Accurate and Scalable Continuum Electrostatics for Large Biomolecular Systems: The pyDelPhi Poisson-Boltzmann...

Electrostatic interactions are central to biomolecular structure, recognition, and assembly, making their efficient and accurate evaluation essential for biophysics, drug discovery, and materials modeling. The Poisson-Boltzmann equation (PBE) provides a rigorous implicit-solvent description, yet large macromolecular systems remain challenging for conventional CPU-based solvers. Here, we introduce pyDelPhi, a modern finite-difference PBE framework designed for numerical accuracy, scalability, and reproducibility across heterogeneous architectures. Implemented in Python with just-in-time compilation via Numba, pyDelPhi preserves the established DelPhi numerical framework while supporting traditional, Gaussian-density, and regularized dielectric models and enabling multiprecision execution on CPUs and NVIDIA GPUs via a CUDA backend. Benchmarking across three curated data sets shows that pyDelPhi reproduces DelPhi reaction-field energies within fractions of a percent and achieves 7-20× GPU acceleration for the linearized PBE. A cuboidal grid-box option reduces memory by up to 80% and accelerates runtimes several-fold for anisotropic systems. A complete viral-capsid calculation (∼5 × 10⁵ atoms, 10⁹ grid points) yields an order-of-magnitude reduction in wall time relative to the single-core DelPhi baseline while maintaining close numerical agreement. These results establish pyDelPhi as a unified and extensible platform that advances continuum electrostatics toward reproducible, high-performance modeling of biomolecular systems at both routine and large scales.