https://arxiv.org/api/Wh1bDZoPYDtE8Vjj4vtpo73B15Y 2026-03-22T17:30:58Z 26775 75 15 http://arxiv.org/abs/2603.15669v1 2026-03-13T06:53:13Z

Density functional theory (DFT) is a cornerstone of computational chemistry and materials science, but its computational cost limits its use in large-scale and high-throughput applications. While machine learning has accelerated energy prediction for specific molecular classes, transferable prediction of electron density across diverse chemical spaces remains challenging. Here, we present a universal framework based on Fourier Neural Operators (FNOs) that directly learns the mapping from external potentials to electron density distributions. Unlike conventional approaches that rely on explicit atomic orbitals, basis sets, or handcrafted descriptors, the proposed method captures global electronic interactions and long-range correlations through operator learning in the spatial-frequency domain. Trained on datasets spanning multiple elements and molecular geometries, the model achieves zero-shot generalization to entirely unseen molecular systems and accurately predicts their electron densities without retraining. This transferability arises from the intrinsic ability of FNOs to represent global structure in continuous fields. Our work establishes neural operator learning as a promising route for fast, accurate, and transferable electronic structure prediction, with potential applications in high-throughput screening and chemical space exploration.

2026-03-13T06:53:13Z 30 pages, 9 figures Yingdi Jin Xinming Qin Ruichen Liu Jie Liu Zhenyu Li Jinlong Yang http://arxiv.org/abs/2602.07075v4 2026-03-13T03:48:49Z

Chemical large language models (LLMs) predominantly rely on explicit Chain-of-Thought (CoT) in natural language to perform complex reasoning. However, chemical reasoning is inherently continuous and structural, and forcing it into discrete linguistic tokens introduces a fundamental representation mismatch that constrains both efficiency and performance. We introduce LatentChem, a latent reasoning interface that decouples chemical computation from textual generation, enabling models to perform multi-step reasoning directly in continuous latent space while emitting language only for final outputs. Remarkably, we observe a consistent emergent behavior: when optimized solely for task success, models spontaneously internalize reasoning, progressively abandoning verbose textual derivations in favor of implicit latent computation. This shift is not merely stylistic but computationally advantageous. Across diverse chemical reasoning benchmarks, LatentChem achieves a 59.88\% non-tie win rate over strong CoT-based baselines on ChemCoTBench, while delivering a 10.84$\times$ average reduction in reasoning overhead. Our results provide empirical evidence that chemical reasoning is more naturally and effectively realized as continuous latent dynamics rather than discretized linguistic trajectories.

2026-02-06T01:28:27Z Xinwu Ye Yicheng Mao Jia Zhang Yimeng Liu Li Hao Fang Wu Zhiwei Li Yuxuan Liao Zehong Wang Yingcheng Wu Zhiyuan Liu Zhenfei Yin Li Yuan Philip Torr Huan Sun Xiangxiang Zeng Mengdi Wang Le Cong Shenghua Gao Xiangru Tang http://arxiv.org/abs/2603.12537v1 2026-03-13T00:36:02Z

Electron confinement within a small volume is intriguing as a realization of the particle-in-a-box system, which appears in every quantum mechanics textbook. While the electron confinement is readily imaginable in solid-state systems, it also occurs in liquids, where the local voids in the liquid serve as confining "boxes." Confinement within these flexible cavities in liquids is expected to differ fundamentally from that in solids. Here, we experimentally investigate the electrons confined in liquid water, which are called hydrated electrons, using transient two-dimensional electronic spectroscopy. Our experiment reveals the large nonuniformity of the shape and the size of hydrated electrons with significant fluctuation at the timescale shorter than 30 fs.

2026-03-13T00:36:02Z Korenobu Matsuzaki Hikaru Kuramochi Tahei Tahara http://arxiv.org/abs/2603.12534v1 2026-03-13T00:26:01Z

Biological polymers, such as intrinsically disordered proteins, play a central role in cellular biology, including mediating phase separation and controlling activity of biological condensates. The physical properties and functions of biopolymers are determined by their residue sequence. Recently, significant computational and theoretical efforts have been devoted to characterizing the combinatorially complex sequence dependence of biopolymer phase diagrams. Here, we quantitatively show that monomer accessibility is central to determining the strength of pair interactions. We formulate an analytical perturbative approach, phenomenologically precluding two polymers' centers of mass from overlapping within a correlation hole. This theory yields the correction to the strength of mean-field interactions in terms of a residue-accessibility parameter (RAP), which accounts for the limited availability of inner monomers to interactions. Despite the simplicity of the approach, RAP rationalizes the variations in critical temperatures found in extensive Monte-Carlo simulations for thousands of two-letter polymer solutions of varying length and sequence. RAP may thus be effective for deciphering the polymer-sequence dependence of phase diagrams given any polymer length, set of monomer types, and polymer mixtures.

2026-03-13T00:26:01Z 10 pages, 4 figures J. Pedro de Souza Benjamin Sorkin Amala Akkiraju Athanassios Z. Panagiotopoulos Howard A. Stone http://arxiv.org/abs/2603.12373v1 2026-03-12T18:51:33Z

Plasmonic nanocavities are a promising platform for strong light-matter coupling and enhanced spectroscopies at the single-molecule level. These nanoscale environments are challenging to model due to their strongly multimodal character and short cavity lifetimes. Herein, we study the effects of these environments using real-time nuclear-electronic orbital time-dependent density functional theory (RT-NEO-TDDFT) coupled to multiple classical cavity modes in a manner that includes cavity loss. In RT-NEO-TDDFT, the quantum mechanical densities of all electrons and specified nuclei, typically protons, are propagated in real time. We show that a cavity with many modes at different frequencies can be used to probe and modify the nuclear-electronic quantum dynamics of chemical systems. Ultrafast excited-state proton transfer reactions can be probed through the time- and energy-resolved cavity emission of a multimode cavity. Under strong coupling conditions, the cavity can modify the dynamics, in some cases suppressing proton transfer and exhibiting Rabi-like oscillations of the cavity emission due to polariton formation. Utilizing the spectral density for an experimentally relevant nanoparticle-on-mirror single-molecule cavity, we show that an excited-state proton transfer system can evolve into resonance with the cavity even when initially out of resonance with the dominant cavity peak. In this case, tuning the dominant cavity peak to be resonant with the electronic transition leads to polariton formation for a small collection of molecules. The RT-NEO framework with multimode cavities enables the efficient simulation of chemical reactions in physically realistic electromagnetic environments, providing fundamental insights into the dynamics and associated spectroscopic signatures.

2026-03-12T18:51:33Z Jonathan H. Fetherolf Tao E. Li Sharon Hammes-Schiffer http://arxiv.org/abs/2603.12233v1 2026-03-12T17:52:15Z

Solving the intricate quantum behavior of interacting particles is key to unlocking the mysteries of condensed matter, but capturing their complex correlations across different scales remains a monumental challenge. We introduce a neural network framework that overcomes this barrier by modeling the full quantum wavefunction of a system, including electrons, nuclei and muons, directly capturing the full quantum effects beyond the Born-Oppenheimer approximation. The neural network approximates joint wavefunction of different interacting particles with a rigorous handling of permutation invariance, enabling simultaneous treatment of nuclear quantum effects and electron-nucleus-muon couplings without explicit excited states. Validated on molecular systems, this approach offers a computationally feasible way to model full quantum phenomena in complex many-body systems, establishing a direct connection between fundamental particle properties and emergent material behavior.

2026-03-12T17:52:15Z Pengzhen Cai Yubing Qian Li Deng Weizhong Fu Lei Yang Zhiyu Sun Xin-Zheng Li En-Ge Wang Liangwen Chen Weiluo Ren Ji Chen http://arxiv.org/abs/2603.12160v1 2026-03-12T16:58:51Z

The coordination complexes of Yb(III) exhibit some of the longest spin coherence times among 4f compounds, making them a promising platform for molecular quantum technologies. While spin-phonon relaxation remains a limiting factor for coherence times even at low temperature, its control through chemical design has the potential to push these spin qubits prototypes beyond current limits. With the aim of providing insights on how to chemically control spin-phonon relaxation, we here present a full ab initio study of spin-phonon dynamics for three Yb(III) molecules exhibiting minimal chemical differences, yet quantitatively different spin relaxation times. Results show that low-temperature relaxation is governed by Raman processes triggered by a small group of largely delocalized low-energy phonons. The analysis of these contributions highlights that the modulation of spin-phonon coupling by molecular structure modifications beyond the first coordination shell are highly non-trivial in nature and hard to rationalize in simple chemical terms. These findings call for a conceptual step change from the attempt to use simple magneto-structural correlations to interpret the effect of molecular structural modifications on spin-phonon relaxation, and present predictive first-principles frameworks as a potential driving force of future chemical design strategies

2026-03-12T16:58:51Z Giacomo Sansone Lorenzo A. Mariano Stefano Carretta Paolo Santini Alessandro Lunghi http://arxiv.org/abs/2507.18930v2 2026-03-12T16:22:25Z

We investigate an approximate sampling scheme that can significantly reduce the cost scaling of variational Monte Carlo when it is employed to predict the energy differences associated with local chemical changes. Inspired by side-chaining and embedding methods, this spotlight sampling approach adopts an approximate fragmented Hamiltonian and correlated sampling to reduce cost scaling to the point that it is essentially linear with system size, with the potential to go sub-linear if certain conditions are met. In tests on bond stretching energies in alcohols, hydrogen dimer chains, and molecules with various degrees of $π$-system delocalization, we observe the anticipated linear scaling as well as an explicit cost crossover with standard variational Monte Carlo.

2025-07-25T03:56:06Z Sonja Bumann Eric Neuscamman 10.1021/acs.jctc.5c01232 http://arxiv.org/abs/2602.09986v2 2026-03-12T15:53:19Z

Thermodynamics is commonly presented as a theory of macroscopic systems in stable equilibrium, built upon assumptions of extensivity and scaling with system size. In this paper, we present a universal formulation of the elementary foundations of thermodynamics, in which entropy and energy are defined and employed beyond equilibrium and without assuming extensivity. The formulation applies to all systems -- large and small, with many or few particles -- and to all states, whether equilibrium or nonequilibrium, by relying on carefully stated operational definitions and existence principles rather than macroscopic idealizations. Key thermodynamic concepts, including adiabatic availability and available energy, are developed and illustrated using the energy-entropy diagram representation of nonequilibrium states, which provides geometric insight into irreversibility and the limits of work extraction for systems of any size. A substantial part of the paper is devoted to the analysis of entropy transfer in non-work interactions, leading to precise definitions of heat interactions and heat-and-diffusion interactions of central importance in mesoscopic continuum theories of nonequilibrium behavior in simple and complex solids and fluids. As a direct consequence of this analysis, Clausius inequalities and the Clausius statement of the second law are derived in forms explicitly extended to nonequilibrium processes. The resulting framework presents thermodynamics as a universal theory whose concepts apply uniformly to all systems, large and small, and provides a coherent foundation for both teaching and modern applications.

2026-02-10T17:12:20Z 84 pages, 41 figures, 68 footnotes, 80 references, contents closely related to the first part of 2.43 Advanced Thermodynamics, MIT OpenCourseWare, Spring 2024, https://ocw.mit.edu/courses/2-43-advanced-thermodynamics-spring-2024/ and https://www.youtube.com/playlist?list=PLUl4u3cNGP6309d0oJDiVo1CvxUQXJ2il Gian Paolo Beretta http://arxiv.org/abs/2603.12045v1 2026-03-12T15:20:06Z

We present a measurement-driven quantum framework for evaluating overlap and Hamiltonian matrix elements in spin-coupled generalized valence bond (SCGVB) wavefunctions. The approach targets a central difficulty of nonorthogonal valence-bond methods: estimating matrix elements between distinct, generally nonorthogonal configuration state functions. Rather than preparing the full wavefunction on quantum hardware, we reformulate the required quantities as vacuum expectation values of Pauli-string operators that can be accessed using shallow, ancilla-free circuits composed of local Clifford rotations and computational-basis measurements. In contrast to Hadamard-test-based matrix-element estimation, this construction avoids ancilla qubits and controlled operations by reducing the problem to local Pauli measurements. This separates the algebraic construction of the SCGVB problem from the measurement task executed on the quantum register and yields a low-depth strategy compatible with near-term architectures. We demonstrate the framework on square and rectangular H4 using quantum-circuit emulation, where the resulting overlap and Hamiltonian matrices reproduce classical Lowdin-based references with good accuracy across the geometries considered, and where derived Coulson-Chirgwin weights remain chemically consistent. These results support the feasibility of measurement-based quantum assistance for nonorthogonal SCGVB expansions and provide a practical route for incorporating quantum measurements into valence-bond electronic-structure workflows.

2026-03-12T15:20:06Z Bruna Gabrielly http://arxiv.org/abs/2603.07977v2 2026-03-12T13:59:35Z

Machine Learning Interatomic Potentials (MLIPs) enable accurate large-scale atomistic simulations, yet improving their expressive capacity efficiently remains challenging. Here we systematically develop Mixture-of-Experts (MoE) and Mixture-of-Linear-Experts (MoLE) architectures for MLIPs and analyze the effects of routing strategies and expert designs. We show that sparse activation combined with shared experts yields substantial performance gains, and that nonlinear MoE formulations outperform MoLE when shared experts are present, underscoring the importance of nonlinear expert specialization. Furthermore, element-wise routing consistently surpasses configuration-level routing, while global MoE routing often leads to numerical instability. The resulting element-wise MoE model achieves state-of-the-art accuracy across the OMol25, OMat24, and OC20M benchmarks. Analysis of routing patterns reveals chemically interpretable expert specialization aligned with periodic-table trends, indicating that the model effectively captures element-specific chemical characteristics for precise interatomic modeling.

2026-03-09T05:27:38Z Yuzhi Liu Duo Zhang Anyang Peng Weinan E Linfeng Zhang Han Wang http://arxiv.org/abs/2603.11893v1 2026-03-12T13:06:00Z

We present the application of the recently developed one-body Møller--Plesset perturbation theory (OBMP2) to the prediction of K-edge excited states. OBMP2 is a self-consistent perturbation theory in which a canonical transformation followed by a cumulant approximation yields an effective one-body Hamiltonian. This resulting operator augments the standard Fock operator with a one-body correlation potential containing double-excitation MP2 amplitudes, allowing molecular orbitals and orbital energies to be optimized in the presence of correlation. This self-consistent framework mitigates convergence and accuracy issues often encountered in standard non-iterative MP2 for open-shell systems and bond-stretching regimes. In this work, we evaluate the performance of an OBMP2-based approach for the calculation of K-edge excitations. Utilizing benchmark test sets of both closed-shell and open-shell molecules, we demonstrate that our method outperforms established standard techniques, including $Δ$DFT, EOM-CCSD, and USTEOM-CCSD. Our findings establish the OBMP2-based $Δ$SCF protocol as a robust and accurate new computational method for the treatment of K-edge excited states.

2026-03-12T13:06:00Z 7 pages, 3 figures, 2 tables Lan Nguyen Tran http://arxiv.org/abs/2603.11891v1 2026-03-12T13:03:04Z

The violation of Hund's rule, resulting in an inverted singlet-triplet (INVEST) gap, represents a paradigm shift in photophysics with major implications for OLED technology. INVEST molecules facilitate barrierless reverse intersystem crossing, theoretically permitting 100\% internal quantum efficiency without thermal activation. However, accurately predicting negative singlet-triplet energy gaps typically demands prohibitive computational costs. In this study, we evaluate the efficacy of our recently developed one-body Møller-Plesset perturbation theory (OBMP2) and its spin-opposite variant (O2BMP2) as efficient alternatives. Benchmarking against 30 INVEST molecules reveals that O2BMP2, with appropriate spin-opposite scaling, achieves the accuracy of ADC(3) and EOM-CCSD. Furthermore, with the possibility of reducing computational complexity to $N^4$, O2BMP2 provides a robust balance of accuracy and efficiency, making it suitable for the high-throughput screening of next-generation INVEST materials.

2026-03-12T13:03:04Z 25 pages, 6 figures, 5 tables Nhan Tri Tran Hoang Thanh Nguyen Lan Nguyen Tran http://arxiv.org/abs/2603.10931v2 2026-03-12T12:26:49Z

Time-slicing has emerged as a strategy for incorporating semiclassical propagation into real-time path integral formulation and recovering full quantum dynamics. A central step is the decomposition of a time-evolved wave function into a superposition of Gaussian wave packets (GWPs). Here we introduce a quadrature-free variational framework for GWP decomposition, reformulating it as an optimization problem in which the GWP parameters are chosen to maximize the overlap with the time-evolving wave function. An autoencoder-decoder neural network is used for this optimization, with the representation being adaptively reoptimized during propagation. Each wave packet in this decomposition represents a localized patch of the underlying semiclassical manifold, while retaining full correlations between all degrees of freedom. This variational adaptive Gaussian decomposition (VAGD) approach yields a compact Gaussian expansion, providing a scalable route to time-sliced semiclassical quantum dynamics. While general, applying VAGD to facilitate time-slicing of thawed Gaussian approximation (TGA) allows a route to improving the semiclassical treatment to the full quantum mechanical result in a systematic manner.

2026-03-11T16:15:56Z 10 pages, 11 figures Rahul Sharma Amartya Bose http://arxiv.org/abs/2511.05897v2 2026-03-12T10:56:01Z

We have developed a reduced-cost non-Dyson third-order algebraic diagrammatic construction theory for the electron-attachment problem based on state-specific frozen natural orbitals. Density fitting and truncated natural auxiliary functions were employed to enhance computational efficiency. The use of state-specific frozen natural orbitals significantly decreases the virtual space and provides a notable speedup over the conventional EA-ADC(3) method with a systematically controllable accuracy. A perturbative correction for the truncated natural orbitals significantly reduces the error in the calculated electron affinity values. The method also shows sufficient accuracy in the case of non-valence correlation-bound anions, where the local approximation-based methods fail. The efficiency of the method is demonstrated by performing an EA-ADC(3) calculation with more than 1300 basis functions.

2025-11-08T07:44:49Z 13 pages, 10 figures, 4 tables Tamoghna Mukhopadhyay Kamal Majee Achintya Kumar Dutta