Lithium diffusion in silicon battery anodes is governed by thermally activated jumps between (meta)stable sites separated by significant energy barriers, making such events rare on ab initio molecular dynamics (AIMD) time scales. To overcome this limitation, we establish a multiscale workflow that links AIMD, machine-learned force fields (MLFFs), and Markov state models (MSMs) to bridge atomistic mechanisms to mesoscale diffusion. Focusing on crystalline Li-Si phases, our MLFFs trained on AIMD data, achieve near-DFT accuracy while enabling large-scale molecular dynamics simulations extending to tens of nanoseconds. From these trajectories, we extract converged lithium-jump statistics to construct MSMs that quantitatively reproduce diffusivities with uncertainties an order of magnitude smaller than those obtained from 100 ps AIMD simulations. Demonstrated here for crystalline LixSiy phases, the AIMD → MLFF → MSM workflow provides a transferable route for quantitative transport modeling in amorphous structures, defect-mediated diffusion, and alternative solid-state anodes.

The valence intermediate effective Hamiltonian (VIEH) approach is a powerful tool for decomposing an overall magnetic coupling strength into ferromagnetic and antiferromagnetic contributions. Herein, we suggest an adaptation of the VIEH framework for complexes with multiple magnetic coupling pathways. This approach is showcased with a homologous series of [M2(μ-O)(NH3)n]2+ complexes, including Cu(II), Ni(II), and Fe(II). Typically, antiferromagnetic coupling progressively weakens from copper to nickel and iron. We demonstrate that this weakening arises primarily from an increase in the ferromagnetic coupling contribution relative to the antiferromagnetic one.

The adaptive derivative-assembled pseudotrotter variational quantum eigensolver (ADAPT-VQE) is a promising hybrid quantum-classical algorithm for molecular ground-state energy calculation, yet its practical scalability is hampered by redundant excitation operators and excessive measurement costs. To address these challenges, we propose Param-ADAPT-VQE, a novel improved algorithm that selects excitation operators based on a parameter-based criterion instead of a traditional gradient-based metric. This strategy effectively avoids redundant operators. We further developed a sub-Hamiltonian technique and integrated a hot-starting VQE optimization strategy, achieving a significant reduction in measurement costs. Numerical experiments on typical molecular systems demonstrate that Param-ADAPT-VQE outperforms the original ADAPT-VQE in ansatz size, computational accuracy, and measurement costs. Furthermore, our scheme retains the fundamental framework of ADAPT-VQE and is thus fully compatible with its various modified versions, enabling further performance improvements in specific aspects. This work presents an efficient and scalable enhancement to ADAPT-VQE, mitigating the core obstacles that impede its practical implementation in molecular quantum chemistry.

Fragmentation-based quantum chemistry offers a practical route to achieving ab initio quality energetics for proteins, but its applicability is often constrained by the high computational cost of fragment-level quantum mechanical (QM) calculations. Here, we introduce GMFCC-UMA, a method that integrates the generalized molecular fractionation with conjugate caps (GMFCC) framework with a fine-tuned foundation neural network potential, thereby eliminating the need for fragment QM calculations. In this approach, the total protein energy is decomposed into adjacent and nonadjacent interaction components. Adjacent terms are assembled in an MFCC II-consistent manner from ACE-NME-capped fragments, while nonadjacent interactions are treated hierarchically: near-contact pairs are evaluated using the fine-tuned UMA model, and longer-range contributions are described via a complementary molecular mechanics (MM) treatment. The underlying neural network model is fine-tuned on a large, protein-focused fragment data set that includes capped monomers, adjacent dimers, and nonadjacent residue pairs and corresponding blocks, ensuring consistent accuracy across all fragment types required by the GMFCC fragmentation scheme. Across benchmark proteins and conformational ensembles, GMFCC-UMA yields relative energies that closely match quantum-based fragmentation references and systematically outperforms conventional force fields in both error reduction and correlation. By replacing fragment-level QM with neural network inference, GMFCC-UMA achieves an order-of-magnitude acceleration while retaining ab initio fidelity, thereby enabling scalable and accurate protein energy evaluation for high-throughput conformational analysis and related applications.

Shallow ensembles provide a convenient strategy for uncertainty quantification in machine-learning interatomic potentials, which is computationally efficient because the different ensemble members share a large part of the model weights. In this work, we systematically investigate training strategies for shallow ensembles to balance calibration performance with computational cost. We first demonstrate that explicit optimization of a negative log-likelihood (NLL) loss improves calibration with respect to approaches based on ensembles of randomly initialized models or on a last-layer Laplace approximation. However, models trained solely on energy objectives yield miscalibrated force estimates. We show that explicitly modeling force uncertainties via an NLL objective is essential for reliable calibration, though it typically incurs a significant computational overhead. To address this, we validate an efficient protocol: full-model fine-tuning of a shallow ensemble originally trained with a probabilistic energy loss or one sampled from the Laplace posterior. This approach yields a calibration quality comparable to training from scratch. While the computational benefits scale with system and model size, it reduces training time by up to 96% for data sets containing large structures. We evaluate this protocol across a diverse range of materials, including amorphous carbon, ionic liquids (BMIM BF4), liquid water (H2O), barium titanate (BaTiO3), and a model tetrapeptide (Ac-Ala3-NHMe), establishing practical guidelines for reliable uncertainty quantification in atomistic machine learning.

The pKa values of titratable residues are fundamental parameters that shape protein behavior in a given solution environment. They govern electrostatic interactions and thereby influence protein folding, conformational dynamics, and molecular recognition. Existing pKa-prediction algorithms typically rely on detailed all-atom structures. However, although coarse-grained models are widely used to simulate protein structure and dynamics, estimating pKa values within these frameworks remains challenging because ionizable groups are often not explicitly represented. In this work, we introduce DeepCGpKa, a deep-learning-based pKa predictor designed for coarse-grained protein structures. Benchmark tests show that DeepCGpKa attains accuracy comparable to state-of-the-art methods based on all-atom structures. It also retains robust predictive performance for partially unfolded protein structures. When coupled with coarse-grained molecular simulations, DeepCGpKa successfully captures the pH dependence of protein conformational changes. Integrating data-driven pKa prediction with physics-based molecular simulations provides a practical route to improve the treatment of electrostatic interactions at the coarse-grained level, which is a critical issue for most currently available coarse-grained biomolecular models.

Chronic infection with the hepatitis B virus (HBV) remains a major global health burden and depends on efficient viral replication, including the assembly of the viral capsid during the HBV life cycle. Capsid assembly proceeds through transient oligomeric intermediates, among which the formation of hexameric units is thought to underlie the energetic bottleneck associated with nucleation during capsid assembly. Despite extensive experimental and computational work, the structural and energetic determinants of hexamer closure remain incompletely understood at the molecular level. Here, we employ a multistage computational approach to investigate the open-to-closed transition of the HBV capsid hexamer in the apo system, in which targeted molecular dynamics is employed to generate diverse open-to-closed transition pathways. These pathways are subsequently refined and sampled by using path-based free-energy methods to construct multidimensional free-energy landscapes of hexamer closure. Across independently sampled pathways, we observe diverse transition routes, while a conserved steric rearrangement in the gate region emerges as the dominant rate-limiting feature. These results provide a qualitative characterization of hexamer closure energetics and establish a general framework for studying complex conformational transitions in large, flexible biomolecular assemblies.

There is considerable interest in being able to address the temperature dependence of enzyme reactions by computer simulations. One reason for this is that enzymes that are adapted to different temperature regimes generally show distinct signatures in terms of their activation enthalpies and entropies, to the extent that it is basically possible to predict whether an enzyme is psychrophilic or mesophilic just by examining these activation parameters. The standard approach to this problem is to calculate reaction free energy profiles at a series of different temperatures. Computational Arrhenius plots can then be constructed from the data, analogous to the experimental procedure. This method has been shown to work well in a number of cases that have examined orthologous pairs of psychrophilic and mesophilic enzymes. The drawback is that the simulations have to be repeated at several temperatures. However, while multitemperature simulations may be computationally demanding they can be informative in revealing deviations from linear Arrhenius behavior. Another issue is that the calculated activation enthalpies obtained in this way cannot be readily decomposed into different energy terms. An alternative approach to the problem would be to just carry out free energy simulations at a single temperature and instead obtain the enthalpy profile by plain averaging of the total energy. The entropy term would then simply be calculated as the difference between free energy and enthalpy. This averaging approach was earlier considered unreliable due to convergence problems for the total energy, even for moderately sized systems. Here, we re-examine the performance of the averaging method for two solution reactions and one enzyme reaction and conclude that it works surprisingly well with sufficient data. This opens up new ways of analyzing nonlinearity of Arrhenius plots in terms of energetics, since the enthalpy is decomposable.

Analyzing routes of transport for open quantum systems with nonequilibrium initial conditions, such as exciton transport aggregates, is extremely challenging. The state-to-state approach [A. Bose, and P. L. Walters, J. Chem. Theory Comput. 2023, 19, 15, 4828-4836] has proven to be a useful method for understanding transport mechanisms in quantum systems interacting with dissipative thermal baths. However, real systems are often exposed to processes that may lead to either increase or decrease of the number of excitations in the aggregate. Such pumping or draining processes can also affect the routes of transport. We extend the state-to-state analysis to account for approximate Lindbladian descriptions of generic dissipative, pumping and decohering processes acting on a system, which is exchanging energy with a thermal bath. The exchange of energy between the system and the environment is incorporated in a numerically exact manner. This Lindblad state-to-state analysis framework is able to unravel the internal transport pathways along with the effect of the external empirical processes and how the thermal solvents modulate these transport routes. Using this new state-to-state formalism, we demonstrate different mechanistic aspects, including the establishment of steady-state excitonic currents in molecular aggregates under the simultaneous influence of pumps and drains whose dynamics is simulated using the path integral Lindblad dynamics [A. Bose, J. Phys. Chem. Lett. 2024, 15, 12, 3363-3368]. It is especially lucrative that in the absence of such processes, the current method reduces to the standard state-to-state approach. We believe that this Lindblad state-to-state method promises to be a unique tool for understanding the dynamics of open quantum systems subject to a host of additional processes with unprecedented granularity, enabling unique questions to be asked about these systems of great complexity.

Machine-learning potentials (MLPs) have become important tools for modern molecular simulations. However, developing models that simultaneously achieve high accuracy and high computational efficiency remains a significant challenge. In this work, we extend the recently proposed MB-PIPNet framework to general covalently bonded molecular systems by combining monomer-based energy decomposition, permutationally invariant polynomial (PIP) descriptors, and neural networks within a fragmentation-based strategy. Within this framework, the total potential energy is represented as a sum of effective monomeric contributions, where PIPs provide compact and chemically motivated descriptions of both monomer internal structures and their local chemical environments. As a proof-of-concept application, we apply the MB-PIPNet framework to linear alkanes, using C14H30 as a representative system, and benchmark its performance against established atomistic machine-learning models. The resulting MB-PIPNet potential accurately reproduces reference ab initio electronic energies and reliably captures key molecular properties, including torsional potential energy profiles, harmonic vibrational frequencies, and vibrational power spectra obtained from molecular dynamics simulations. Importantly, MB-PIPNet demonstrates a substantial advantage in computational efficiency over other MLP models for combined energy and force evaluations. These results establish MB-PIPNet as a scalable and efficient framework for constructing MLPs, providing an additional route for large-scale quantum and classical simulations of complex molecular systems.