Cornell Workshop

Earlier this week (starting last Sunday), I was at Cornell University for a workshop on “Building a Collaborative Framework for Nanoscale Simulations”. In large part, most of the attendees were doing MD (molecular or materials dynamics) or some variety of electronics structure calculations.

Eventually most of the talks and slides will be available online – I’ll post when that happens.

Last Sunday evening, I gave a quick walkthrough of Avogadro. It was somewhat redundant, as over half of the participants had already used Avogadro. I emphasized some of the more unique features (display of spectra, super cell generation, interactive editing). Key questions:

  1. How big of a system can we handle?
  2. Who is funding development?
  3. Future directions & integration with other codes (e.g., Abinit).
  4. How hard it is to code extensions.

Fortunately, one of the Abinit developers, Matthieu Verstraete (Liege) was interested enough to start hacking an Abinit input generator on his Mac.

On Monday, I gave a presentation on Open Babel, Avogadro, ChemSpotlight, Quixote, and a few other emerging projects.

  1. This merges well with other new efforts for data repositories. I emphasized that I think Quixote’s model of a “federated” repository system held with RSS is very strong and likely to win out. Other talks had emphasized the risks of allowing untrusted contributors donate to a carefully-curated repository.
  2. Many people asked for an Open Babel for nanoscience and solid state. I assured them that almost all, if not all data models already exist in libopenbabel – it’s mostly an issue of adding file formats.
  3. There is a clear desire from solid-state DFT codes like Abinit to have a good visualization tool. If we build a file format or a code library for visualizing electron density, orbitals, etc., then they will convert to write that type of output.

One last note. The developer of NanoHub.org gave a talk – their infrastructure is really good. It allows users to run simulations through the browser using VMWare (or something) with no install needed.

  1. We should try to add an “export to NanoHub” extension to allow people to build in Avogadro and then export to run Abinit or GAMESS, etc. on NanoHub. They use authenticated HTTPS WebDav.
  2. They want us to try getting Avogadro on NanoHub directly, but we’d need to work with them a bit to run calculations directly.

-Geoff