Other than about mathematics and physics, I am very passionate about the tools I use to carry on my research.
My workflow roughly goes as follows: I write my Monte Carlo simulation code in C++. C++ is an horrible language for academic research because it’s very complex and there are too many ways to shoot yourself in the foot, e.g. by stepping on undefined behavior. Despite this, I spent quite a long time learning C++ and, along with Python, is one of the languages I am most productive in. I have recently fall in love with Rust but I find Scala quite attractive.
My simulations usualy end up writing down a big HDF5 ready for data analysis.
I do all the analysis on IPython notebooks using a combination of Python libraries (numpy, matplotlib, PyTables, Pandas, etc) and code I have written of my own that automatise the most repetitive tasks.
When an open-source tool that I use is missing an important feature, I do my best to contribute to its development. This was the case with PyTables, to which I have contributed the code to support extended precision datasets on both 32 and 64 bits arches (where extended precision is 12 and 16 bytes respectively). My contribution to the project is now lagging behind and I do feel pretty bad about it.