The DFT Roadmap: From Theory to Quantum ESPRESSO and AKAI KKR — Part 1

April 6, 2026

This is Part 1 of a multi-part series. Here we will develop the full mathematical framework of Density Functional Theory (DFT) from first principles. Part 2 will cover the practical implementation inside Quantum ESPRESSO and AKAI KKR.

1 The Many-Body Problem

The motion of electrons in atoms, molecules, and solids is governed by the Schrödinger equation. The computational complexity to obtain the exact solution grows exponentially with the number of electrons $N$ . When $N$ exceeds a few dozen, calculating the exact ground-state wave function becomes extremely expensive [1]. Two strategies have emerged:

Wave-function methods: It works directly with the $3N$ -variable many-body wave function $\Psi(\mathbf{r}_1,\ldots,\mathbf{r}_N)$ and attempt systematic approximations (Configuration Interaction, Coupled Cluster, Quantum Monte Carlo) [2].
Density Functional Theory: It replaces the $3N$ -variable wave function with the electron density $\rho(\mathbf{r})$ , a function of only three spatial coordinates, drastically reducing the computational cost [1].

We begin with the full, non-relativistic Hamiltonian that describes all kinetic energies and interactions between the electrons and the atomic nuclei in a system:

\hat{H} = -\sum_{I} \frac{\hbar^2}{2M_I} \nabla_{\mathbf{R}_I}^2 \;-\; \sum_{i} \frac{\hbar^2}{2m} \nabla_{\mathbf{r}_i}^2 \;+\; V(\{\mathbf{r}\},\{\mathbf{R}\}) \tag{1}

where the first sum runs over nuclei (mass $M_I$ , position $\mathbf{R}_I$ ), the second over electrons (mass $m$ , position $\mathbf{r}_i$ ), and the interaction potential is

V(\{\mathbf{r}\},\{\mathbf{R}\}) = \underbrace{\sum_{i<j} \frac{e^2}{|\mathbf{r}_i - \mathbf{r}_j|}}_{\text{electron–electron}} \;-\; \underbrace{\sum_{i,I} \frac{Z_I e^2}{|\mathbf{r}_i - \mathbf{R}_I|}}_{\text{electron–nucleus}} \;+\; \underbrace{\sum_{I<J} \frac{Z_I Z_J e^2}{|\mathbf{R}_I - \mathbf{R}_J|}}_{\text{nucleus–nucleus}} \tag{2}

with $\{\mathbf{r}\} = \{\mathbf{r}_1, \mathbf{r}_2, \ldots, \mathbf{r}_N\}$ and $\{\mathbf{R}\} = \{\mathbf{R}_1, \mathbf{R}_2, \ldots, \mathbf{R}_M\}$ being the sets of electron and nucleus coordinates respectively [1,3].

2 The Born-Oppenheimer Approximation

The Born-Oppenheimer approximation [4] exploits the enormous mass ratio $M_I / m \sim 10^3$ – $10^5$ : nuclei move so slowly compared to electrons that their motions can be treated separately. Thus the total wave function factorises as

\Psi_{\text{total}}(\{\mathbf{r}\},\{\mathbf{R}\}) \;\approx\; \psi_e(\{\mathbf{r}\}\,|\,\{\mathbf{R}\})\;\Phi_n(\{\mathbf{R}\}) \tag{3}

where $\psi_e$ is the electronic wave function (parametrically dependent on the nuclear positions) and $\Phi_n$ is the nuclear wave function.

2.1 The Electronic Schrödinger Equation

By neglecting the nuclear kinetic energy and treating the nuclei as fixed classical charges, the nucleus-nucleus repulsion becomes a constant and we obtain the electronic Schrödinger equation [1]:

\left[ -\sum_i \frac{\hbar^2}{2m} \nabla_{\mathbf{r}_i}^2 + \sum_{i<j} \frac{e^2}{|\mathbf{r}_i - \mathbf{r}_j|} - \sum_{i,I} \frac{Z_I e^2}{|\mathbf{r}_i - \mathbf{R}_I|} \right] \psi_e = E(\{\mathbf{R}\})\,\psi_e \tag{4}

This is solved for fixed nuclear positions $\{\mathbf{R}\}$ . The eigenvalue $E(\{\mathbf{R}\})$ is the electronic energy as a function of nuclear geometry.

2.2 The Nuclear Schrödinger Equation

The nuclear dynamics, operating on a smaller energy scale, are described by:

\left[ -\sum_{I} \frac{\hbar^2}{2M_I} \nabla_{\mathbf{R}_I}^2 + E(\{\mathbf{R}\}) \right] \Phi_n(\{\mathbf{R}\}) = \mathcal{E}\,\Phi_n(\{\mathbf{R}\}) \tag{5}

Here $E(\{\mathbf{R}\})$ serves as the potential for the nuclear dynamics: the Born-Oppenheimer energy surface. The eigenvalue $\mathcal{E}$ is the total energy of the coupled electron-nucleus system [1].

From this point forward, the central problem of electronic structure theory is about solving Eq. (4); finding the ground-state energy and density of the electronic system for a given nuclear configuration. DFT offers an elegant route for that.

3 The Hohenberg-Kohn Theorems (1964)

The electronic Hamiltonian (Eq. 4) can be written more compactly as

\hat{H}_e = \hat{T} + \hat{V}_{ee} + \hat{V}_{\text{ext}} \tag{6}

where $\hat{T}$ is the kinetic energy operator, $\hat{V}_{ee}$ is the electron-electron repulsion, and $\hat{V}_{\text{ext}} = \sum_i v(\mathbf{r}_i)$ is the external (electron-nucleus) potential. In DFT the electron density

\rho(\mathbf{r}) = N \sum_{\sigma_1,\ldots,\sigma_N} \int |\Psi(\mathbf{r}\sigma_1, \mathbf{r}_2\sigma_2, \ldots, \mathbf{r}_N\sigma_N)|^2 \, d\mathbf{r}_2 \cdots d\mathbf{r}_N \tag{7}

is promoted to the fundamental variable [1,3].

3.1 Theorem I: One-to-One Correspondence

Statement: The ground-state electron density $\rho(\mathbf{r})$ uniquely determines the external potential $v(\mathbf{r})$ , up to an additive constant [5].

Since the density determines the potential and the total electron number $N = \int \rho(\mathbf{r})\,d\mathbf{r}$ , it completely fixes the Hamiltonian. Therefore every ground-state observable is a unique functional of $\rho(\mathbf{r})$ .

Proof (Reductio ad Absurdum)

Assume there exist two different external potentials $v_1(\mathbf{r})$ and $v_2(\mathbf{r})$ (differing by more than a constant) that produce the same ground-state density $\rho(\mathbf{r})$ .

These potentials define two different Hamiltonians $\hat{H}_1$ and $\hat{H}_2$ with different ground-state wave functions $\Psi_1$ and $\Psi_2$ and ground-state energies $E_1$ and $E_2$ .

Step 1. Since $\Psi_2$ is not the ground state of $\hat{H}_1$ , the variational principle gives a strict inequality:

E_1 < \langle \Psi_2 | \hat{H}_1 | \Psi_2 \rangle \tag{8}

We decompose $\hat{H}_1 = \hat{H}_2 + (\hat{V}_1 - \hat{V}_2)$ . The difference $\hat{V}_1 - \hat{V}_2$ is a one-body operator whose expectation value depends only on $\rho(\mathbf{r})$ :

E_1 < E_2 + \int \rho(\mathbf{r})\left[v_1(\mathbf{r}) - v_2(\mathbf{r})\right] d\mathbf{r} \tag{9}

Step 2. Swap the labels $1 \leftrightarrow 2$ and apply the identical argument:

E_2 < E_1 + \int \rho(\mathbf{r})\left[v_2(\mathbf{r}) - v_1(\mathbf{r})\right] d\mathbf{r} \tag{10}

Step 3. Add inequalities (9) and (10). The integrals cancel exactly:

E_1 + E_2 < E_2 + E_1 \tag{11}

This is a contradiction. Therefore, two different potentials cannot yield the same ground-state density. Proving then our point : every ground-state observable is a unique functional of $\rho(\mathbf{r})$ .

3.2 Theorem II: The Variational Principle

Statement. There exists a universal functional $F[\rho]$ such that the total energy functional

E_v[\rho] = F[\rho] + \int \rho(\mathbf{r})\,v(\mathbf{r})\,d\mathbf{r} \tag{12}

satisfies the inequality

E_v[\rho] \geq E_v[\rho_0] = E_0 \tag{13}

for all trial densities $\rho(\mathbf{r})$ , where $\rho_0$ is the true ground-state density and $E_0$ the true ground-state energy [1,5].

Demonstration

Define the universal functional:

F[\rho] = E_{\text{kin}}[\rho] + E_{ee}[\rho] \tag{14}

This functional is called “universal” because it does NOT depend on the external potential. It is the same for a single atom or a macroscopic crystal. It can be rigorously defined via the constrained-search formalism of Levy [6]:

F[\rho] = \min_{\Psi \to \rho} \langle \Psi | \hat{T} + \hat{V}_{ee} | \Psi \rangle \tag{15}

where the minimisation is over all antisymmetric $N$ -electron wave functions that yield the density $\rho(\mathbf{r})$ .

From Theorem I, any trial density $\tilde{\rho}(\mathbf{r})$ corresponds to its own external potential and thus its own Hamiltonian and ground-state wave function $\tilde{\Psi}$ . By the standard variational principle of quantum mechanics:

\langle \tilde{\Psi} | \hat{H} | \tilde{\Psi} \rangle \geq E_0 \tag{16}

Writing this in terms of the energy functional:

E_v[\tilde{\rho}] = F[\tilde{\rho}] + \int \tilde{\rho}(\mathbf{r})\,v(\mathbf{r})\,d\mathbf{r} \geq E_0 = E_v[\rho_0] \tag{17}

Therefore, the ground-state density can be obtained by minimising the energy functional $E_v[\rho]$ .

The problematic now The Hohenberg-Kohn theorems guarantee that all ground-state properties are functionals of $\rho(\mathbf{r})$ . The difficulty is that the exact forms of $E_{\text{kin}}[\rho]$ and $E_{ee}[\rho]$ are unknown. The Kohn-Sham scheme provides the crucial practical solution.

4 The Kohn-Sham Equations (1965)

4.1 The Central Idea

Kohn and Sham [7] introduced a brilliant reformulation. Instead of minimising over $\rho(\mathbf{r})$ directly (for which we would need the unknown $E_{\text{kin}}[\rho]$ ), they introduced an auxiliary system of non-interacting electrons. Those are some fictitious particles that, by construction, reproduce the exact ground-state density of the interacting system.

The key insight is: for non-interacting fermions, the kinetic energy is known exactly and the ground state is a single Slater determinant built from one-electron orbitals $\{\varphi_i(\mathbf{r})\}$ .

4.2 Derivation of the Kohn-Sham Equations

Step 1: The Non-Interacting Kinetic Energy

Define the Kohn-Sham kinetic energy for the auxiliary non-interacting system:

T_s[\rho] = \sum_{i=1}^{\text{occ}} \int \varphi_i^*(\mathbf{r}) \left( -\frac{\hbar^2}{2m} \nabla^2 \right) \varphi_i(\mathbf{r})\,d\mathbf{r} \tag{18}

This is the exact kinetic energy of the non-interacting reference system given orbital occupation. It captures the dominant part of the true kinetic energy.

Step 2: The Hartree Energy

The classical electrostatic (Coulomb) repulsion of a charge distribution with itself:

E_H[\rho] = \frac{e^2}{2} \iint \frac{\rho(\mathbf{r})\,\rho(\mathbf{r}')}{|\mathbf{r} - \mathbf{r}'|}\,d\mathbf{r}\,d\mathbf{r}' \tag{19}

Step 3: The Exchange-Correlation Functional

All the many-body complexity is folded into the exchange-correlation functional, defined as the sum of corrections between the true system and the Kohn-Sham reference [1]:

E_{xc}[\rho] = \underbrace{\left( E_{\text{kin}}[\rho] - T_s[\rho] \right)}_{\text{kinetic correction}} + \underbrace{\left( E_{ee}[\rho] - E_H[\rho] \right)}_{\text{interaction correction}} \tag{20}

This is the exact definition. It accounts for:

The difference between the true interacting kinetic energy and the non-interacting one (correlation kinetic energy),
Exchange effects (Pauli exclusion),
Correlation effects beyond Hartree (quantum many-body correlations),
Self-interaction correction (the Hartree term $E_H$ includes an unphysical self-repulsion that $E_{xc}$ must cancel) [3].

Step 4: The Total Energy Functional

The total energy is now partitioned as:

E[\rho] = T_s[\rho] + \int \rho(\mathbf{r})\,v(\mathbf{r})\,d\mathbf{r} + E_H[\rho] + E_{xc}[\rho] \tag{21}

Step 5: Minimisation under the Constraint of Fixed $N$

We minimise $E[\rho]$ with respect to the orbitals $\varphi_i$ , subject to the normalisation constraint $\langle \varphi_i | \varphi_j \rangle = \delta_{ij}$ . Using the method of Lagrange multipliers, we arrive at the Kohn-Sham equations [1,7]:

\boxed{\left[ -\frac{\hbar^2}{2m}\nabla^2 + v_{\text{eff}}(\mathbf{r}) \right] \varphi_i(\mathbf{r}) = \varepsilon_i\,\varphi_i(\mathbf{r})} \tag{22}

where $\varepsilon_i$ are the Kohn-Sham eigenvalues and the effective potential is:

v_{\text{eff}}(\mathbf{r}) = v(\mathbf{r}) + e^2 \int \frac{\rho(\mathbf{r}')}{|\mathbf{r} - \mathbf{r}'|}\,d\mathbf{r}' + v_{xc}(\mathbf{r}) \tag{23}

with the exchange-correlation potential defined as the functional derivative:

v_{xc}(\mathbf{r}) = \frac{\delta E_{xc}[\rho]}{\delta \rho(\mathbf{r})} \tag{24}

Step 6: The Density

The density is reconstructed from the occupied Kohn-Sham orbitals:

\rho(\mathbf{r}) = \sum_{i=1}^{\text{occ}} |\varphi_i(\mathbf{r})|^2 \tag{25}

4.3 The Self-Consistent Cycle

Equations (22), (23), and (25) form a set of self-consistent equations:

Guess an initial density $\rho^{(0)}(\mathbf{r})$ .
Construct $v_{\text{eff}}(\mathbf{r})$ from Eq. (23).
Solve the Kohn-Sham equations (22) to obtain $\{\varphi_i\}$ and $\{\varepsilon_i\}$ .
Compute a new density $\rho^{(n+1)}(\mathbf{r})$ from Eq. (25).
Check convergence: if $|\rho^{(n+1)} - \rho^{(n)}|$ is below a threshold, stop. Otherwise, mix the densities and return to step 2.

This iterative procedure is the computational backbone of every practical DFT code.

5 Approximate Exchange-Correlation Functionals

The exact form of $E_{xc}[\rho]$ is unknown. The success of DFT depends critically on developing accurate approximations. These are organised on what Perdew calls “Jacob’s Ladder” of density functional approximations [8].

5.1 Local Density Approximation (LDA)

The simplest approximation assumes the exchange-correlation energy density at each point equals that of a uniform electron gas with the same local density [1]:

E_{xc}^{\text{LDA}}[\rho] = \int \rho(\mathbf{r})\,\varepsilon_{xc}(\rho(\mathbf{r}))\,d\mathbf{r} \tag{26}

where $\varepsilon_{xc}(\rho)$ is the exchange-correlation energy per particle of a homogeneous electron gas at density $\rho$ . Accurate values of $\varepsilon_{xc}$ are obtained from quantum Monte Carlo simulations of the uniform electron gas [9]. The LDA is exact for uniform systems and surprisingly accurate for many solids where the electron density varies slowly.

5.2 Generalised Gradient Approximation (GGA)

The GGA improves upon LDA by incorporating the first-order gradient of the density [1]:

E_{xc}^{\text{GGA}}[\rho] = \int \rho(\mathbf{r})\,\varepsilon_{xc}\!\left(\rho(\mathbf{r}),\,\nabla\rho(\mathbf{r})\right) d\mathbf{r} \tag{27}

The most widely used GGA functional is PBE (Perdew-Burke-Ernzerhof) [8], which is constructed by imposing exact constraints (correct uniform-gas limit, Lieb-Oxford bound, etc.) without empirical fitting.

5.3 Meta-GGA

The meta-GGA further adds the Laplacian of the density and the kinetic energy density $\tau(\mathbf{r}) = \sum_i |\nabla\varphi_i(\mathbf{r})|^2$ [1]:

E_{xc}^{\text{meta-GGA}}[\rho] = \int \rho(\mathbf{r})\,\varepsilon_{xc}\!\left(\rho,\,\nabla\rho,\,\nabla^2\rho,\,\tau\right) d\mathbf{r} \tag{28}

5.4 Hybrid Functionals and Exact Exchange

Adding higher-order gradients may not efficiently capture quantum effects like the Pauli exclusion. To address this, the exact exchange (EXX) energy is computed from the Kohn-Sham orbitals [1]:

E_{\text{EXX}} = -\frac{e^2}{2} \sum_{i,j}^{\text{occ}} \iint \frac{\varphi_i^*(\mathbf{r})\,\varphi_j^*(\mathbf{r}')\,\varphi_i(\mathbf{r}')\,\varphi_j(\mathbf{r})}{|\mathbf{r} - \mathbf{r}'|}\,d\mathbf{r}\,d\mathbf{r}' \tag{29}

Hybrid functionals mix a fraction of this exact exchange with GGA exchange-correlation, e.g. the B3LYP or PBE0 functionals. The mixing coefficient is tuned to optimally reproduce reference data (cohesive energies, band gaps, etc.) [1].

5.5 The Adiabatic Connection and the RPA

An exact formula relates $E_{xc}$ to the density-density response function via the adiabatic connection [10]:

E_c = -\frac{e^2}{2} \int_0^1 d\lambda \iint d\mathbf{r}\,d\mathbf{r}' \int \frac{d\omega}{2\pi} \;\frac{\chi_\lambda(\mathbf{r}',\mathbf{r},i\omega) - \chi_0(\mathbf{r}',\mathbf{r},i\omega)}{|\mathbf{r} - \mathbf{r}'|} \tag{30}

where $\chi_\lambda$ is the response function of the system with electron-electron interaction scaled by $\lambda$ , and $\chi_0$ is the Kohn-Sham response function. The random phase approximation (RPA) approximates $\chi_\lambda$ and successfully captures van der Waals interactions [1].

6 Extensions of DFT

6.1 Spin-DFT

When spin degrees of freedom are relevant (magnetic materials, open-shell systems), the framework is extended to include the spin density $\mathbf{m}(\mathbf{r})$ as an additional variable alongside $\rho(\mathbf{r})$ [11]. The Kohn-Sham equation becomes spin-dependent, with separate exchange-correlation potentials for spin-up and spin-down channels.

6.2 Time-Dependent DFT (TDDFT)

For systems driven by time-dependent external fields, the Runge-Gross theorem [12] establishes a one-to-one correspondence between the time-dependent density $\rho(\mathbf{r},t)$ and the external potential $v(\mathbf{r},t)$ , given a known initial state. The time-dependent Kohn-Sham equation reads:

i\hbar \frac{\partial}{\partial t} \varphi_i(\mathbf{r},t) = \left[ -\frac{\hbar^2}{2m}\nabla^2 + v_{\text{eff}}(\mathbf{r},t) \right] \varphi_i(\mathbf{r},t) \tag{31}

TDDFT is widely used to simulate excitation dynamics and optical spectra [1].

7 Model Hamiltonians in DFT

The standard Kohn-Sham scheme maps the interacting system to a fully non-interacting reference. As Gori-Giorgi, Toulouse, and Savin have emphasised [3], this can be generalised: one may choose model Hamiltonians that retain partial electron-electron interactions, reducing the burden placed on the exchange-correlation functional.

7.1 Motivation: When KS-DFT Struggles

The Kohn-Sham Slater determinant $\Phi$ has a fundamentally different structure from the true many-body wave function $\Psi$ . This difference is measured by the spherically and system-averaged pair density (APD) $f^\lambda(r_{12})$ , obtained by integrating $|\Psi^\lambda|^2$ over all variables except the inter-electronic distance $r_{12} = |\mathbf{r}_2 - \mathbf{r}_1|$ [3]:

f^\lambda(r_{12}) = \frac{N(N-1)}{2} \sum_{\sigma_1,\ldots,\sigma_N} \int \frac{d\Omega_{r_{12}}}{4\pi} \int |\Psi^\lambda|^2 \, d\mathbf{R}\,d\mathbf{r}_3\cdots d\mathbf{r}_N \tag{32}

where $\mathbf{R} = (\mathbf{r}_1+\mathbf{r}_2)/2$ .

When the KS APD $f^{\lambda=0}$ and the physical APD $f^{\lambda=1}$ are similar (as for H $_2$ at equilibrium), simple functionals work well. But when they differ dramatically (as for stretched H $_2$ , where near-degeneracy effects dominate), the functional $E_{Hxc}[\rho]$ must correct for a huge qualitative difference between model and reality — making universal approximations extremely difficult [3].

7.2 Partially-Interacting Model Systems

The idea is to introduce a modified electron-electron interaction $w^\mu(r_{12})$ controlled by a parameter $\mu$ , satisfying [3]:

$w^\mu(r_{12}) \to 0$ as $\mu \to 0$ (recovers KS)
$w^\mu(r_{12}) \to 1/r_{12}$ as $\mu \to \infty$ (recovers the physical system)

A common choice is the error-function interaction [3]:

w^\mu(r_{12}) = \frac{\operatorname{erf}(\mu\, r_{12})}{r_{12}} \tag{33}

This retains the long-range Coulomb tail but softens the short-range singularity. The corresponding density functional is:

F^\mu[\rho] = \sup_v \left\{ \min_\Psi \langle \Psi | \hat{T} + \hat{W}^\mu + \hat{V} | \Psi \rangle - \int \rho(\mathbf{r})\,v(\mathbf{r})\,d\mathbf{r} \right\} \tag{34}

7.3 The Short-Range Functional

The quantity that must be approximated is the complement to the partially-interacting functional [3]:

\bar{F}^\mu[\rho] = F[\rho;\hat{W}_{ee},\hat{T}] - F[\rho;\hat{W}^\mu,\hat{T}] \tag{35}

This can be decomposed into Hartree and exchange-correlation parts:

\bar{E}_H^\mu[\rho] = \frac{1}{2} \iint \rho(\mathbf{r})\,\rho(\mathbf{r}')\left[\frac{1}{|\mathbf{r}-\mathbf{r}'|} - w^\mu(|\mathbf{r}-\mathbf{r}'|)\right] d\mathbf{r}\,d\mathbf{r}' \tag{36}

\bar{E}_{xc}^\mu[\rho] = \bar{F}^\mu[\rho] - \bar{E}_H^\mu[\rho] \tag{37}

Only $\bar{E}_{xc}^\mu$ requires approximation. Crucially, it describes short-range correlation effects, which are more transferable between systems and thus easier to approximate with local or semi-local functionals [3].

7.4 Adiabatic Connection for the Short-Range Functional

An exact expression for $\bar{F}^\mu[\rho]$ follows from the adiabatic connection [3]:

\bar{F}^\mu[\rho] = \int_\mu^\infty d\mu' \int_0^\infty dr_{12}\; 4\pi r_{12}^2\, f^{\mu'}(r_{12})\,\frac{\partial w^{\mu'}(r_{12})}{\partial \mu'} \tag{38}

Note the integration over the coupling constant starts from the positive value $\mu$ rather than zero — the short-range regime only.

7.5 Large- $\mu$ Asymptotics and the On-Top Pair Density

The large- $\mu$ behaviour of the functionals involves the on-top pair density $f(0)$ (the probability of finding two electrons at the same point in space). Gori-Giorgi and Savin showed [3,13]:

\bar{E}_x^\mu[\rho] \;\xrightarrow{\mu\to\infty}\; -\frac{\pi}{4\mu^2} \int \rho(\mathbf{r})^2\,d\mathbf{r} + \mathcal{O}(\mu^{-4}) \tag{39}

\bar{E}_c^\mu[\rho] \;\xrightarrow{\mu\to\infty}\; \frac{\pi}{\mu^2}\left[f(0) - \frac{1}{4}\int \rho(\mathbf{r})^2\,d\mathbf{r}\right] + \frac{4\sqrt{2\pi}}{3\mu^3}\,f(0) + \mathcal{O}(\mu^{-4}) \tag{40}

These exact asymptotics constrain the construction of approximate short-range functionals. The on-top pair density $f(0)$ can itself be approximated via the LDA (transfer from the uniform electron gas) or self-consistently from $f^\mu(0)$ [3].

7.6 Practical Advantage

The model wave function $\Psi^\mu$ is multideterminantal but, since $w^\mu$ is much weaker than $1/r_{12}$ , far fewer Slater determinants are needed for an accurate expansion. Gori-Giorgi et al. demonstrated for He and Be atoms that at $\mu = 1$ , a handful of configurations already achieve errors below 0.005 Hartree — a dramatic reduction compared to the full Coulomb problem [3].

This approach — combining a multideterminantal wave function for the long-range part with a local/semi-local functional for the short range — provides a systematic way to improve DFT for strongly-correlated and near-degenerate systems [3].

8 Summary of Part 1

We have built the mathematical framework of DFT from the ground up:

Layer	Key Equations	What it achieves
Many-body Hamiltonian	Eqs. (1)–(2)	Full quantum-mechanical description
Born-Oppenheimer	Eqs. (3)–(5)	Decouples electronic and nuclear motion
Hohenberg-Kohn I	Eqs. (8)–(11)	$\rho(\mathbf{r}) \leftrightarrow v(\mathbf{r})$ bijection proved by contradiction
Hohenberg-Kohn II	Eqs. (12)–(17)	Variational principle for $E[\rho]$
Kohn-Sham	Eqs. (18)–(25)	Exact reformulation as non-interacting eigenvalue problem
XC Approximations	Eqs. (26)–(30)	LDA → GGA → meta-GGA → hybrids → RPA
Model Hamiltonians	Eqs. (33)–(40)	Partially-interacting reference for strong correlation

In Part 2, we move from equations to implementations: how these ideas are realised in the plane-wave pseudopotential framework of Quantum ESPRESSO and the KKR-Green’s function method of AKAI KKR, with hands-on examples.

References

[1] Y. Nomura and R. Akashi, “Density functional theory,” Encyclopedia of Condensed Matter Physics, 2nd ed., Elsevier, 2024, pp. 867–878. doi: 10.1016/B978-0-323-90800-9.00148-7

[2] W. M. C. Foulkes, L. Mitas, R. J. Needs, and G. Rajagopal, “Quantum Monte Carlo simulations of solids,” Rev. Mod. Phys. 73, 33–83 (2001).

[3] P. Gori-Giorgi, J. Toulouse, and A. Savin, “Model Hamiltonians in Density Functional Theory,” CRM Proceedings and Lecture Notes, vol. 41, American Mathematical Society, 2006.

[4] M. Born and R. Oppenheimer, “Zur Quantentheorie der Molekeln,” Ann. Phys. 389, 457–484 (1927).

[5] P. Hohenberg and W. Kohn, “Inhomogeneous Electron Gas,” Phys. Rev. 136, B864–B871 (1964).

[6] M. Levy, “Universal variational functionals of electron densities, first-order reduced density matrices, and natural spin-orbitals and solution of the v-representability problem,” Proc. Natl. Acad. Sci. U.S.A. 76, 6062–6065 (1979).

[7] W. Kohn and L. J. Sham, “Self-Consistent Equations Including Exchange and Correlation Effects,” Phys. Rev. 140, A1133–A1138 (1965).

[8] J. P. Perdew, K. Burke, and M. Ernzerhof, “Generalized Gradient Approximation Made Simple,” Phys. Rev. Lett. 77, 3865–3868 (1996).

[9] D. M. Ceperley and B. J. Alder, “Ground State of the Electron Gas by a Stochastic Method,” Phys. Rev. Lett. 45, 566–569 (1980).

[10] D. Langreth and J. Perdew, “The exchange-correlation energy of a metallic surface,” Solid State Commun. 17, 1425–1429 (1975).

[11] U. von Barth and L. Hedin, “A local exchange-correlation potential for the spin polarized case,” J. Phys. C: Solid State Phys. 5, 1629–1642 (1972).

[12] E. Runge and E. K. U. Gross, “Density-Functional Theory for Time-Dependent Systems,” Phys. Rev. Lett. 52, 997–1000 (1984).

[13] P. Gori-Giorgi and A. Savin, “Properties of short-range and long-range correlation energy density functionals from electron-electron coalescence,” Phys. Rev. A 73, 032506 (2006).