Week 1: Renee and I met with our mentors and discussed possible directions for the project. Worked on a basic website design and continued reading through background information. Prepped for meeting #2 and worked on presentation.
Week 2: Renee and I met with our graduate mentor and discussed the details of the project further. We gave our powerpoint presentation during the second round on Tuesday. Presentation 1 (pdf)
Since then, I have obtained and read through the dissertation that we will be working off of. Researched various methods of optimization for non convex problems. Created a file to implement the code provided in the appendix to the dissertation.
Week 3: Renee and I met with Dr. Elsayed and Dr. Xie again. They instructed us to learn about genetic algorithms and neural networks. We formulated a new model for the optimization function, adding the element of earliness. We are in contact with Dr. Young and are hoping to obtain all of the necessary files to get her program running. We are hoping to get the genetic algorithm in Matlab working for our new optimization function.
Week 4: This week, we continued to write the function file for our new optimization formulation. We also ran the Genetic Algorithm with simple functions, to get a feel for how it works. We also worked with some scheduling heuristics, finding the optimal schedule manually with a small number of containers and stations. Finally, we read up on metaheuristics and papers that solved problems with penalties for both earliness and tardiness.
Week 5: I got my function file running in Matlab and used the GA on it. The GA returned the optimal allocation of containers to inspection stations given the risk levels that I inputted. This is the first half of a complete solution. Next, we will find the the optimal schedule for this allocation and see if we can find a better risk score when considering the allocation and the schedule at the same time. Additionally, we are about to begin work on our final presentation.
Week 6: This week, I manually completely enumerated a few examples of scheduling situations. Using these, it was easier to see, in general, which heuristics work well. I was impressed by how quickly the problem becomes large. I was able to use one of these examples in our final powerpoint. Renee and I presented our powerpoint to Drs. Elsayed & Xie on Friday, received criticism, and will be polishing it in the upcoming week.
Week 7: As final presentations approached, this week was spent mainly on preparing our powerpoint. Renee and I presented once again to Drs. Elsayed and Xie on Tuesday and for the rest of the group on Thursday. With the presentation finished, I then began work on my paper. Final Presentation (pdf)
Week 8: We have reached the final week of the REU. This week was spent working on my final write-up. I spent time deciding how to formulate our algorithm and deciding which parts of the research to continue in the fall.
