Benchmark Data and Non-Simulation Predictive Methods

Description

Our vision is that prediction methods will be an important part of the industrial toolbox for property prediction and will be used in conjunction with Simulation methods to solve important industrial problems. Innovative predictive methods must be encouraged, and established methods must be improved. Simulation methods will be routinely and effectively used for validation of these predictive methods.
 
Definition of non-simulation methods
 
any method which does not neatly fit into the definition of molecular simulation (MS):
 
Molecular simulation refers to any method that generates a trajectory through phase space for the system of interest by executing a series of deterministic and/or stochastic steps that obey the thermodynamic constraints. (Ilja Siepmann, 2006)
 
Goals
 
Obtain buy-in to strategic plan by key stakeholders.

Identify publicly available resources (commercial and academic); provide links and test cases.

Best place for this might be our web site, or use existing NIST framework from NIST internal web site.

Explore the possibility of developing a flow chart/web tool for best methods based on property, material and conditions.

  • DIPPR has started this.
  • Best to hire a post-doc at NIST to complete.
  • Start with a list of properties and methods, then prioritize based on industrial need.
  • This activity aligns well with the Challenge.
  • Need to ask public for input!
  • Dow has put together a straw man on this (recently published by Jim Olson et al.) and will share with team.
  • Note that John O'Connell tried to set up a similar scheme for Eqn. of State methods about 4-5 years ago. We don't think that came to fruition but we assume that John would be interested in our proposal.
  • Tool could be web-based and owned/administered by NIST.

Develop a process to provide benchmark data for periodic Challenges.
  • ensure that the resources are available to perform testing
  • ensure that the materials are available in sufficient purity
  • ensure that the state points are experimentally accessible within reason
  • ensure that the "answers" are not already in the open literature (if that is desired)
  • ensure that the testing is completed in a timely fashion
  • Offer a periodic challenge in some form to test methods and stimulate development of new methods.

Develop a flow chart/web tool for best methods based on property, material and conditions.

 

Subscribe to Comments for "Benchmark Data and Non-Simulation Predictive Methods"