Lab Home | Phone | Search
Center for Nonlinear Studies  Center for Nonlinear Studies
 Home 
 People 
 Current 
 Executive Committee 
 Postdocs 
 Visitors 
 Students 
 Research 
 Publications 
 Conferences 
 Workshops 
 Sponsorship 
 Talks 
 Seminars 
 Postdoc Seminars Archive 
 Quantum Lunch 
 Quantum Lunch Archive 
 P/T Colloquia 
 Archive 
 Ulam Scholar 
 
 Postdoc Nominations 
 Student Requests 
 Student Program 
 Visitor Requests 
 Description 
 Past Visitors 
 Services 
 General 
 
 History of CNLS 
 
 Maps, Directions 
 CNLS Office 
 T-Division 
 LANL 
 
Wednesday, June 05, 2013
3:00 PM - 4:00 PM
CNLS Conference Room (TA-3, Bldg 1690)

Seminar

Data-Driven Discovery of Physical, Chemical, and Pharmaceutical Materials

Barbara Jones
IBM Almaden Research Center

Data-driven insights have aided materials discovery in the pharmaceutical and related chemical areas for some years now, with even commercial products available. I will describe some successes in these areas, and derive lessons which might be applicable to the areas of condensed matter and polymeric materials. Just as experiment, theory, and computation must guide each other for MGI to succeed, a triangle of inter-relationships, for MGI it is really a tetrahedron, with computer science forming the fourth vertex. I will discuss the benefits which modern computer science can bring in the areas of modern data mining, machine learning, and big data analytics techniques. The volume of data on materials is fast-growing and scattered across many sources. While new tools and platforms have allowed the processing of vast volumes of data, our ability to integrate heterogeneous and unstructured data sets is still developing. The ability to correlate data from multiple sources deepens the value of data and allows new insights to emerge. A second point I will make is that the elements of accelerated materials discovery are different in the different scientific fields. Pharmaceutical discovery involves extracting chemical constituents and structures from patents; polymer data is scattered, unstructured, statistical and often ambiguous; and in condensed matter we tend to look at materials properties as a function of some parameter such as doping or temperature, often in graph form. Understanding the needs of both soft and hard condensed matter will help common tools and synergies to develop. Finally, there are many challenges ahead in fully incorporating data-enabled scientific discovery, as well as learning on both computer science and materials science sides. Getting scientific insights from both computer scientists and from data mining and databases is net yet common, and requires some work ahead in both communities to familiarize themselves to opportunities and to optimize the tools needed for future materials by design.

Host: Josephine Olivas