arXiv subjects

Dario Rocca

Publications and source records attributed to Dario Rocca.

18 recordsLinked to original sources

arXiv2026

RLEASE: Reinforcement Learning Efficient Active Space Engine

Selecting the active space for multireference electronic-structure calculations is a long-standing bottleneck that often requires expert chemical intuition and costly trial-and-error. We introduce RLEASE (Reinforcement Learning Efficient Active Space Engine), a low-cost method for automatic, geometry-dependent active-space selection. A neural network predicts per-orbital diagnostic scores ($\hat{s}_{1}$) from inexpensive Hartree-Fock orbital descriptors, and a learned threshold partitions orbitals into active and inactive sets. The threshold policy is optimized with proximal policy optimization, using the discrepancy between sc-NEVPT2 energies computed with the selected active space and DMRG reference energies as the reward. After training, the same RLEASE-selected active spaces can be used with multireference perturbation theory or composite coupled-cluster energy estimators. Despite being trained on a small set of molecules and geometries, RLEASE transfers to chemically diverse test systems, producing compact active spaces and competitive potential-energy surfaces relative to established entropy-based selectors. Because deployment requires only inexpensive orbital descriptors and neural-network inference, RLEASE enables high-throughput multireference workflows without molecule-specific retraining or target-system pilot DMRG calculations.

physics.chem-ph↗

arXiv2026

MLIPilot: LLM-Driven Auto-Research for Machine-Learned Interatomic Potentials

Constructing production-quality machine-learned interatomic potentials (MLIPs) requires balancing accuracy, dynamical stability, and computational throughput under constraints that are not captured by a single training loss. We introduce MLIPilot, an auto-research framework in which tool-calling large language models propose hypotheses, edit MLIP training code, launch HPC jobs, and accept or revert changes using a fixed, physically constrained scorecard. We evaluate MLIPilot on MACE potential optimization using both commercial and open-weight LLM agents, including GPT-5.5, GPT-4.1, Mistral-24B, and Qwen3-32B. The benchmarks span molecular and periodic settings: a QM7-derived dataset for which we generated B3LYP/6-31G(d) energies and forces, and a Cu EMT dataset with periodic copper supercells labeled by ASE's Effective Medium Theory calculator. Across these benchmarks, the strongest agents move initially constraint-violating baselines to accepted models by discovering useful training strategies, including output normalization, loss-function changes, progressive training schedules, and model-capacity adjustments. These results suggest that LLM agents can serve as autonomous operators for scientific machine-learning workflows when their search is constrained by domain-specific validation criteria, shifting part of MLIP development from manual trial-and-error toward auditable, automated experimentation.

physics.chem-ph↗

arXiv2025

Quantum simulation of carbon capture in periodic metal-organic frameworks

Carbon capture is vital for decarbonizing heavy industries such as steel and chemicals. Metal-organic frameworks (MOFs), with their high surface area and structural tunability, are promising materials for CO2 capture. This study focuses on Fe-MOF-74, a magnetic Mott insulator with exposed metal sites that enhance CO2 adsorption. Its strongly correlated electronic structure challenges standard DFT methods, which often yield inconsistent predictions. We initially benchmark adsorption energies using various DFT functionals, revealing substantial variability and underscoring the need for more accurate approaches, such as those provided by quantum computing. However, practical quantum algorithms are far less established for simulations of periodic materials, particularly when the plane-wave basis set, often comprising tens of thousands of basis vectors, is used. To address this, we employ an active space reduction strategy based on Wannier functions and natural orbital selection. Localized orbitals around the adsorption site are identified, and MP2 natural orbitals are used to improve convergence of correlation energies. Adsorption energies are then computed using a quantum number-preserving ansatz within the variational quantum eigensolver framework. In addition to classical simulations, we conduct quantum experiments using the sample-based quantum diagonalization method. Although current hardware limits the size of feasible simulations, our approach offers a more efficient and scalable path forward. These results advance the applicability of quantum algorithms to realistic models of carbon capture and periodic materials more broadly.

quant-ph↗

arXiv2025

Improving the runtime of quantum phase estimation for chemistry through basis set optimization

Quantum phase estimation (QPE) is a promising quantum algorithm for obtaining molecular ground-state energies with chemical accuracy. However, its computational cost, dominated by the Hamiltonian 1-norm $\lambda$ and the cost of the block encoding, scales at least quadratically with the number of molecular orbitals, making it challenging to incorporate dynamic correlation by enlarging the active space. In this work, we investigate two strategies to mitigate this cost through the optimization of the basis set. First, we investigate whether adjusting the coefficients of Gaussian basis functions can minimize the 1-norm while preserving the accuracy of the ground state energy. Although this method leads to a reduction in the 1-norm up to 10%, this reduction is system-dependent and diminishes with increasing molecular size. Second, we demonstrate that employing a large-basis-set frozen natural orbital (FNO) strategy results in a substantial reduction in QPE resources without compromising accuracy. We study a dataset of 58 small organic molecules and the dissociation curve of N2, and demonstrate that an active space constructed from orbitals derived from larger basis sets captures correlation effects more effectively. This approach yields up to an 80% reduction in the 1-norm $\lambda$ and also leads to a 55% reduction in the number of orbitals. Our results highlight that improving the quality, not just the size, of the orbital basis is a viable strategy for extending QPE to include dynamical correlation, making progress toward scalable and chemically accurate quantum simulations with tractable resource requirements.

quant-ph↗

arXiv2024

Assessing the query complexity limits of quantum phase estimation using symmetry aware spectral bounds

The computational cost of quantum algorithms for physics and chemistry is closely linked to the spectrum of the Hamiltonian, a property that manifests in the necessary rescaling of its eigenvalues. The typical approach of using the 1-norm as an upper bound to the spectral norm to rescale the Hamiltonian suits the most general case of bounded Hermitian operators but neglects the influence of symmetries commonly found in chemical systems. In this work, we introduce a hierarchy of symmetry-aware spectral bounds that provide a unified understanding of the performance of quantum phase estimation algorithms using block-encoded electronic structure Hamiltonians. We present a variational and numerically tractable method for computing these bounds, based on orbital optimization, to demonstrate that the computed bounds are smaller than conventional spectral bounds for a variety of molecular benchmark systems. We also highlight the unique analytical and numerical scaling behavior of these bounds in the thermodynamic and complete basis set limits. Our work shows that there is room for improvement in reducing the 1-norm, not yet achieved through methods like double factorization and tensor hypercontraction, but highlights potential challenges in improving the performance of current quantum algorithms beyond small constant factors through 1-norm reduction techniques alone.

quant-ph↗

arXiv2024

Enhancing initial state overlap through orbital optimization for faster molecular electronic ground-state energy estimation

The quantum phase estimation algorithm stands as the primary method for determining the ground state energy of a molecular electronic Hamiltonian on a quantum computer. In this context, the ability to initialize a classically tractable state that has a strong overlap with the desired ground state is critical as it directly affects the runtime of the algorithm. However, several numerical studies have shown that this overlap decays exponentially with system size. In this work, we demonstrate that this decay can be alleviated by optimizing the molecular orbital basis, for an initial state constructed from a single Slater determinant. We propose a practical method to achieve this optimization without knowledge of the true molecular ground state and test this method numerically. By comparing the resulting optimized orbitals to the natural orbitals, we find improved overlap. Specifically, for four iron-sulfur molecules, which are known to suffer from the mentioned decay, we show that our method yields one to two orders of magnitude improvement compared to localized molecular orbitals.

quant-ph↗

arXiv2024

Reducing the runtime of fault-tolerant quantum simulations in chemistry through symmetry-compressed double factorization

Quantum phase estimation based on qubitization is the state-of-the-art fault-tolerant quantum algorithm for computing ground-state energies in chemical applications. In this context, the 1-norm of the Hamiltonian plays a fundamental role in determining the total number of required iterations and also the overall computational cost. In this work, we introduce the symmetry-compressed double factorization (SCDF) approach, which combines a compressed double factorization of the Hamiltonian with the symmetry shift technique, significantly reducing the 1-norm value. The effectiveness of this approach is demonstrated numerically by considering various benchmark systems, including the FeMoco molecule, cytochrome P450, and hydrogen chains of different sizes. To compare the efficiency of SCDF to other methods in absolute terms, we estimate Toffoli gate requirements, which dominate the execution time on fault-tolerant quantum computers. For the systems considered here, SCDF leads to a sizeable reduction of the Toffoli gate count in comparison to other variants of double factorization or even tensor hypercontraction, which is usually regarded as the most efficient approach for qubitization.

quant-ph↗

arXiv2021

Assessing the Accuracy of Machine Learning Thermodynamic Perturbation Theory: Density Functional Theory and Beyond

Machine learning thermodynamic perturbation theory (MLPT) is a promising approach to compute finite temperature properties when the goal is to compare several different levels of ab initio theory and/or to apply highly expensive computational methods. Indeed, starting from a production molecular dynamics trajectory, this method can estimate properties at one or more target levels of theory from only a small number of additional fixed-geometry calculations, which are used to train a machine learning model. However, as MLPT is based on thermodynamic perturbation theory (TPT), inaccuracies might arise when the starting point trajectory samples a configurational space which has a small overlap with that of the target approximations of interest. By considering case studies of molecules adsorbed in zeolites and several different density functional theory approximations, in this work we assess the accuracy of MLPT for ensemble total energies and enthalpies of adsorption. The problematic cases that were found are analyzed and it is shown that, even without knowing exact reference results, pathological cases for MLPT can be detected by considering a coefficient that measures the statistical imbalance induced by the TPT reweighting. For the most pathological examples we recover target level results within chemical accuracy by applying a machine learning-based Monte Carlo (MLMC) resampling. Finally, based on the ideas developed in this work, we assess and confirm the accuracy of recently published MLPT-based enthalpies of adsorption at the random phase approximation level, whose high computational cost would completely hinder a direct molecular dynamics simulation.

cond-mat.mtrl-sci↗

arXiv2020

Self-consistent potential correction for charged periodic systems

Supercell models are often used to calculate the electronic structure of local perturbations from the ideal periodicity in the bulk or on the surface of a crystal or in wires. When the defect or adsorbent is charged, a jellium counter charge is applied to maintain overall neutrality, but the interaction of the artificially repeated charges has to be corrected, both in the total energy and in the one-electron eigenvalues and eigenstates. This becomes paramount in slab or wire calculations, where the jellium counter charge may induce spurious states in the vacuum. We present here a self-consistent potential correction scheme and provide successful tests of it for bulk and slab calculations.

cond-mat.mtrl-sci↗

arXiv2020

Hybrid localized graph kernel for machine learning energy-related properties of molecules and solids

Nowadays, the coupling of electronic structure and machine learning techniques serves as a powerful tool to predict chemical and physical properties of a broad range of systems. With the aim of improving the accuracy of predictions, a large number of representations for molecules and solids for machine learning applications has been developed. In this work we propose a novel descriptor based on the notion of molecular graph. While graphs are largely employed in classification problems in cheminformatics or bioinformatics, they are not often used in regression problem, especially of energy-related properties. Our method is based on a local decomposition of atomic environments and on the hybridization of two kernel functions: a graph kernel contribution that describes the chemical pattern and a Coulomb label contribution that 1encodes finer details of the local geometry. The accuracy of this new kernel method in energy predictions of molecular and condensed phase systems is demonstrated by considering the popular QM7 and BA10 datasets. These examples show that the hybrid localized graph kernel outperforms traditional approaches such as, for example, the smooth overlap of atomic positions (SOAP) and the Coulomb matrices.

physics.chem-ph↗

arXiv2020

Substrate screening approach for quasiparticle energies of two-dimensional interfaces with lattice mismatch

Two-dimensional (2D) materials are outstanding platforms for exotic physics and emerging applications by forming interfaces. In order to efficiently take into account the substrate screening in the quasiparticle energies of 2D materials, several theoretical methods have been proposed previously, but only applicable to interfaces of two systems' lattice constants with certain integer proportion. In this work, we analytically showed the equivalence and distinction among different approximate methods for substrate dielectric matrices. We evaluated the accuracy of these methods, by applying them to calculate quasi-particle energies of hexagonal boron nitride interface systems (heterojunctions and bilayers), and compared with explicit interface calculations. Most importantly, we developed an efficient and accurate interpolation technique for dielectric matrices that made quasiparticle energy calculations possible for arbitrarily mismatched interfaces, which is extremely important for practical applications.

cond-mat.mtrl-sci↗

arXiv2019

Bridging molecular dynamics and correlated wave-function methods for accurate finite-temperature properties

We introduce the "selPT" perturbative approach, based on ab initio molecular dynamics (AIMD), for computing accurate finite-temperature properties by efficiently using correlated wave-function methods. We demonstrate the power of the method by computing prototypical molecular enthalpies of adsorption in zeolite (CH$_4$ and CO$_2$ on protonated chabazite at 300~K) using the random phase approximation. Results are in excellent agreement with experiment. The improved accuracy provided by selPT represents a crucial step towards the goal of truly quantitative AIMD prediction of experimental observables at finite temperature.

cond-mat.mtrl-sci↗

arXiv2019

Dimensionality and Anisotropicity Dependence of Radiative Recombination in Nanostructured Phosphorene

The interplay between dimensionality and anisotropicity leads to intriguing optoelectronic properties and exciton dynamics in low dimensional semiconductors. In this study we use nanostructured phosphorene as a prototypical example to unfold such complex physics and develop a general first-principles framework to study exciton dynamics in low dimensional systems. Specifically we derived the radiative lifetime and light emission intensity from 2D to 0D systems based on many-body perturbation theory, and investigated the dimensionality and anisotropicity effects on radiative recombination lifetime both at 0 K and finite temperature, as well as polarization and angle dependence of emitted light. We show that the radiative lifetime at 0 K increases by an order of $10^3$ with the lowering of one dimension (i.e. from 2D to 1D nanoribbons or from 1D to 0D quantum dots). We also show that obtaining the radiative lifetime at finite temperature requires accurate exciton dispersion beyond the effective mass approximation. Finally, we demonstrate that monolayer phosphorene and its nanostructures always emit linearly polarized light consistent with experimental observations, different from in-plane isotropic 2D materials like MoS2 and h-BN that can emit light with arbitrary polarization, which may have important implications for quantum information applications.

cond-mat.mes-hall↗

arXiv2017

First-principles Engineering of Charged Defects for Two-dimensional Quantum Technologies

Charged defects in 2D materials have emerging applications in quantum technologies such as quantum emitters and quantum computation. Advancement of these technologies requires rational design of ideal defect centers, demanding reliable computation methods for quantitatively accurate prediction of defect properties. We present an accurate, parameter-free and efficient procedure to evaluate quasiparticle defect states and thermodynamic charge transition levels of defects in 2D materials. Importantly, we solve critical issues that stem from the strongly anisotropic screening in 2D materials, that have so far precluded accurate prediction of charge transition levels in these materials. Using this procedure, we investigate various defects in monolayer hexagonal boron nitride (h-BN) for their charge transition levels, stable spin states and optical excitations. We identify $C_BN_V$ (nitrogen vacancy adjacent to carbon substitution of boron) to be the most promising defect candidate for scalable quantum bit and emitter applications.

cond-mat.mtrl-sci↗

arXiv2016

Dielectric matrix formulation of correlation energies in the Random Phase Approximation (RPA): inclusion of exchange effects

Starting from the general expression for the ground state correlation energy in the adiabatic connection fluctuation dissipation theorem (ACFDT) framework, it is shown that the dielectric matrix formulation, which is usually applied to calculate the direct random phase approximation (dRPA) correlation energy, can be used for alternative RPA expressions including exchange effects. Within this famework, the ACFDT analog of the second order screened exchange (SOSEX) approximation leads to a logarithmic formula for the correlation energy similar to the direct RPA expression. Alternatively, the contribution of the exchange can be included in the kernel used to evaluate the response functions. In this case the use of an approximate kernel is crucial to simplify the formalism and to obtain a correlation energy in logarithmic form. Technical details of the implementation of these methods are discussed and it is shown that one can take advantage of density fitting or Cholesky decomposition techniques to improve the computational efficiency; a discussion on the numerical quadrature made on the frequency variable is also provided. A series of test calculations on atomic correlation energies and molecular reaction energies shows that exchange effects are instrumental to improve over direct RPA results.

physics.chem-ph↗

arXiv2014

turboTDDFT 2.0 - Hybrid functionals and new algorithms within time-dependent density-functional perturbation theory

We present a new release of the turboTDDFT code featuring an implementation of hybrid functionals, a recently introduced pseudo-Hermitian variant of the Liouville-Lanczos approach to time-dependent density-functional perturbation theory, and a newly developed Davidson-like algorithm to compute selected interior eigenvalues/vectors of the Liouvillian super-operator. Our implementation is thoroughly validated against benchmark calculations performed on the cyanin (C$_{21}$O$_{11}$H$_{21}$) molecule using the Gaussian09 and turboTDDFT 1.0 codes.

physics.chem-ph↗

arXiv2008

Turbo charging time-dependent density-functional theory with Lanczos chains

We introduce a new implementation of time-dependent density-functional theory which allows the \emph{entire} spectrum of a molecule or extended system to be computed with a numerical effort comparable to that of a \emph{single} standard ground-state calculation. This method is particularly well suited for large systems and/or large basis sets, such as plane waves or real-space grids. By using a super-operator formulation of linearized time-dependent density-functional theory, we first represent the dynamical polarizability of an interacting-electron system as an off-diagonal matrix element of the resolvent of the Liouvillian super-operator. One-electron operators and density matrices are treated using a representation borrowed from time-independent density-functional perturbation theory, which permits to avoid the calculation of unoccupied Kohn-Sham orbitals. The resolvent of the Liouvillian is evaluated through a newly developed algorithm based on the non-symmetric Lanczos method. Each step of the Lanczos recursion essentially requires twice as many operations as a single step of the iterative diagonalization of the unperturbed Kohn-Sham Hamiltonian. Suitable extrapolation of the Lanczos coefficients allows for a dramatic reduction of the number of Lanczos steps necessary to obtain well converged spectra, bringing such number down to hundreds (or a few thousands, at worst) in typical plane-wave pseudopotential applications. The resulting numerical workload is only a few times larger than that needed by a ground-state Kohn-Sham calculation for a same system. Our method is demonstrated with the calculation of the spectra of benzene, C$_{60}$ fullerene, and of chlorofyll a.

cond-mat.mtrl-sci↗

arXiv2007

Weak binding between two aromatic rings: feeling the van der Waals attraction by quantum Monte Carlo methods

We report a systematic study of the weak chemical bond between two benzene molecules. We first show that it is possible to obtain a very good description of the C_2 dimer and the benzene molecule, by using pseudopotentials for the chemically inert 1s electrons, and a resonating valence bond wave function as a variational ansatz, expanded on a relatively small Gaussian basis set. We employ an improved version of the stochastic reconfiguration technique to optimize the many-body wave function, which is the starting point for highly accurate simulations based on the lattice regularized diffusion Monte Carlo (LRDMC) method. This projection technique provides a rigorous variational upper bound for the total energy, even in the presence of pseudopotentials, and allows to improve systematically the accuracy of the trial wave function, which already yields a large fraction of the dynamical and non-dynamical electron correlation. We show that the energy dispersion of two benzene molecules in the parallel displaced geometry is significantly deeper than the face-to-face configuration. However, contrary to previous studies based on post Hartree-Fock methods, the binding energy remains weak (~ 2 kcal/mol) also in this geometry, and its value is in agreement with the most accurate and recent experimental findings.

cond-mat.other↗