TeraChem: Accelerating electronic structure and ab initio molecular dynamics with graphical processing units.

Developed over the past decade, TeraChem is an electronic structure and ab initio molecular dynamics software package designed from the ground up to leverage graphics processing units (GPUs) to perform large-scale ground and excited state quantum chemistry calculations in the gas and the condensed phase. TeraChem's speed stems from the reformulation of conventional electronic structure theories in terms of a set of individually optimized high-performance electronic structure operations (e.g., Coulomb and exchange matrix builds, one- and two-particle density matrix builds) and rank-reduction techniques (e.g., tensor hypercontraction). Recent efforts have encapsulated these core operations and provided language-agnostic interfaces. This greatly increases the accessibility and flexibility of TeraChem as a platform to develop new electronic structure methods on GPUs and provides clear optimization targets for emerging parallel computing architectures.

Generating Efficient Quantum Chemistry Codes for Novel Architectures.

We describe an extension of our graphics processing unit (GPU) electronic structure program TeraChem to include atom-centered Gaussian basis sets with d angular momentum functions. This was made possible by a "meta-programming" strategy that leverages computer algebra systems for the derivation of equations and their transformation to correct code. We generate a multitude of code fragments that are formally mathematically equivalent, but differ in their memory and floating-point operation footprints. We then select between different code fragments using empirical testing to find the highest performing code variant. This leads to an optimal balance of floating-point operations and memory bandwidth for a given target architecture without laborious manual tuning. We show that this approach is capable of similar performance compared to our hand-tuned GPU kernels for basis sets with s and p angular momenta. We also demonstrate that mixed precision schemes (using both single and double precision) remain stable and accurate for molecules with d functions. We provide benchmarks of the execution time of entire self-consistent field (SCF) calculations using our GPU code and compare to mature CPU based codes, showing the benefits of the GPU architecture for electronic structure theory with appropriately redesigned algorithms. We suggest that the meta-programming and empirical performance optimization approach may be important in future computational chemistry applications, especially in the face of quickly evolving computer architectures.

Controllable synthetic molecular channels: biomimetic ammonia switch.

We use molecular dynamics simulations combined with iterative screening to test if one can design mechanically controllable and selective molecular pores. The first model pore is formed by two stacked carbon nanocones connected by aliphatic chains at their open tips, in analogy to aquaporins. It turns out that when one nanocone is gradually rotated with respect to the other, the molecular chains alter the size of the nanopore formed at the cone tips and control the flow rates of liquid pentane through it. The second model pore is formed by two carbon nanotubes joined by a cylindrical structure of antiparallel peptides. By application of a torque to one of the nanotubes, while holding the other, we can reversibly fold the peptides into forward or backward-twisted barrels. We have modified the internal residues in these barrels to make these pores selective and controllable. Eventually, we found a nanopore that in the two folded configurations has very different transmission rates for hydrated NH(3) molecules.

Sandwiched graphene--membrane superstructures.

We demonstrate by molecular dynamics simulations that graphene sheets could be hosted in the hydrophobic interior of biological membranes formed by amphiphilic phospholipid molecules. Our simulation shows that these hybrid graphene--membrane superstructures might be prepared by forming hydrated micelles of individual graphene flakes covered by phospholipids, which can be then fused with the membrane. Since the phospholipid layers of the membrane electrically isolate the embedded graphene from the external solution, the composite system might be used in the development of biosensors and bioelectronic materials.

Modeling the self-assembly of colloidal nanorod superlattices.

We model the self-assembly of superlattices of colloidal semiconducting nanorods horizontally and vertically oriented on material substrates. The models include van der Waals and Coulombic coupling between nanorods with intrinsic electric dipoles and their coupling to the substrates. We also investigate the effect of external electric fields on the self-assembly processes. Our theoretical predictions for stable self-assembled superlattices agree well with the available experimental data.

Dipole-dipole interactions in nanoparticle superlattices.

Nanoparticles often self-assemble into hexagonal-close-packed (hcp) structures although it is predicted to be less stable than face-centered-cubic (fcc) packing in hard-sphere models. In addition to close-packed fcc and hcp superlattices, we observe formation of nonclose-packed simple-hexagonal (sh) superlattices of nearly spherical PbS, PbSe, and gamma-Fe2O3 nanocrystals. This surprisingly rich phase diagram of monodisperse semiconducting nanoparticles is explained by considering the interactions between nonlocal dipoles of individual nanoparticles. By calculating the total electrostatic and dispersive energies, we explain stability of the hcp and sh nanoparticle superlattices, introduce the superlattice phase diagram, and predict antiferroelectric ordering in dipolar nanoparticle superlattices.