Journal of Chemical Theory and Computation最新文献

{"title":"Incorporating Anharmonicity of Selected Modes into Fermi’s Golden Rule Rate Calculations for Large Molecular Systems","authors":"Takuma Ito,&nbsp;Yuki Kurashige","doi":"10.1021/acs.jctc.6c00696","DOIUrl":"10.1021/acs.jctc.6c00696","url":null,"abstract":"<p>Properties of photofunctional materials are governedby nonadiabatictransitions, and theoretical methods to predict reaction rate constantsfor nonadiabatic transitions are desired. Fermi’s golden rule(FGR) is one of the most useful approaches for computing rate constants.In particular, FGR combined with the harmonic approximation to thepotential energy surface (PES) is widely used, since it requires onlygeometry optimizations and Hessian calculations for the initial andfinal states. However, real PESs are inherently anharmonic, and suchanharmonicity can affect the rate constant by altering vibrationalenergy levels and wave functions. In this study, we propose a methodfor computing rate constants that incorporates anharmonicity intothe conventional harmonic-approximation framework. In this method,only a subset of important modes is treated anharmonically, whereasthe remaining modes are treated harmonically. We applied the proposedmethod to two systems: the T₁ → S₀ nonradiativedecay process of 4-pyrone molecule, and the S₁ →T₁ intersystem crossing process of 3-(9,9-dimethylacridin-10(9<em>H</em>)-yl)-9<em>H</em>-xanthen-9-one (ACRXTN) molecule.The calculation for 4-pyrone revealed that the rate constant decreasesby 3 orders of magnitude compared with the harmonic rate when anharmonicityof all vibrational modes is taken into account, indicating the importanceof considering the anharmonicity when calculating the rate constant.In addition, it was found that the fully anharmonic rate constantcan be reproduced by considering the anharmonicity of only two vibrationalmodes. Further analysis based on the sum-over-states formalism revealedthat this decrease in the rate constant originates from a reductionin the overlaps between the initial- and final-state vibrational eigenfunctions.It was also found that this decrease in the rate constant does notappear unless the anharmonicity of multiple vibrational modes is takeninto account simultaneously, even though coupling between vibrationalmodes is not considered. The application to ACRXTN showed that theanharmonicity associated with dihedral torsional modes has littleinfluence on the rate constant. Comparison of these results revealedthat the anharmonicity of vibrational modes with high Huang–Rhysfactors and high frequencies, or those whose harmonic frequenciesare different between initial and final states, has an impact on therate constant. The proposed method is expected to be useful for thetheoretical design of new photofunctional materials.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11456–11470"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149221626","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Force Fields and Simulation Engines: Lessons Learned from Porting Amber ff14SB and ff19SB to GROMACS","authors":"Vedran Miletić,&nbsp;Miłosz Wieczór,&nbsp;Carlos Simmerling,&nbsp;Markus Rampp,&nbsp;Bert L. de Groot,&nbsp;Carsten Kutzner,&nbsp;Vytautas Gapsys","doi":"10.1021/acs.jctc.6c01592","DOIUrl":"10.1021/acs.jctc.6c01592","url":null,"abstract":"<p>The GROMACS molecular dynamics simulation package hashistoricallysupported a number of molecular mechanics force fields, includingAmber. Force fields have evolved over time and continue to do so throughrefinement of existing interactions and the definition of additionalones to more faithfully match real-world experiments and higher-levelquantum chemistry calculations. Although users have provided a numberof conversions of newer Amber force fields over the years, the twomost recent Amber protein force fields, ff14SB and ff19SB, had notpreviously been officially ported to GROMACS. A technical challengeis that ff19SB introduces residue-specific backbone dihedral energycorrection maps (CMAPs), which required new support in GROMACS forresidue-aware force-field preprocessing and topology generation. Unlikeolder force fields such as CHARMM27, in which CMAPs are parameterizedusing only atom types, ff19SB extends this scheme with additionalamino-acid typing. Here, we describe how we ported and validated theAmber ff14SB and ff19SB force fields for use with GROMACS as AMBER14SBand AMBER19SB, respectively. The ported force fields reproduce theAmber reference topologies exactly, with per-interaction energiesmatching to within the single-precision floating-point limit (&lt;10^–7 kJ/mol). To achieve this level of consistency betweenAmber and GROMACS, we resolved discrepancies in their handling offundamental constants, parameter precision, and improper-dihedralatom ordering logic. In long unrestrained ubiquitin simulations, theconverted ff19SB reproduces the conformational free-energy profileobtained with the native Amber implementation. This work providesthe first official port of an Amber force field to GROMACS in morethan a decade, improving feature parity between the two simulationpackages. We provide a reproducible porting and validation procedurethat can be used for modified variants and future versions of Amberforce fields.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11142–11152"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149221787","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Augmented Roothaan-Hall Hessian Applied to Spin-Restricted Open-Shell Density Functional Theory","authors":"Yichi Zhang,&nbsp;Jun Yang","doi":"10.1021/acs.jctc.6c01383","DOIUrl":"10.1021/acs.jctc.6c01383","url":null,"abstract":"<p>We generalize the augmented Roothaan–Hall (ARH)Hessianformalism to the self-consistent field (SCF) optimization of spin-restrictedopen-shell (RO) wave functions, encompassing high-spin, low-spin,and two-determinant electronic states. A detailed ARH formulationis presented. We demonstrate that ARH is a highly efficient optimizationalgorithm for rapidly identifying accurate SCF solutions, primarilyowing to its systematic construction of an effective Hessian, particularlyin the case of Euclidean quadratic energy functions. The ARH is builtupon a universal energy formulation, including grid-based integration,for spin-restricted closed-shell, spin-unrestricted, and RO densityfunctional theory (DFT), thereby unifying and simplifying their numericalimplementation. The performance of the present method is evaluatedusing two benchmarking studies. First, for a series of iron–sulfurclusters exhibiting different spin states, which represent notoriouslychallenging SCF problems, the ARH algorithm demonstrates superiorconvergence efficiency relative to L-BFGS and truncated Newton methods,requiring much fewer RO-SCF iterations to achieve convergence. Second,the ARH approach tends to avoid convergence to higher-energy stationarypoints in two-determinant RO-SCF calculations for singlet excitedstates of selected photoactive compounds. Finally, an applicationof the ARH-based RO-SCF is illustrated by an investigation of themechanistic origin of the spin-crossover phenomenon in a Ni(II)-porphyrincomplex utilized as a contrast agent.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11319–11330"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149241649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Electron Spin Decoherence due to Hindered Rotation in Donor-Bridge-Acceptor Linked Molecules","authors":"Tomoaki Miura","doi":"10.1021/acs.jctc.6c01847","DOIUrl":"10.1021/acs.jctc.6c01847","url":null,"abstract":"<p>Extraordinarilybroad line widths for the magnetic field effects(MFEs) on radical pair recombination in rigid oligoaromatic donor-bridge-acceptormolecules have been successfully simulated by the full-hybrid MonteCarlo (MC) MD simulation, in which modulation of the exchange couplingdue to hindered rotation is directly calculated. It has been demonstratedthat the single-site Liouville equation gives nearly identical resultsat general diffusional rotation times (&lt; a few nanoseconds) ifthe effective exchange coupling and singlet–triplet dephasingrate constants are estimated from the electron-only MC simulationof spin coherence. Furthermore, the single-site model is applicableto multinuclear simulations of the so-called low-field effect, whichis due to the coherent superposition of the degenerate zero-fieldeigenstates. These findings ensure MD-assisted spin dynamics simulationswith high accuracy and fewer computational resources.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11193–11205"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149241652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Can DFT-Trained Neural Network Potentials Reproduce Structure, Solvation, and Water-Exchange Properties in Aqueous Magnesium Solutions?","authors":"Sebastian Falkner,&nbsp;Pablo Montero de Hijes,&nbsp;Christoph Dellago,&nbsp;Nadine Schwierz","doi":"10.1021/acs.jctc.6c01270","DOIUrl":"10.1021/acs.jctc.6c01270","url":null,"abstract":"<p>Magnesium ions play an essential role in many biologicalprocessesbut remain challenging to model in biomolecular simulations. Despiteconsiderable scientific effort, classical force fields fail to simultaneouslyreproduce key structural, thermodynamic, and kinetic solution properties,likely due to their inability to explicitly account for quantum many-bodyeffects. Here, we develop and systematically benchmark MACE neuralnetwork potentials (NNPs) for aqueous MgCl₂ solutions,trained on revPBE-D3/zd and revPBE0-D3/zd density functional theoryreference data, and assess their ability to reproduce a broad rangeof experimental solution properties including the structure of thefirst hydration shell, diffusion coefficient, activity derivative,water-exchange rate, and mechanism as well as solvation free energy.Both NNPs accurately reproduce the octahedral structure of the firsthydration shell, ion-pairing properties, and diffusion coefficients.Combining the NNPs with transition path sampling and other enhancedsampling techniques allows us to capture the rare event of water exchangein the first hydration shell of Mg²⁺, revealing a dissociativeexchange mechanism. Transition interface sampling yields exchangerates within 1 order of magnitude of experiment, representing a substantialimprovement over classical dissociative force fields. In contrast,the NNP-derived solvation free energy significantly underestimatesthe experimental value, revealing a limitation of the present localNNP architecture for describing ion solvation thermodynamics. Ourresults demonstrate that DFT-trained NNPs can accurately describeMg²⁺ hydration structure, diffusion, ion pairing, and exchangekinetics while highlighting limitations of local NNP architecturesand the importance of long-range electrostatic contributions to achievequantitative agreement with experimental ion solvation free energies.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11432–11441"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149241848","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enabling Structure-Only Initialization and Out-of-Distribution Generalization in GNN-Based Molecular Dynamics Simulators","authors":"Sergey A. Shteingolts,&nbsp;Salman N. Salman,&nbsp;Dan Mendels","doi":"10.1021/acs.jctc.6c01145","DOIUrl":"10.1021/acs.jctc.6c01145","url":null,"abstract":"<p>Machine-learning-based simulators offer the potentialto modelthe dynamics of complex systems more efficiently than classical approacheswhile retaining differentiability, a key property for materials design.Graph neural network (GNN)-based simulators have shown strong performanceacross a range of physical domains, including molecular dynamics.However, their reliance on temporal context for accurate predictionlimits their use in inverse design settings, where simulations mustbe initialized from a single static configuration. Moreover, inversedesign requires robust out-of-distribution (OOD) generalization, ascandidate structures typically lie outside the training domain. Here,we address both challenges by introducing two complementary strategiesthat enable the stable and accurate structure-only initializationof GNN-based simulations. To directly target OOD generalization, wepropose an inference-time physics-based optimization framework thatconstrains model predictions to remain physically consistent duringthe rollout. In addition, we introduce a differentiable, GNN-basedbarostat that enables accurate tracking of system dimensions and pressure,which is critical for capturing macroscopic responses and supportingOOD generalization. We evaluated these approaches in the context ofuniaxial compression of disordered elastic networks spanning a broadrange of geometries, Poisson’s ratios, and microscopic behaviors.We find that, together, these methods substantially improve rolloutstability and enable reliable OOD generalization, including regimeswith distinct, more complex dynamics than those the in training data.These results show that when properly initialized and constrained,GNN-based simulators can serve as efficient and generalizable toolsfor materials discovery and structural optimization, advancing theiruse in materials, molecular, and dynamical system design.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11101–11110"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149251035","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Approaching Coupled Cluster Accuracy with Positive Semidefinite Vertex Corrected Self-Energies","authors":"Yaroslav Pavlyukh,Fabien Bruneval,Arno Förster","doi":"10.1021/acs.jctc.6c01644","DOIUrl":"https://doi.org/10.1021/acs.jctc.6c01644","url":null,"abstract":"Hedin’s formalism of functional derivatives is a successful method for systematically constructing correlated electronic theories, largely due to the success of its lowest-order self-energy expansion, the GW approximation. Beyond GW, diagrammatic resummation schemes attempt to mix correlations simultaneously across all particle-particle and particle-hole channels. Because such a comprehensive treatment is computationally prohibitive for realistic molecular systems, a highly effective alternative is to fully account for electronic correlations in one specific channel, typically the particle-hole channel. This idea was recently implemented for molecular systems, yielding a self-energy expressed in terms of excited-state energies and transition amplitudes from the solution of the Bethe–Salpeter equation, rather than the random-phase approximation used in GW. While this approach is predictive and numerically efficient, it violates the fundamental positive-definiteness constraint on the electron spectral function in certain energy ranges. In this study, we resolve this physical flaw by deriving a positive-semidefinite (PSD) extension of the theory using a rigorous framework based on the nonequilibrium Green’s function formalism. The PSD constraint introduces new scattering channels and triplet intermediate states and restores the correct physical behavior. We demonstrate that it consistently improves quasiparticle energies across standard molecular benchmarks, with an accuracy comparable to coupled-cluster reference calculations.","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"32 1","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149374730","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Local Correlation-Based Machine Learning Enables Coupled-Cluster-Level Energetics for Thousand-Atom Molecular Systems","authors":"Hua Feng,&nbsp;Shuhua Li,&nbsp;Jing Ma,&nbsp;Wei Li","doi":"10.1021/acs.jctc.6c01569","DOIUrl":"10.1021/acs.jctc.6c01569","url":null,"abstract":"<p>Data-intensive molecular workflows increasingly requireconsistentenergies for many molecules and configurations. High-accuracy dataare much scarcer because the small energy differences governing conformationalpreferences and molecular recognition can require an accurate treatmentof electron correlation. Coupled-cluster singles and doubles withperturbative triples [CCSD(T)] provides a widely used benchmark, butrepeated calculations become prohibitive for medium-to-large molecularsystems. Machine learning has enabled rapid coupled-cluster-qualitypredictions for small molecules, particularly by learning high-levelcorrections to lower-cost calculations, yet its extension to largesystems remains limited by expensive full-system labels and increasinglycomplex electronic representations. Here we develop cluster-in-moleculemachine learning (CIM-ML), which learns orbital-pair corrections fromCIM resolution-of-the-identity second-order Møller–Plessetperturbation theory (CIM-RI-MP2) to CIM domain-based local pair naturalorbital CCSD(T) [CIM-DLPNO–CCSD(T)]. Fixed CIM locality criteriadefine bounded learning environments, and the CIM energy expressionassembles local predictions into an extensive total energy. A modeltrained on one (H₂O)₁₂₈ conformer transfersto additional conformers and (H₂O)₂₃₇, reproducingconformational-energy separations within approximately 1 kcal mol^–1. For a 1,027-atom protein–ligand complex,training on 500 local clusters reduces the interaction-energy meanabsolute error from 12.92 to 3.98 kcal mol^–1. CIM-MLthus provides a domain-adaptable route for reusing high-level correlationinformation across large molecular ensembles.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11676–11686"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149151274","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A High-Performance Mesoscale Molecular Dynamics Framework for Polar Ordering in Dipolar Liquid Crystals","authors":"Yu-Wei Sun,&nbsp;Yong-Qing Chen,&nbsp;Tong Li,&nbsp;Zhan-Wei Li","doi":"10.1021/acs.jctc.6c01260","DOIUrl":"10.1021/acs.jctc.6c01260","url":null,"abstract":"<p>Polar ordering in liquid crystals arises from the interplaybetweenmolecular anisotropy and long-range dipole–dipole interactions,making molecular dynamics simulations of formation kinetics computationallydemanding due to the cost of evaluating these interactions. Here,we develop a high-performance mesoscale molecular dynamics frameworkfor efficient large-scale simulations of dipolar liquid-crystal systems.The framework incorporates GPU- and DCU-accelerated algorithms forefficiently computing anisotropic Gay-Berne forces and long-rangedipole–dipole interactions, and supports generalized Gay-Berneparticles carrying an arbitrary number of embedded dipoles with freelyassignable positions and orientations. The algorithm employs a multithreadedparallelization strategy with adaptive optimization, yielding significantspeedups on heterogeneous computing platforms. We validate the frameworkagainst representative ferroelectric and antiferroelectric systems,faithfully reproducing known phase behavior while maintaining computationalefficiency. This work provides a versatile molecular dynamics frameworkfor investigating polar ordering and its formation kinetics acrossa wide range of liquid-crystal systems and dipolar assemblies.</p>","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"22 21","pages":"11545–11556"},"PeriodicalIF":5.8,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149230155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Equivariant Interatomic Potentials without Tensor Products","authors":"Thiago Reschützegger,Sarp Aykent,Gabriel Jacob Perin,Bruno Henrique Nunes,Flaviu Cipcigan,Rodrigo Neumann Barros Ferreira,Mathias Steiner,Fabian L. Thiemann","doi":"10.1021/acs.jctc.6c00712","DOIUrl":"https://doi.org/10.1021/acs.jctc.6c00712","url":null,"abstract":"Foundational machine-learned interatomic potentials have emerged as powerful tools for atomistic simulations, promising near first-principles accuracy across diverse chemical spaces at a fraction of the cost of quantum-mechanical calculations. However, the most accurate equivariant architectures rely on Clebsch–Gordan tensor products whose computational cost scales steeply with angular resolution, creating a trade-off between model expressiveness and inference speed that ultimately limits practical applications. Here we introduce Geodite, an equivariant message-passing architecture that replaces tensor products while incorporating physical priors to ensure smooth, well-behaved potential energy surfaces. Trained on the Materials Project trajectories dataset of inorganic crystals, Geodite-MP achieves accuracy competitive with leading methods on benchmarks for materials stability prediction, thermal conductivity, phonon-derived properties, and nanosecond-scale molecular dynamics, while running 3–5× faster than models performing similarly. By combining predictive accuracy, computational efficiency, and physicality, Geodite enables faster large-scale atomistic simulations and high-throughput screening that would otherwise be computationally prohibitive.","PeriodicalId":45,"journal":{"name":"Journal of Chemical Theory and Computation","volume":"506 1","pages":""},"PeriodicalIF":5.5,"publicationDate":"2026-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149374731","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}