Next week Back to topDaily Life
After much anticipation and preparation, I boarded my flight in Albany, NY on 29 May 2011 and arrived in Albuquerque, NM. Dr. Tapia had graciously offered to pick me up and sent me a few reassuring texts around the time of my arrival. She arranged on-campus housing for us (my good luck just keeps going!), and she drove me to the campus and waited to be certain that I was able to check in. She also gave me a copy of a paper to read that we would discuss later in the week.
On my first day here, I walked about the campus a bit and found a Wendy's for dinner. On my second day, I learned the mailing address and was able to order a laptop. Later, I walked to Walgreen's for groceries. When I returned, the other undergraduate intern in the program, Rachel Webster, had checked into a room in the same suite. Her report can be found in the DREU intership links as well. On day three, we began work on the internship.
The weekend arrived, as weekends tend to do... On Saturday, I tried out the Albuquerque Public Bus system and made my way to a grocery store. It was an interesting experience for me as I'm from an area with very limited public transportation. Consistant with my experience since arriving here, everyone I spoke with was helpful and nice.
Research
The project is a three year project to investigate the role of protein motion in determining binding affinities between peptide, MHC, and TCR molecules. This information will be used to understand the specificity of immune responses which can then be used in vaccine research. Over the course of this first week, we were introduced to the project and to a road map program. We ran this program and changed a number of parameters to attempt to learn which paramaters affected node generation.
On Tuesday and Wednesday we familiarized ourselves with a node generator, BasicPRMStategy for pathfinding, and viewed the results of changes to parameters using Vizmo. We presented our findings on Wednesday. It appeared the best method to use is MAPRM, using the distance metric for euclidean scaling and reducing the euclidean scale from 0.5 to 0.25, and increasing the K value for Neighborhood Connections from 10 to 50. The combinations of changes were run resulting in: Prior to change Parameters: Default Connected Components : 2(99,1) Number of Edges : 582, 0 Comments/Visual Assessment:There are large unexplored areas to the sides. The nodes appear evenly distributed to the top and bottom of the obstacles with some of the larger component nodes appearing between the obstacles as well. After change to Parameters: See above Connected Components : 1(100) Number of Edges : 1319 Comments/Visual Assessment: There are still under explored areas. The nodes are otherwise evenly distributed with more nodes in general and more nodes and edges than the default appearing between the obstacles. These results were unexpected and I was asked to repeat the experiments and to make some modifications in the experiments to try and determine how additional nodes were being generated.
On Thursday and Friday we began to develop our websites. We also read the paper, 'Probabilistic Roadmaps for Path Planning in High-Dimensional Configuration Spaces (L. Kavraki, P. Svestka, J.-C. Latombe, M. Overmars, 1996)' and discussed it at a meeting on Thursday. At this meeting, we also discussed how to use the editor Vi and were asked to read the Project Proposal for discussion next Thursday.
Next Week
For week 2, I will finish my initial website and submit the URL to DREU and will also read the Project Proposal prior to the group meeting on Thurday. I will continue the experiments on folding and will repeat the summary experiments to verify that the changes as noted (and not some residual changes) were responsible for the node increases and to attempt to determine what parameter changes caused the node increases. At the end of the week, I will update the website and submit a weekly report on the work done.