news       |       manual       |       developers       |       downloads

Psi4Education     |     contacts     |     history     |     user forum


Psi4 Developers' Conference Nov 2014


Saturday afternoon session at Georgia Tech

Meeting Summary

The meeting was be held Friday and Saturday, Nov 14-15, at Georgia Tech featuring participants from 12 institutions and 4 countries. 30 people attended in person at Georgia Tech. The following are outcomes of the meeting:

  • A new distributed tensor class will be written for use by Psi4 going forward; work on this started at the meeting and will continue in November 2014. We will wrap existing distributed tensor libraries to maintain flexibility in the future; the API will be based on preliminary work by Francesco Evangelista
  • The SCF code will be rewritten, based on the new tensor class; it will incorporate GTFock
  • Chemical Semantics file (CSX) output will be added to Psi4 in the near future
  • Roberto DiRemigio's CMake system will be adopted, without file globs
  • Automatic Psi4 testing on various platforms will be set up, using ctest instead of current test system; Roberto DiRemigio has a prototype of this working
  • Compilation using Python 2.6 will continue to be supported
  • CC and CI codes will incorporate Wavefunction object more fully in the near future to (a) bring these codes more completely into Psi4 style, to (b) move forward with deprecating Psi3 checkpoint library, and to (c) test Wavefunction and make sure it has desired features; member reference_wavefunction will be kept for now in case it is needed later (but no clear use for it was identified)
  • C++11 will be incorporated more fully into Psi4, but serious efforts along these lines will be postponed until the standard is more widely supported by compilers; not all our users have upgraded to the latest compilers yet
  • Some compilation issues with Python 2.6 were resolved and segfaults on failed compare_values() checks were fixed
  • The next release will be made available after the CMake system has been thoroughly tested
  • Addition of psiArray counterpart to PsiVariable planned
  • Basis set handling extensively changed thanks to new Python object (report problems to Lori Burns)
  • Cleaner data handover from psi4.Matrix to NumPy imminent
  • We need a way to specify relative orientations of rigid fragments in input for use in molecular cluster optimizations (similar to EFP handling)
  • Agreement to update Psi4 logo and change tagline from "Ab Initio Quantum Chemistry" to "Open-Source Quantum Chemistry"
  • Developers provided status updates on a number of projects (see below)

Participating Institutions

  • Georgia Tech
  • Virginia Tech
  • University of Georgia
  • Emory University
  • Chemical Semantics, Inc
  • ETH, Zurich
  • Auburn University
  • NIH
  • Florida State University
  • Ataturk University, Turkey
  • University of Tromso, Norway
  • University of Memphis

Meeting Location

Georgia Tech, room 4202A of the Molecular Science and Technology building, 901 Atlantic Drive NW, Atlanta, GA. See here for directions for visitors and a list of nearby hotels.

Agenda

Friday, Nov 14

Saturday, Nov 15

  • 9:30am- 9:45am: Group discussion of any general topics from Friday
  • 9:45am-10:00am: Eugene DePrince: 2-RDM methods
  • 10:00am-12:00pm: Followup on Friday discussions, coding, debugging
  • 12:00pm: Meeting concludes, lunch on your own
  • Afternoon: optional continued discussions