Having dabbled with everything from Excel to Matlab I settled on using Python for the main part of the data analysis that I do. This includes everything from curve fitting to processing some tens of GB of data from hybrid pixel detectors. With a large user base lots of open source tools it is the ideal tool for students. While Python does most of the things I need, I still use some functions from ROOT sunch as filling histograms and fitting curves. Using the PyROOT bindings with some encapsulations of my own I can seamlessly integrate ROOT in my analysis work flow.
Latex is the only real option for me here. While changes and comments in MS Word can be usefull its just too annoying for me to handle references, table of contents, formatting etc. Go with Latex from the begining, even though it might be some initial threshold to get started. Pair it up with bibtex and some tool like Mendeley or Zootero to manage your library of references. For the actual writing I use the Texworks editor which gives easy compilation, syntax highlight and spellchecking without adding a lot of fluff…
Posters is going to be an integral part of your work. Please make them look nice, too many researchers don’t. Before you start no MS Power Point is not the tool to use. Adobe Illustrator is of course nice and will make collaborating with your cool designer friends easier… but quite pricey! My choice is Scribus, featuring templates and precise placement of graphics it will make your life easier.
Here I stick with MS Power Point… unfortunately. I used the Latex beamer package before, and it is still my first choice for math heavy presentations, but for normal presentations it’s just too time consuming. Also borrowing slides from colleagues is much easier using the same tool.
GIT is an excellent choice for version control and collaborative coding. I use it on almost all my code and writing. Get familiar with it and use it from beginning! Using a platform like bitbucket you can also access your code form anywhere and get a good backup.