Our latest paper, PSI4 1.1: An Open-Source Electronic Structure Program Emphasizing Automation, Advanced Libraries, and Interoperability, is available from the Journal of Chemical Theory and Computation.
PSI4 1.1 is released!
The new version is a proper Python library (with C++ underneath) that can
be called from a Python script using
This version adds RHF analytic Hessians, RHF CCSD(T) gradients,
functional group and intramolecular SAPT, high-spin open-shell SAPT,
Grimme's semi-empirical HF-3C and PBEh-3C, optional integrals package
SIMINT, density-fitted MCSCF, and
Eugene DePrince, Lori Burns, Ryan Fortenberry, and David Sherrill hosted the first PSI4 Users' Workshop at the American Chemical Society's Florida Meeting and Exposition (FAME) conference on May 4, 2017.
David Sherrill hosted a PSI4Education workshop Dec 19-20, 2016 at Georgia Tech. We discussed some new computational chemistry lab modules using PSI4 1.1.
Justin Turney hosted the annual PSI4 workshop at the University of Georgia, Nov 4-5, 2016. Meeting summary and talks
Ryan Richard gave a talk on the Pulsar project and interfacing it to PSI4 as part of the symposium on Emerging Technologies in Computational Chemistry at the 2016 Fall National ACS meeting in Philadelphia.
PSI4 1.0 is now available!
Lead developers Lori Burns and Daniel Smith gave talks at the SciPy 2016 conference: PSI4: A Case Study on Modernizing and Modularizing Quantum Chemsitry with Python and C++, and QCDB Database Tools for Managing and Harmonizing Quantum Chemistry.
Our YouTube Channel is now active with news and tutorial videos.
A PSI4 Developers' Workshop was held at Emory Nov 13-14, 2015.
The PSI4 team is part of a collaborative, multi-PI grant recently funded by the NSF for software development under the Sustainable Software Infrastructure (SSI) program. The project, Removing Bottlenecks in High Performance Computational Science, is led by Mark Gordon (Iowa State) and involves interoperability between multiple program packages (PSI4, GAMESS, NWChem, AIMS, and others) and the creation of reusable software components.
PSI4 contributor Daniel Smith (Patkowski group, Auburn University) is a winner of the Chemical Computing Group Excellence Award for Graduate Students and will present a poster on his project, Psi4NumPy: A Hybrid C++/Python Interpreted Quantum Chemistry Programming Environment at the Fall 2015 National ACS Meeting in Boston.
The PSI4Education team's book chapter, "Psi4Education: Computational Chemistry Labs using Free Software," has been accepted for publication in The Promise of Chemical Education: Addressing our Students' Needs a volume of the ACS Symposium Series.
A Psi4 Developers' Workshop was held at Georgia Tech Nov 14-15, 2014.
Psi4 co-PI David Sherrill was interviewed by Chemical & Engineering News about reusable software components.
The Psi4Education project has launched their open source lab manual. Psi4Education aims to provide free resources for chemistry educators to enable them to include computational chemistry lab activities at all levels of the chemistry curriculum. The lab manual currently features 7 labs, which use the Psi4/WebMO interface, with plans to expand to 12 labs by the end of 2014. A Psi4Education workshop was held at the Biennial Conference on Chemical Education at Grand Valley State University in Grand Valley, Michigan on August 5, 2014. At the workshop, 25 chemistry educators from across the US got hands-on experience using the Psi4/WebMO interface and the Psi4Education labs.
Our popular SAPT code contains a number of different truncations of the SAPT perturbation expansion, offering users quite a few choices. We examine the accuracy vs. efficiency tradeoffs for all these choices in J. Chem. Phys. 140, 094106 (2014) (DOI: 10.1063/1.4867135) Note: SAPT0 works best with the jun-cc-pVDZ basis (it's built into Psi4). Errors for SAPT0 grow with larger basis sets. We are currently updating the SAPT printouts to conform to recommendations in this paper.
The Psi4 code was used to develop a graphics processing unit (GPU) accelerated density-fitted coupled-cluster (DF-CCSD) code. See Mol. Phys. 112, 844-852 (2014) (DOI: 10.1080/00268976.2013.874599). The GPU-accelerated code is available from GitHub here.
The Beta5 release includes very efficient Density Fitted and Cholesky Decomposition (DF/CD) CCSD(T), written by Eugene DePrince. These techniques represent the usual four-index electron repulsion integrals by three-index quantities, tremendously reducing time spent in I/O reading or writing the integrals from/to disk. The speedup is especially noticeable on machines without fast striped local disks. The DF/CD approximations work very well in concert with Frozen Natural Orbitals, which are also available. An analysis of the new algorithm is presented in J. Chem. Theory Comput. 9, 2687-2696 (2013) (DOI: 10.1021/ct400250u)
Our frozen natural orbital (FNO) coupled-cluster module substantially speeds up accurate CCSD(T) computations. This module, FNO-CC, was written by Eugene DePrince and is very fast for RHF-based coupled-cluster computations, even without the FNO approximation. However, FNO's speed up computations even more. A thorough analysis of how the FNO approximation affects non-covalent interaction energies is presented in J. Chem. Theory Comput. 9, 293 (2013) (DOI: 10.1021/ct300780u)
Analytic gradients of optimized-orbital second-order perturbation theory (OMP2) have been added to Psi4. The theory and code are described in J. Chem. Phys. 138, 184103 (2013) (DOI: 10.1063/1.4803662)
A preview of the Psi4 package is provided in Wiley Interdisciplinary Reviews: Computational Molecular Science, 2, 556 (2012) (DOI: 10.1002/wcms.93)