Deriving density-matrix functionals for excited states

Julia Liebert, Christian Schilling

Abstract

This work initiates the recently proposed $w$-ensemble one-particle reduced density matrix functional theory ($w$-RDMFT) by deriving the first functional approximations and illustrating how excitation energies can be calculated in practice. By examining the symmetric Hubbard dimer and a homogeneous Bose gas, we reveal that the gradient of the functional diverges repulsively at the boundary of its domain, extending the "Bose-Einstein condensation force" to excited states. Our findings highlight the physical relevance of the generalized exclusion principle and the challenges posed by the "curse of universality" in functional theories.

Introduction

Density Functional Theory (DFT) has long been the cornerstone of ground-state quantum property calculations in chemistry and condensed matter physics. However, accurately predicting excited states—the behavior of a system when it absorbs energy—remains a significant challenge. Reduced Density Matrix Functional Theory (RDMFT), which utilizes the one-particle reduced density matrix (1RDM) as its primary variable, offers a promising alternative.

Recently, a $w$-ensemble version of RDMFT was proposed specifically for excited states. However, until now, it lacked explicit "functionals"—the mathematical maps from density matrices to energy. This paper bridges that gap by deriving the first analytical functionals for two fundamental systems, providing a roadmap for calculating excitation energies in complex materials.

Methodological Approaches

The researchers utilized three conceptually distinct mathematical strategies to derive their functionals:

  1. Levy-Lieb Constrained Search: A rigorous minimization procedure over all possible quantum states that yield a specific 1RDM. This was applied to the symmetric Hubbard dimer, a two-site model that serves as a building block for complex lattice models.
  2. Legendre-Fenchel Transformation: A tool from convex analysis used to relate the total energy of a system to its universal functional. This was employed for the homogeneous Bose gas.
  3. $w$-ensemble $v$-representability: This involves inverting the mapping between kinetic energy operators and the resulting momentum occupation numbers.

By demonstrating these different routes, the authors show that $w$-RDMFT is a computationally viable framework rather than just a theoretical curiosity.

Key Insights: The BEC Force and Universality

The "BEC Force" in Excited States

A major discovery in ground-state RDMFT was the "Bose-Einstein Condensation (BEC) force"—a geometric effect that prevents all bosons from occupying the same state (quantum depletion). This paper proves that this force persists in excited states. The gradient of the functional diverges as $1/\sqrt{D}$ (where $D$ is the distance to the boundary), acting as a repulsive barrier that prevents complete condensation even when the system is energized.

The Curse of Universality

The authors identify a subtle challenge: the "curse of universality." As system parameters change, energy levels can cross. This causes the "universal" functional to become non-analytic, meaning it must be defined piecewise in different "cells" of its domain. This suggests that seeking a single, smooth global functional for all possible systems might be mathematically impossible.

Results and Validation

The authors derived explicit, closed-form expressions for the $w$-ensemble functionals of the symmetric Hubbard dimer and the homogeneous Bose gas in the Bogoliubov regime. To validate these new functionals, the calculated excitation energies were compared against exact diagonalization of the Hamiltonian, showing perfect agreement. These findings emphasize that many physical phenomena, like the generalized exclusion principle for bosons and fermions, are consequences of the underlying geometry of reduced quantum states.

Conclusion and Future Outlook

While the current derivations for the Bose gas rely on the Bogoliubov approximation (assuming small quantum depletion), they provide a foundation for future work in strongly interacting systems. The methodologies are also directly applicable to fermionic systems, which are crucial for understanding electronic excitations in molecules and solids. As the complexity of state degeneracies increases with higher excitations ($r > 2$), the mathematical framework established here will be essential for navigating the non-analytic nature of universal functionals.

Summary updated

Abstract

We initiate the recently proposed w <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"> <mml:mi>w</mml:mi> </mml:math> -ensemble one-particle reduced density matrix functional theory ( w <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"> <mml:mi>w</mml:mi> </mml:math> -RDMFT) by deriving the first functional approximations and illustrate how excitation energies can be calculated in practice. For this endeavour, we first study the symmetric Hubbard dimer, constituting the building block of the Hubbard model, for which we execute the Levy-Lieb constrained search. Second, due to the particular suitability of w <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"> <mml:mi>w</mml:mi> </mml:math> -RDMFT for describing Bose-Einstein condensates, we demonstrate three conceptually different approaches for deriving the universal functional in a homogeneous Bose gas for arbitrary pair interaction in the Bogoliubov regime. Remarkably, in both systems the gradient of the functional is found to diverge repulsively at the boundary of the functional’s domain, extending the recently discovered Bose-Einstein condensation force to excited states. Our findings highlight the physical relevance of the generalized exclusion principle for fermionic and bosonic mixed states and the curse of universality in functional theories.
BibTeX
@inproceedings{liebert2023deriving,
  author = {Julia Liebert and Christian Schilling},
  title = {{Deriving density-matrix functionals for excited states}},
  booktitle = {SciPost},
  year = {2023},
  doi = {10.21468/scipostphys.14.5.120},
}
Figure