Finishing my undergrad Bsc in Physics this tear, and am thinking

>>8640573
You're probably going to want to learn C and Fortran. Python alone just won't cut it for HPC. As for books, I can't think of any (although I'll admit I'm not that familiar with the subject).

>background reading
Ask your supervisor

>>8640573
>Should I work through a Computational physics book or a computer science book?

Why not both? In my (limited) experience, Computational Physics does not need super advanced programming skills, but it does require some knowledge about numerical mathematics.

>>8640624
I think C and C++ is whats used mainly for the course but I heard Python is really useful in computational Physics too. If I keep learning Python will it help understand C/C++ when I start the course?

Also can anyone tell me what kind of job this could lead to, as in what a work day would be like? Is it a good career?

>>8640630
Yes. Python is a very good language to learn for beginners and will teach you the fundamentals of programing.

>>8640630
i took a course in university that taught C++ and used that in all later courses that required any sort of programming since i was familiar with it. recently, i started learning python on my own just for the fuck of it. python is significantly easier to use, and much more forgiving in terms of getting anything done, but it's also far slower. i didn't think this would be a problem until i got to the point where i have scripts with a few hundred lines of code and a few million iterations in loops, but even having a nested loop working over a list with a few hundred thousand elements takes a while to run while in C++ it would be nearly instant.

so what i'm getting at is that python is really great to use. you can make loops where you iterate over characters in a string or elements in a list effortlessly, but performance wise C++ is far better.

>>8640573
>I want to develop my computational physics skills this year before I start the masters so is there a recomended textbook to work through?

For the "computational" portion, learning C++, Data Structures and Algorithms, Computer Organization, Systems Programming and Parallel Programming would be helpful and useful.

See the wikia for books:
http://4chan-science.wikia.com/wiki/Computer_Science_and_Engineering

>>8640573
Knowing Python to a certain degree will most likely be a 'must' in the future. Even if the raw core of the algorithm is written in Fortran or C++, Python is used to interface various modules or to parse hueg inputfiles that jumpstart the simulations.

What matters really in scientific computation?

Know your way around Linux shell scripting (bash or equivalent), know how Linux works (where are libraries, how do I install new ones, how do I find them etc.) have an idea about job schedulers (e.g. LSF or Torque), and then be somewhat proficient in the main languages (which at the moment are: Fortran, C/C++ with Blitz, Python). If you want to go the extra mile, learn about unit tests and how to do them properly (although you will rarely have the time to do them).

Essentially you want to minimise the time between having an idea and implementing that idea into working code.

Source: physicist doing earth science, I run HPC jobs on a daily basis.