748 papers
This paper introduces SmileyLlama, a large language model transformed via supervised fine-tuning and direct preference optimization to function as a chemical language model that reliably generates novel drug-like molecules with user-specified properties and optimized 3D conformations.
This paper introduces the Seniority-zero Linear Canonical Transformation (SZ-LCT) method, which employs a specialized unitary transformation to map the electronic Hamiltonian into the computationally efficient seniority-zero space, achieving highly accurate results for strongly correlated systems with submilliHartree errors and favorable computational scaling.
This paper introduces a Seniority-Zero Canonical Transformation Theory that achieves high accuracy ( Hartree) for small- to medium-sized systems by exactly evaluating the first three commutators of a unitary transformation applied to a seniority-zero reference wavefunction, while approximating higher-order terms to reduce truncation errors.
This paper demonstrates that applying a quantum eigenvalue estimation algorithm to non-Hermitian transcorrelated Hamiltonians with the xTC approximation can achieve chemical accuracy comparable to standard qubitization on significantly larger basis sets for small atoms, though the method's accuracy degrades for larger second-row elements.
This paper demonstrates that combining Sample-based Quantum Diagonalization (SQD) with Density Matrix Embedding Theory (DMET) enables accurate, chemically precise ground-state energy calculations for complex, low-symmetry ligand-like molecules on IBM's Eagle R3 quantum hardware by effectively managing subsystem-dependent entanglement variations.
This study investigates the fluorescence lifetime response of molecular rotors in ternary PEG mixtures, demonstrating that a linear mixing rule applies to these systems and providing data to critically evaluate the semi-quantitative accuracy of free volume theory in characterizing microviscosity.
This paper introduces a model charge density approach to Ewald summations that accelerates convergence by canceling multipole moments of the crystalline charge distribution, a method validated on gallium arsenide and shown to clarify long-standing implementations in the CRYSTAL code.
The paper introduces El Agente Estructural, a multimodal, natural-language-driven AI agent that mimics human experts to perform precise, context-aware 3D molecular editing and geometry manipulation through the integration of vision-language models and domain-specific tools, thereby enabling complex chemical tasks like site-selective functionalization and stereochemical control without rebuilding entire molecular frameworks.
This paper demonstrates that incorporating quantum-chemical bonding descriptors into machine learning models significantly improves the prediction of elastic, vibrational, and thermodynamic properties of approximately 13,000 solid-state materials while also enabling the discovery of intuitive physical expressions for these properties.
This paper introduces a physics-aware machine learning surrogate for the 3D stochastic Cahn-Hilliard equation that parameterizes inter-cell fluxes to guarantee mass conservation and thermodynamic interpretability, enabling the accurate simulation of noise-driven phenomena like nucleation and coarsening with significant generalization to larger spatial and temporal scales.
By employing interpretable machine learning models on first-principles molecular descriptors, this study reveals that human odor perception lacks a universal physicochemical determinant, instead relying on heterogeneous, odor-specific patterns of feature importance that reflect unique structure-odor relationships for each scent.
This paper investigates how interference effects in dense molecular nanolayers near reflecting surfaces lead to non-monotonic absorption behavior, revealing that while free-standing films are limited to 50% absorption due to symmetry, mirror-backed configurations can achieve unity absorption through critical coupling by balancing radiative leakage with intrinsic loss.
This first-principles study investigates how sodium intercalation into layered potassium birnessite alters its structural stability, ion diffusion barriers, vibrational modes, and electronic properties, revealing that the process induces significant lattice distortions and transforms the material into a tunable bipolar magnetic semiconductor with potential applications in energy storage and spintronics.
This paper explores how concepts from symplectic topology, particularly Gromov's non-squeezing theorem, offer a novel geometric framework for understanding classical reaction dynamics near saddle points by identifying symplectic width scales that reveal how initial phase-space distributions influence reactivity and reaction bottlenecks beyond traditional flux-based metrics.
CovAngelo is a hybrid quantum-classical platform that utilizes a novel QM/QM/MM embedding model and quantum-information metrics to accurately and scalably model ligand-protein binding reactions, such as the covalent docking of zanubrutinib, while demonstrating potential speedups on current and future quantum hardware to improve drug discovery efficiency.
By combining deep potential molecular dynamics with high-fidelity electronic-structure data, this study resolves decades of uncertainty to precisely locate aluminum's liquid-vapor critical point at approximately 6531–6576 K, 0.637 g/cm³, and 1.6 kbar, establishing a transferable framework for predicting critical phenomena in metals under extreme conditions.
This paper investigates quantum reaction dynamics at an index-1 saddle using a Weyl-symbol formulation of quantum normal forms, revealing that extreme squeezing of transverse bath modes induces a geometric suppression of transmission by depleting effective reactive energy, thereby establishing a concrete link between squeezed-state covariance geometry and quantum reactivity consistent with classical symplectic width concepts.
This theoretical study compares wavepacket interference and impulsive stimulated Raman scattering (ISRS) descriptions of pump-probe spectroscopy, demonstrating that accurate modeling of excited-state absorption requires accounting for non-adjacent vibrational coherences and highlighting the dominant role of the coherent anti-Stokes pathway under specific spectral conditions.
This paper introduces opt-DDAP, a reformulated density-derived atomic point charge method that leverages automatic differentiation to optimize Gaussian basis parameters and reciprocal-space cutoffs, thereby overcoming the numerical instability and reliance on fixed heuristics of the original approach to produce robust charges for long-range electrostatic modeling.
This paper proposes a computational method and algorithm that models molecular cages as constrained graphs to automatically generate substrate-specific structures by connecting binding patterns into the smallest possible molecular paths, capable of constructing cages with over a hundred atoms.