https://arxiv.org/api/Gz7v9E+SRqGE2u1n/5kTMR+1yxM 2026-06-13T15:34:24Z 27442 105 15 http://arxiv.org/abs/2606.05127v1 2026-06-03T17:31:32Z

Foundation models in atomistic machine learning encode interaction physics across diverse atomic environments, but whether that structure can be transferred when building specialist potentials at quantum-chemical accuracy remains open. Here we show that knowledge distillation from a pretrained universal machine-learning interatomic potential (MLIP), followed by coupled-cluster fine-tuning with single and double excitations and perturbative triples [CCSD(T)], transfers not only low-cost labels but a physically meaningful prior on interaction length scales, anisotropy, and the repulsive-dispersive balance, which CCSD(T) data then sharpens to quantum-chemical accuracy. For He--benzene, fine-tuning with 30% of the CCSD(T) data outperforms direct training using the full 80%; a 60% reduction in the high-fidelity compute budget. A symmetry-adapted perturbation theory (SAPT)-informed adaptive short-range/long-range architecture further lowers the validation MAE from 0.75 1/cm to 0.49 1/cm. Across a circumarene series of polycyclic aromatic hydrocarbons (PAHs), swapping the MLIP teacher under an otherwise identical pipeline changes the coronene error by an order of magnitude while leaving the larger PAHs stable, direct evidence that distillation transfers physical structure, not labels alone. Together, these results identify the choice of pretrained teacher as a primary design axis for data-efficient quantum-chemical-accuracy potentials, alongside architecture and training protocol.

2026-06-03T17:31:32Z 10 pages, 5 figures plus supplemental material. For associated data and code repository see: https://github.com/DelMaestroGroup/papers-code-mlip-distillation-sapt Yulin Shen Shahzad Akram Louis Primeau Gen Zu Konstantinos D. Vogiatzis Yang Zhang Adrian Del Maestro http://arxiv.org/abs/2604.02121v2 2026-06-03T16:19:13Z

Stochastic kinetic models are ubiquitous in physics, yet inferring their parameters from experimental data remains challenging. For deterministic models, parameter inference often relies on gradients, which can be obtained efficiently through automatic differentiation (AD). However, AD cannot be applied directly to the Gillespie stochastic simulation algorithm (SSA), since sampling from a discrete set of reactions introduces non-differentiable operations. In this work, we adopt three gradient estimators from machine learning for the Gillespie SSA: the Gumbel-Softmax Straight-Through (GS-ST) estimator, the Score Function estimator, and the Alternative Path estimator. We use the estimators to evaluate gradients of steady-state and time-dependent observables, and compare their performance in representative biophysical systems with relaxation dynamics (bimolecular association) and oscillatory dynamics (repressilator). We find that the GS-ST estimator generally yields well-behaved gradient estimates, but exhibits diverging variance in challenging parameter regimes, which can cause parameter inference to fail. In these cases, other estimators provide more robust, lower variance gradients. Our results demonstrate that gradient-based parameter inference can be effectively combined with the Gillespie SSA, with different estimators offering complementary advantages.

2026-04-02T14:56:38Z 19 pages, 9 figures Ludwig Burger Annalena Kofler Lukas Heinrich Ulrich Gerland http://arxiv.org/abs/2606.05050v1 2026-06-03T16:10:58Z

Theoretical heterogeneous catalysis promises rapid catalyst discovery, yet computational and machine-learning predictions often deviate from experiment and stay confined to narrow material families, for want of a faithful, condition-aware catalytic simulator. We present CatDT (Catalysis Digital Twin), a self-evolving multi-agent system that builds an autonomous digital twin of a working catalyst, unifying gas-solid and liquid-solid modeling. From only a bulk crystal and a natural-language reaction description, eight specialized agents and 27 scientific tools predict stable facets, reconstruct working surfaces, enumerate and rank reaction pathways, locate transition states, and compute kinetics in 5-30 min on a single GPU. Two innovations address the hardest steps: UniMech finds dominant pathways for novel materials at over $10^3\times$ lower cost than exhaustive enumeration by fusing agent-guided proposals with energy-cached graph search, and a memory-augmented reinforcement loop raises barrier-calculation success from 41\% to 84\% across 600 catalytic surfaces. Across seven gas-solid benchmarks -- stepped metals, single-atom catalysts, ordered intermetallics, vacancy-rich 2D sulfides and carbides, and a strong-metal--support-interaction (SMSI) interface -- every CatDT prediction lies within 0.5-2 times experiment over four orders of magnitude. For propane dehydrogenation, CatDT independently discovers non-precious candidates rivaling the Pt-based industrial benchmark, with a proposed Ni@ZrO$_2$ SMSI overlayer reaching a simulated TOF of $1.63~\text{s}^{-1}$ at $\sim$100\% selectivity. More broadly, the decisive factor for a faithful catalyst digital twin -- or any multi-stage scientific simulator -- is not raw LLM capability but the engineered harness around it: deterministic tools, persistent memory, and verified self-improvement that compound across models, tools, and runs.

2026-06-03T16:10:58Z Zhilong Song Zongmin Zhang Lixue Cheng http://arxiv.org/abs/2606.04786v1 2026-06-03T12:08:45Z

The energy-based approach to operator selection in ADAPT-VQE relies on reconstructing the one-parameter energy landscape for each operator in the pool. In fermionic implementations, the cost of reconstructing this energy landscape often becomes a bottleneck. We address this issue through an exact Hamiltonian transformation that reformulates the one-parameter energy landscape according to a generator-dependent fragmentation of the transformed Hamiltonian. While our method is mathematically identical to standard fermionic Rotoselect, it effectively reduces its cost by about a factor of two, bringing it close to that of gradient-based ADAPT-VQE. We use this formulation to benchmark the gradient-based and energy-based selection approaches in combination with two ansatz-optimization strategies -- `last', where only the appended operator is optimized, or `full', where the full ansatz is re-optimized -- and with both fixed-orbital and orbital-optimized formulations. The benchmark comprises $\text{LiH}$, $\text{BeH}_2$, and $\text{H}_2\text{O}$ at both equilibrium and stretched geometries. In the weakly correlated regime, the `last' optimization strategy combined with energy-based selection enables the efficient construction of an accurate ansatz, while avoiding any VQE optimization. As correlation increases, full ansatz re-optimization and orbital optimization become the main factors governing convergence and overall resource cost. These results show that exact Hamiltonian transformations provide an effective route to reducing the measurement overhead of fermionic energy-based ADAPT-VQE. Moreover, the benchmark clarifies the relative role of operator scoring approach, re-optimization strategy, and orbital treatment in the performance of ADAPT-VQE.

2026-06-03T12:08:45Z Emanuele Rossi Erik Rosendahl Kjellgren Artur F. Izmaylov Stephan P. A. Sauer Karl Michael Ziems Sonia Coriani http://arxiv.org/abs/2510.25508v2 2026-06-03T12:02:34Z

Generating tunable, high-intensity mid-infrared (MIR) to terahertz (THz) radiation on-chip remains a formidable challenge due to the rigid spectral limits of conventional thermal emitters. While graphene has emerged as a promising platform for light-matter interaction, active control of its radiative properties has been largely confined to surface-limited phenomena mostly associated with plasmons. Here, we introduce a new MIR radiation platform where multi-layer chemical vapor deposition (CVD) graphene is integrated with modular, vibrationally active dielectric substrates, ranging from organic thin films and inorganic matrices. A pivotal discovery is that the long-range de Broglie wavelength of drift carriers enables coherent coupling with vibrational transition dipoles deep within the substrate bulk. This transforms the substrate into a three-dimensional volume emission source, where complex spectra of characteristic molecular and lattice vibration energies are additively combined on demand. The exponential scaling of radiation intensity appears when the electrons' drift velocity in graphene exceeds the sound velocity of the substrates, consistent with quantum stimulated amplification associated with Cerenkov electron-phonon instability. Our work redefines the passive dielectric substrate as an active, programmable component driven by electron waves, paving the way for next-generation system-on-a-chip MIR-THz photonics, environmental and biomedical sensing, and highly efficient mode-specific electrothermal applications.

2025-10-29T13:32:01Z 12 pages,15 figures Sunhwa Hong Moo Jin Kwak Yunseok Lee Chan-Jin Kim Sung Jin Hong Ha Eun Lee Yejun Lee Koeun Kim Juhyen Lee Minkyung Lee Youngdeog Koh Joonhyun Lee Miyoung Kim Zee Hwan Kim Myung Jin Park Hoon Wee Byung Hee Hong Konstantin S. Novoselov http://arxiv.org/abs/2606.04667v1 2026-06-03T09:45:40Z

We performed numerical (restricted) path integral Monte Carlo experiments on metallic Hydrogen from first principles. We study a quantum two component plasma where one component is made of pointwise particles of negative unitary charge and the other is made of charged hard spheres of positive unitary charge. We study both the additive mixture and a nonadditive mixture where we only keep a hard core between unlike species. We specialize to the case of the electron-proton plasma with a 1:1 ratios between the molar fraction of the two species. We measured thermodynamic and structural properties of the plasma. From an analysis of the structure we see a transition from a metallic Hydrogen phase, to a molecular Hydrogen phase as the temperature is lowered. As expected at high density the correlations are diminished.

2026-06-03T09:45:40Z 12 pages, 1 table, 7 figures Riccardo Fantoni http://arxiv.org/abs/2606.07656v1 2026-06-03T08:57:00Z

Solubility prediction is a standard benchmark in computational chemistry, yet multi-solvent models which reportedly approach the experimental-noise ceiling (i.e. the aleatoric limit) are not yet reliable enough to be deployed. We argue that this gap is partly artefactual: published benchmarks differ in curation policies, evaluate on count-weighted RMSE that hides failure on tail-heavy solvent distributions, and treat the widely cited 0.6-0.8 log S inter-laboratory figure as the aleatoric ceiling even though it reflects worst-case, not expected, disagreement. We introduce SC3, a multi-solvent solubility benchmark built on BigSolDB v2.1 with three contributions: (i) a reproducible curation pipeline yielding 101,535 measurements over 1,327 solutes and 206 solvents, with a recalibrated aleatoric floor of 0.106 log S-roughly 6 times tighter than the conventional figure; (ii) nested Gold/Silver/Bronze consensus tiers with per-point standard deviation, three leakage-checked splits, and a multi-solvent metric suite (PS-RMSE, Z-RMSE); and (iii) a 31-model benchmark across six families, whose best Bronze PS-RMSE sits at 5 times the aleatoric limit, and we observe this is a gap unclosed by any deep alternative tested. We perform three follow-on analyses: data scaling, transfer from quantum-chemistry solvation energies, and feature-level attribution, which demonstrates that calibrated per-point uncertainty is a reusable infrastructure for diagnosis beyond point prediction.

2026-06-03T08:57:00Z 34 pages, 16 tables, 22 figures Vansh Ramani Har Ashish Arora Dhairya Kuchhal Sergei Tatarin Lev Krasnov Sayan Ranu Tarak Karmakar http://arxiv.org/abs/2601.20259v2 2026-06-03T06:59:49Z

Phycobiliproteins serve as highly efficient light-harvesting antennae in cyanobacteria, yet the molecular factors governing their ultrafast energy relaxation and coherence dynamics remain incompletely understood. In this study, we investigate the role of pigment arrangement and pigment-protein interactions by combining recombinant protein engineering with two-dimensional electronic spectroscopy (2D-ES). In addition to wild-type allophycocyanin (APC) and C-phycocyanin (CPC), we artificially synthesized a β153 phycocyanobilin (PCB)-deficient CPC mutant, enabling direct experimental isolation of the contribution of this peripheral pigment. The absorption and fluorescence spectra show that removal of the β153 pigment primarily eliminates its spectral contribution without significantly altering the excitonic coupling between the α84 and β84 pigments. Time-resolved 2D-ES reveals the close similarity between the dynamics of wild-type and β153-deficient CPCs, which demonstrate that the β153 pigment plays a minor role in ultrafast relaxation dynamics. Instead, the difference between APC and CPC arise primarily from pigment-protein interactions that modulate pigment geometry and vibronic structure. These results highlight the critical importance of local protein environments in controlling energy relaxation and coherence in photosynthetic light-harvesting proteins.

2026-01-28T05:13:54Z Masaaki Tsubouchi Takatoshi Fujita Motoyasu Adachi Ryuji Itakura http://arxiv.org/abs/2511.18837v2 2026-06-03T05:33:51Z

The crystal field theory as explained by Abragam and Bleaney in their landmark 1970 book on transition-ion electron paramagnetic resonance remains a cornerstone in the development of luminescence applications and molecular magnets based on the $f$-elements. The modern numerical derivation of the 27 $B_k^q$ Stevens crystal field parameters (CFPs), which describe the splitting of the energy levels of a central ion, is traditionally achieved through the effective Hamiltonian theory and multiconfiguration wavefunction theory calculations, insofar as the lowest $J$ level fully captures the targeted low-energy physics. In this work, we present a novel theoretical approach for determining the CFPs. The procedure resembles the traditional extraction path but crucially accounts for the full $\ket{J,M_J}$ space of an ion configuration with $L=3$ and $S=\nicefrac{1}{2}$. By demonstrating the extraction procedure using the simplest case of a Ce$^\text{III}$ 4f$^1$ ion with a crystal-field split $J \in \{\nicefrac{5}{2}, \nicefrac{7}{2}\}$ manifold, it is shown for the first time that a unique set of CFPs describes the splitting and mixing both the $J$ manifolds. In fact, this $J/J^\prime$ mixing is analogous to the ``spin mixing'' in binuclear transition metal complexes. At the employed level of calculation, we demonstrate that there is no spin-orbit coupling influence on the CFP values, contrary to previous beliefs. This study represents the first step of a larger effort in reviewing the theory and extraction procedures of CFPs in f-element complexes.

2025-11-24T07:16:17Z 17 pages with SI included Journal of Chemical Physics, 164, 144304 (2026) Dumitru-Claudiu Sergentu Gwenhaël Duplaix-Rata Ionel Humelnicu Boris Le Guennic Rémi Maurice 10.1063/5.0313869 http://arxiv.org/abs/2606.04452v1 2026-06-03T04:55:53Z

Determining mutant protein structures is critical for understanding the mechanistic roles of mutations in biochemical processes. However, experimental characterization and conventional theoretical modeling are often expensive and time-consuming. Recent advances in machine learning provide new opportunities to efficiently predict protein structures from primary sequences. Nevertheless, applying these models to proteins with single-site or few-site mutations remains challenging because mutant sequences are often highly similar to their wild-type counterparts. Here, we introduce DeltaDiff, a physics-guided inference framework for mutant-structure generation that incorporates mutation-aware physical guidance into a baseline diffusion model. We evaluate DeltaDiff on three representative systems: Chignolin T8P, Novispirin G-10, and BBL D162N. All three examples involve nonlocal structural changes, making accurate mutant-structure prediction challenging. DeltaDiff captures key mutation-induced conformational changes without requiring retraining or fine-tuning of the baseline model. These results establish a foundation for efficient mutant-structure prediction at a fraction of the cost of conventional methods, facilitating rational mutant design.

2026-06-03T04:55:53Z Yajie Cai Yanbin Wang Ming Chen http://arxiv.org/abs/2605.26594v2 2026-06-03T04:29:34Z

While spin-adapted time-dependent density functional theory (TDDFT) approaches significantly improve the excitation energies and gradients of open-shell molecules, the effect of spin-adaptation on non-adiabatic coupling matrix elements (NACMEs) remains unknown for spin-conserving excitations. In this article, we report the derivation, implementation and benchmark studies of the ground state-excited state and excited state-excited state NACMEs of our spin-adapted TDDFT method, X-TDDFT; to our best knowledge, this represents the first implementation of the analytic NACMEs of a spin-adapted TDDFT method. Similar to the X-TDDFT analytic gradients, X-TDDFT NACMEs can be easily implemented on top of an existing U-TDDFT NACME implementation taking into account the restricted open-shell Kohn-Sham (ROKS) reference and the implicit involvement of doubly excited determinants, with acceptable computational overhead. Benchmark calculations reveal that X-TDDFT reduces the error of U-TDDFT NACMEs by 1/3-2/3 (referenced against high-level multireference NACMEs), which leads to large corrections of internal conversion rates (up to two orders of magnitude). In particular, for copper(II) porphyrin, X-TDDFT leads to qualitative revisions of the relative importance of the excited state relaxation pathways, as well as the substituent effects of the internal conversion (IC) rates, suggesting that the error of U-TDDFT NACMEs is not only large but also unsystematic. It is therefore expected that X-TDDFT NACMEs will prove useful in the photophysics/photochemistry studies of open-shell systems such as radicals and transition metal complexes.

2026-05-26T06:28:04Z 61 pages, 7 figures, 5 tables Xiaoli Wang Xingwen Wang Zikuan Wang Wenjian Liu http://arxiv.org/abs/2606.04337v1 2026-06-03T01:29:24Z

Light-matter interactions provide versatile routes for probing and controlling chemical reactivity, charge transport, and material properties. Time-periodic external fields can reshape electronic states and open new dynamical pathways beyond the field-free Born-Oppenheimer (BO) picture. Floquet nonadiabatic dynamics has consequently emerged as an important framework for describing coupled electron-nuclear dynamics under periodic driving. In this Perspective, we first discuss recent developments in Floquet nonadiabatic dynamics methods for closed and open quantum systems. We then highlight how this framework provides mechanistic insights into electron transfer at molecule-metal interfaces, quantum transport in molecular junctions, carrier dynamics in crystalline solids, and multicolor Floquet engineering. Finally, we outline key conceptual and computational challenges that must be addressed to transform Floquet nonadiabatic dynamics from model-based demonstrations into predictive, first-principles simulations of realistic light-driven processes.

2026-06-03T01:29:24Z Jiayue Han Yu Wang Vahid Mosallanejad Wei Liu Wenjie Dou http://arxiv.org/abs/2606.04316v1 2026-06-03T00:43:36Z

High-accuracy correlated four-component relativistic electronic structure methods are typically formulated in terms of integrals over molecular orbital (MO). Consequently, an efficient and scalable strategy is required to deal with the complexity of transforming relativistic two-electron integrals from the atomic orbital (AO) to the MO basis. The transformation bottleneck is particularly acute for approaches that include Breit interaction integrals, whose computational and memory demands further exacerbate the transformation cost. To overcome this challenge, we develop a quaternion-based, AO-driven direct integral transformation scheme. The method operates on scalar AO integrals and combines quaternion density-based contractions with direct Cauchy-Schwarz screening to systematically exploit integral locality. As a result, the proposed framework substantially lowers the practical computational scaling and provides an efficient, memory-conscious, and highly parallelizable pathway for the routine inclusion of relativistic Dirac-Coulomb-Breit integrals in large-scale four-component correlated calculations.

2026-06-03T00:43:36Z 18 pages, 4 figures Martijn Oele Rajat Majumder Shiv Upadhyay Tianyuan Zhang Ryan A Beck Agam Shayit Lucas Visscher Xiaosong Li http://arxiv.org/abs/2606.04285v1 2026-06-02T23:25:33Z

The $G3W2$ approximation -- the second-order self-energy beyond $GW$ -- is known to violate some fundamental analytic properties of the self-energy. In particular, its lack of positive semi-definiteness leads to unphysical features such as negative spectral functions. In this work, we reformulate the $G3W2$ approximation within the algebraic-diagrammatic construction (ADC) framework. The resulting ADC-$G3W2$ formalism enforces the same analytic form as the exact self-energy, namely a sum-over-state representation, and, consequently, guarantees positive semi-definiteness. Starting from the $GW$ self-energy, we construct a hierarchy of ADC-based approximations of increasing sophistication, including ADC-2SOSEX, ADC(3)-$G3W2$, and a full ADC-$G3W2$ scheme. These methods can be interpreted as nonperturbative resummations of vertex corrections to the self-energy, yielding Hermitian effective Hamiltonians whose diagonalization provides quasiparticle and satellite energies. This establishes a formal bridge between many-body perturbation theory formulated in terms of the screened interaction $W$ and conventional ADC schemes based on the bare Coulomb interaction. The performance of these ADC-based approximations is gauged for valence ionization potentials and benchmarked against their parent method.

2026-06-02T23:25:33Z 12 pages, 2 figures (Supporting Information available) Antoine Marie Johannes Tölle Pierre-François Loos http://arxiv.org/abs/2606.04242v1 2026-06-02T21:42:16Z

The spin dynamics of water ice in the presence of external magnetic fields are investigated. The employed model builds upon the approach introduced by Buntkowsky et al. [Z. Phys. Chem. 222, 1049 (2008)], which considers two nearest-neighbor water molecules and yields a four-spin system, as the abundant oxygen isotope has zero nuclear spin. The model is extended to include coupling to external magnetic fields, allowing us to analyze the interplay between magnetic dipole-dipole interactions and magnetic field coupling. Two types of configurations are examined: (i) static, homogeneous fields, corresponding to a time-independent interaction, and (ii) spatially varying sinusoidal fields in relative motion with the molecules, leading to a time-dependent interaction. All computations are performed within the density operator formalism. The ortho/para populations and the total spin projections are evaluated during the first tens of milliseconds following the gas-to-solid phase transition. For static homogeneous fields, we show that increasing field strength suppresses dipolar-induced depolarization. Assuming that all molecules are initially in the para state, we show that static homogeneous fields can drive the ortho population up to approximately $50\%$, whereas suitably chosen sinusoidal-field configurations can increase it beyond $90\%$. These results are relevant for schemes aiming to preserve or manipulate nuclear-spin polarization during deposition.

2026-06-02T21:42:16Z Chrysovalantis S. Kannis Ralf Engels Nicolas Faatz Simon J. Pütz Markus Büscher