I will be studying simple SUSY models using the razor variables. In particular I will be trying to distinguish between these models and the common ttbar background process, but using a multi-b-jet analysis. Some of the models of interest include RPV-gluino-gluino, T2bw, T2tt dna T6ttHH.

My research will focus on exploring how machine learning, specifically reinforcement learning, can be used to better handle matching of the workflows of the LHC experiments with LHC Grid resources. I will be using the RLPypackage (written in Python) developed at MIT to facilitate the investigation into reinforcement learning by implementing my own simulations of computer systems. The aim is to understand which learning systems better fit with a distributed system as the LHC Grid as well as to evaluate the speed with which such systems can be learned.

Python for Reinforcement Machine Learning

Machine learning has found wide applications in systems where decisions have to be made based on data. This includes control systems, computer systems, data analysis, etc. One of the tools of choice is often the python programming language, due to the very large amount of existing python libraries as well as to the general popularity of the language itself. During my FSRI research I plan to write python modules and programs that implement machine learning scenarios in the context of simple computer systems.