arXiv subjects

P. D. Varuna S. Pathirage

Publications and source records attributed to P. D. Varuna S. Pathirage.

2 recordsLinked to original sources

arXiv2026

Time-Dependent Relativistic Two-Component Equation-of-Motion Coupled-Cluster for Open-Shell Systems: TD-EA/IP-EOMCC

We present a combined imaginary-time/real-time time-dependent (TD) approach for evaluating linear absorption spectra of open-shell systems at the electron attachment (EA) and ionization potential (IP) equation-of-motion coupled-cluster (EOMCC) levels of theory and within the exact two-component relativistic framework. The absorption lineshape is given by the Fourier transform of the electric dipole autocorrelation function, which is obtained from a real-time simulation. Approximations of the lowest-energy EA- and IP-EOMCC eigenstates, which are required as initial states for the real-time simulation, are generated by propagating a Koopmans EA/IP state in imaginary time. TD-EA/IP-EOMCC linear absorption spectra of open-shell atomic (Na, K, Rb, F, Cl, and Br) and diatomic (SiH and ClO) systems closely reproduce those obtained from standard TD-EA/IP procedures carried out in the frequency domain. We find that the existence of low-lying states with non-negligible overlap with the Koopmans determinant impacts the length of the imaginary-time propagation required to obtain an initial state that produces correct absolute energies and peak height intensities in spectra extracted from the subsequent real-time TD-EA/IP-EOMCC calculations.

physics.chem-ph↗

arXiv2026

DDCCNet: Physics-enhanced Multitask Neural Networks for Data-driven Coupled-cluster

We present the data-driven coupled-cluster deep network (DDCCNet), a family of multitask, physics-enhanced deep learning architectures designed to predict coupled-cluster singles and doubles (CCSD) amplitudes and correlation energies from lower-level electronic structure methods. The three DDCCNet variants (termed as v1, v2, and v3) progressively incorporate architectural refinements ranging from parallel subnetworks for t_1 and t_2 amplitudes to feature-partitioned blocks and physics-enhanced intermediate prediction layers that are structured in accordance with coupled-cluster equations to enhance physical consistency and multitask learning efficiency. These models jointly learn correlated amplitude patterns while embedding symmetry and orbital-level interactions directly into the network structure. Applied to methanol conformers, CO2 clusters, and small organic molecules, DDCCNet_v2 delivered the most accurate and transferable performance, achieving chemically precise correlation energies across diverse molecular systems. Collectively, DDCCNet establishes a scalable, physically grounded framework that unifies machine learning and ab initio theory for efficient, data-driven electronic structure prediction.

physics.chem-ph↗