DREU Andrea Howells

Daily Life

The bus trips are going smoothly now. I'm still without my phone though. I had ordered a new charger, but when it arrived, my phone still didn't take a charge. After doing some research on the model, I learned the connection in the phone frequently breaks, so I've ordered a battery charger.

On Wednesday, I met up with a friend that I haven't seen in almost 30 years! My best friend from then lives about 3 hours from Albuquerque and was even in town for training. It was so nice to see her and catch up on her life. She is the same sweet person she was then.

She took me to Old Albuquerque for some souvenir shopping, a mini tour, and dinner. The square we shopped in was beautiful. There was an open air concert at the time, so we got to listen to some nice music as we walked.

Research

The number of nodes rejected for various Pflex, Prigid and Angle Deviation values were graphed to evaluate the mapping process for three proteins, MHC, TCR and 1QLP. The sizes of the layers for were also graphed for evaluation.

MHC (A Chain) - with simultaneous decreases in Prigid and Pflex values, there is a corresponding decreasing number of rejected nodes. Setting the Node Generation value to 100 and the Num Sample value to 10 resulted in a decrease of rejects compared to setting the Node Generation value to 10 and the Num Sample value to 100.

TCR (D Chain)- with decreasing Prigid values, there is a corresponding decreasing number of reject nodes. (Variations in both Prigid and Pflex for D Chain was evaluated by Rachal).

The generation of the values took an increasing amount of time for TCR, MHC, and 1QLP, correlating to the size of these proteins. TCR took an average of ~ 2 hrs to run, MHC took an average of ~ 3.5 hrs to run, and 1QLP took an average of ~ 28 (Rachel's tests) and ~ 46.5 hrs (Andrea's tests) to run. We weren't able to immediately explain why the runs were so much longer for one set of 1QLP than for the other.

There was a noticeable disjoint when the 1QLP rejects for Angle Deviations of 0.5 to 2.5 were compared to the Angle Deviations for 3 to 40. It was expected that the number of rejects would be linear.

The code used to generate the mapping for single robots with articulated links was evaluated for changes that would be required to map multiple robots with articulated links. This will help determine behavior of the proteins when other proteins are present.

My website for DREU was accepted, with some suggestions for changes, which I implemented.

On Thursday, we attended a Cyber-Physical Systems presentation on Verification of Hybrid and Distributed Systems. We also attended a Brown Paper Bag Lunch session regarding the Application and Advising Process of UNM.

Next Week

For week 4, we will modify the make file and evaluate the number of joints seen after running the PRMRoadmap_testSerial.map file. The papers 'A Machine Learning Approach for Feature-Sensitive Motion Planning (M. Morales, L. Tapia, R. Pearce, S. Rodriguez, N. Amato) , which was postponed from week 3 and 'First-Class Control of HIV-1 (A. McMichael and E. Jones, 2010)' will be discussed, I will continue to revise and add to my DREU website and at the end of the week, I will submit a weekly report on the work done.