Publications by authors named "P A Houston"
Hydrocarbons are the central feedstock of fuels, solvents, lubricants, and the starting materials for many synthetic materials, and thus the physical properties of hydrocarbons have received intense study. Among these, the molecular flexibility and the power and infrared spectroscopies are the focus of this paper. These are examined for the linear alkane CH using molecular dynamics (MD) calculations and recent machine-learned potentials.
View Article and Find Full Text PDFPurpose: To investigate and explain observed features of the placental DWI signal in healthy and compromised pregnancies using a mathematical model of maternal blood flow.
Methods: Thirteen healthy and nine compromised third trimester pregnancies underwent pulse gradient spin echo DWI MRI, with the results compared to MRI data simulated from a 2D mathematical model of maternal blood flow through the placenta. Both sets of data were fitted to an intravoxel incoherent motion (IVIM) model, and a rebound model (defined within text), which described voxels that did not decay monotonically.
Hydrocarbons are ubiquitous as fuels, solvents, lubricants, and as the principal components of plastics and fibers, yet our ability to predict their dynamical properties is limited to force-field mechanics. Here, we report two machine-learned potential energy surfaces (PESs) for the linear 44-atom hydrocarbon CH using an extensive data set of roughly 250,000 density functional theory (DFT) (B3LYP) energies for a large variety of configurations, obtained using MM3 direct-dynamics calculations at 500, 1000, and 2500 K. The surfaces, based on Permutationally Invariant Polynomials (PIPs) and using both a many-body expansion approach and a fragmented-basis approach, produce precise fits for energies and forces and also produce excellent out-of-sample agreement with direct DFT calculations for torsional and dihedral angle potentials.
View Article and Find Full Text PDFArticle Synopsis
- Advances in machine learning have led to potentials that combine first-principles accuracy with faster evaluations, specifically using Δ-machine learning to enhance potential energy surfaces (PES) based on low-level methods like DFT.
- The study showcases successes with various molecules, demonstrating that the approach can achieve near-coupled cluster accuracy while testing its effectiveness using different functionals like PBE and M06 with ethanol as a case study.
- Results indicate significant optimizations in DFT gradients without needing coupled cluster corrections, and the findings highlight potential applications for improving molecular mechanics force fields.