Sandia National Labs FY20 LDRD Annual Report


Aerial cross-polarized nuclear quadrupole and magnetic resonance allows for explosive detection at a safe distance. New technology now allows for the safe and remote detection of nitrogen-containing explosive materials possibly buried in landmines or hidden in contraband in airport luggage or in/on passengers. During this LDRD, the PI conducted additional experiments, doping a polymer with quadrupolar nuclei. He showed that a stress in the polymer stresses the dopant particles, shifting the nuclear quadrupole resonance frequency. The techniques used to detect small amounts of explosive material quickly and safely will significantly enhance our nation’s ability to prevent casualties in warzones and improve security in the transportation sector. (PI: Eric Sorte) Automated analysis finds stable graph data. As communication is evaluated, relationships can be seen to change over time (people join/leave or new relationships form). Graphs formed from all communication events in a limited time period provide “snapshots” of the relationships. As change events are detected, a new graph snapshot should be generated from the communication events. Since graph snapshots are often used by downstream analysis tools to facilitate important national security decisions, it is critical to ensure that they properly represent the underlying communication

stream. A new technique developed at Sandia can now efficiently identify these fundamental changes in real-time. It was tested against synthetic, known-answer datasets (providing significant improvements over previous techniques) and against a variety of real-world datasets (highlighting previously unreported features). The technique is expected to be foundational to future graph analyses. (PI: Jeremy Wendt) Sandia’s technique automatically splits years of the Enron Email Dataset into 44 distinct static graphs of varying duration based on underlying shifts in communication patterns. A second step highlights periods of high similarity (lighter, white colors). This visualization shows that some of those high- similarity periods repeat across many months, interrupted by periods of low similarity.



Made with FlippingBook - professional solution for displaying marketing and sales documents online