VIRTUAL
 
Thursday, September 17th , 2020
3:45 – 4:45 p.m.
WEBEX
 
Join meeting
 
  
Speaker: Michael Ham
 
  P-23: NEUTRON SCIENCE & TECHNOLOGY
LANL
 
 
“(U) Data Curation and Standards - Legacy Data, Existing Data, and Future Data.”
 
 
 
Abstract: 
 
At this time, LANL is identifying data sources necessary for the Common Model Framework (CMF) and establishing ownership of data and pedigree of data for reliability, optimization of codes and preparation for a machine learning future. However, data management needs at LANL are much bigger extend beyond the CMF and this talk delves into our data management issues; past, present and future.  
 
For example, predictive simulations depend on a wide variety of data sources for setting variables and validating output. Due to legacy practice, data are typically ‘owned’ and maintained by the analysts or the group the analysts once worked for. These vital data sources live in a many separate locations and have various levels of documentation (pedigree). The ASC V&V Data Validation and Archiving team and CMF are undertaking the process of documenting the state of data and establishing who owns what data and where it currently resides. The end goal is to ensure that paths to data are accurate and link to the most up to date information along with establishing a data owner that can assist when questions arise. 
 
To set an example for best practice, ASC V&V DV&A is focusing on integrating legacy UGT gamma and neutron data into the CMF in the most seamless way possible. This data set in particular is tricky as there may be multiple analysis of an event due to decades of analysts addressing different issues for designers. The current effort is identifying the most trusted analysis and assigning each one a pedigree. That data is then uploaded to a repository that CMF accesses. Here we discuss setting up the repository such that legacy data are easily integrated and future data can also be added in a robust manner.
 
In addition to making the CMF workflow more robust and trustworthy, the effort is anticipated to allow machine learning and AI algorithms to make better use of this invaluable experimental and simulation data. As the laboratory evolves to address the opportunities laid out in the Strategic Investment Plan IS&T section, being able to quickly and accurately access the treasure trove of unique LANL data will allow LANL and the NNSA greater agility and faster response time to serve America’s defense needs.
 
 
 
 
WEBEX etiquette:
Please mute your microphones.
Log into meeting few minutes before talk starts
Refrain to use chat for side conversations
Use chat exclusively for questions that organizer/host will ask to speaker
Only turn your mic on when asked to do so by organizer/host
If you have questions, or comments please contact allobet@lanl.gov