Physics. Available online: https://www.arielcaticha.com/ (accessed on 20 November 2023).
Umegaki, H. Conditional expectation in an operator algebra. IV. Entropy and information. Kodai Math Sem. Rep. 1962, 14, 59–85. [Google Scholar] [CrossRef]
Lieb, E.H. Density functionals for Coulomb systems. Int. J. Quantum Chem. 1983, 24, 243. [Google Scholar] [CrossRef]
Fukuda, R.; Kotani, T.; Suzuki, Y.; Yokojima, S. Density functional theory through Legendre transformation. Prog. Theor. Phys. 1994, 92, 833–862. [Google Scholar] [CrossRef]
Rajeev, S.G. A Hamilton-Jacobi formalism for thermodynamics. Annals of Physics 2008, arXiv:0711.4319323, 2265–2285. [Google Scholar] [CrossRef]
Balian, R.; Valentin, P. Hamiltonian structure of thermodynamics with gauge. Eur. Phys. J. B 2001, 21, 269–282. [Google Scholar] [CrossRef]
Feynman, R.P. Statistical mechanics: a set of lectures; W. A. Benjamin: Reading, MA, USA, 1972. [Google Scholar]
Caticha, A. The Entropic Dynamics Approach to Quantum Mechanics. Entropy 2019, arXiv:1908.0469321, 943. [Google Scholar] [CrossRef]
Caticha, A. Quantum Mechanics as Hamilton-Killing Flows on a Statistical Manifold. Phys. Sci. Forum 2021, arXiv:2107.085023, 12. [Google Scholar]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.

MDPI Initiatives

Important Links

Choose an area of interest and we will send you notifications of new preprints at your preferred frequency.

Disclaimer

Entropic Density Functional Theory

Abstract

1. Introduction

2. Preliminaries

2.1. The Quantum MaxEnt Method

2.2. Optimal Approximations of Density Operators

3. Density functional formalism

3.1. Introducing density as the relevant variable

3.2. The entropic DFT variational principle

3.3. The DFT Theorem

4. The Kohn-Sham approximation scheme

5. Conclusion

Acknowledgments

Conflicts of Interest

References

MDPI Initiatives

Important Links

Subscribe