With Computers from Atoms to Macromolecular Systems

We review and update selected contributions to computational chemistry made since the late 1950s. Introductory remarks are given to place our work in the context of contemporary science. We start with a classical benchmark, the H(2) wave-function constructed with a new one-particle representation, the Chemical spin-Orbitals, which replaces the traditional Atomic and Molecular spin-Orbitals. Computations from diatomic to small polyatomic molecules, obtained with the Hartree-Fock-Heitler-London (HF-HL) model, are compared to those obtained from the traditional Hartree-Fock (HF) and Heitler-London (HL) methods; we conclude that the hierarchy of solutions within the HF-HL approach represents a general and reasonable choice for computational quantum chemistry. Further, we show that a wave function constructed with Chemical spin-Orbitals is equivalent to a wave-function obtained with the HF-HL model. These simulations are complemented with a critical analysis on the correlation energy, and on Wigner and Coulomb Hole functionals. The above studies follow the early Hartree-Fock period (1960-1970) characterised by pioneering computations on atomic and molecular systems, including basis set optimisation, atomic energy tabulation at the Hartree-Fock and post Hartree-Fock level, and potential energy surface computations obtained with the super-molecular approach. However, to deal with large molecular systems and to explicitly consider temperature and time, we must turn to statistical methods; we recall simulations using Monte Carlo, Molecular Dynamics, and Langevin dynamics, first at equilibrium, then for open systems at non-equilibrium. A concatenation of these models constitutes to the Global Simulation approach, discussed in detail. The above work requires both computer hardware and application codes in different areas of computational chemistry. We recall the quantum chemical atomic and molecular codes and the statistical mechanics codes written, documented and freely distributed for the last half century. Further, we recall our pioneering efforts in the early 1980s in computer architecture, with the design and assembly of a parallel supercomputer extensively used to perform the first parallel applications in computational chemistry.