BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//ZContent.net//ZapCalLib 1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
SUMMARY:The Future of Enhanced Sampling Methods
LOCATION:Chemistry A101
TZID:America/Denver
DTSTART:20191001T000000
UID:2026-04-25-03-08-29@natsci.colostate.edu
DTSTAMP:20260425T030829
Description:Literature Seminar\nComputational modeling techniques\, such as
  molecular dynamics\, have provided valuable insight into chemistry over t
 he years and continue to advance the field concurrently as these methods a
 re developed to be more efficient and accurate. One of the greatest limita
 tions molecular dynamics faces is computational efficiency\; many interest
 ing dynamics occur on a timescale that is currently unfeasible to simulate
  within a reasonable amount of time. A variety of approaches to combat thi
 s issue exist\, including enhanced sampling techniques. Enhanced sampling 
 requires the definition of a reaction coordinate that describes the proces
 s of interest\, and subsequently the ability to bias simulations to sample
  along the reaction coordinate while reducing the natural energy barrier\,
  thus increasing the speed of that process. While this method has been suc
 cessful\, the reaction coordinate is not always clear\, and choosing a poo
 r reaction coordinate will result in researcher-frustration and wasted tim
 e. The Tiwary group at the University of Maryland has developed a robust m
 achine learning model formalized on the Predictive Information Bottleneck 
 theory that optimizes a reaction coordinate and subsequently performs bias
 ed simulations to sample the process of interest. The application of machi
 ne learning in the field of molecular dynamics is still new\, however I pr
 opose this combination will continue to develop and impact the future of e
 nhanced sampling by minimizing the amount of human input required. 4:00 pm
END:VEVENT
END:VCALENDAR