[Pycon] [new paper] "Giordano Mancini" - One formula, many molecules: using metaheuristics and Python to analyze chemical data.

[info a pycon.it]

Title: One formula, many molecules: using metaheuristics and Python to analyze chemical data.
Duration: 45 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: Molecules are flexible objects that may arrange themselves in many different “shapes” (conformers). Very often these yield different properties that is, a different chemistry and different macroscopic effects (colour, toxicity, drug activity to name a few). Hence the properties of a formula may actually depends on what structure do exist under given conditions. To perform this exploration using first principles amounts to generate many candidate molecular forms and evaluate them, sampling a very rugged surface. I will show how I use Python to walk surfaces, in the form of a wrapper able to call existing quantum chemistry codes (to provide energy or other properties) and run meta-heuristic optimizations (such as Genetic Algorithms or Particle Swarm Optimization) using this data.

Tags: [u'physics', u'statistics', u'python3']