Computational Physics: A Career in the Intersection of Physics and Coding?

[Halindor]

I think I've decided on this career field (I really love physics and coding) with electrical engineering as my backup in case I change my mind. Is there anyone out there who is either on their degree track or has a career in that can give me info on their thoughts/experiences/projects you work on? Thanks!

EDIT: First post yay! =D

 

[Pythagorean]

I got my BS in physics and throughout my MS in comp neuro, I worked as a computational physicist in geophysical infrasound.

I loved it! Low stress job, chose my own hours, good pay. I did everything in MATLAB.

The first big project I had there was characterization of several manmade acoustic signals, the last project I had was designing a detection algorithm for waves coming across the array of sensors. There was a lot of geometry and signal processing involved.

Am now doing a PhD in comp neuro in another country and I often miss having the job, often wonder if my future prospects will be as bright :/ especially with all the disparaging talk around here about working in your field, heehee. Maybe I'll go for pharamaceutical while I wait to be accepted into a tenure track position (fingers crossed). Or maybe I'll just go back to infrasound :)

 

[Halindor]

That's good to hear Pythagorean! I'm hoping when I go to grad school there will be a wide range of projects to work on so I can narrow down exactly what I want to specialize in. I'll have to research those things you were mentioning.

 

[Astronuc]

How about computational physics as an EE.

http://www.comsol.com/offers/mphsim13/


Computational physics is a broad area. Each scienctific and engineering discipline has their particular version of computational physics. Often, some FEA code is applied, or it could be FD or FV. Basically one will use numerical methods/analysis to solve a coupled system of ODEs/PDEs, or ususally PDEs, possibly with some ODEs. Such problems/systems are usually time dependent, and frequently non-linear.

 

[Pythagorean]

That sounds like a good plan. It's a pretty broad field. Pretty much every field of subjects has topics that need compuational power to study: it often comes down to either analzying theoretical systems that are too complex to be modeled with pen and paper or analyzing data from experiments.

Both cases include good programming and data visualization skills, but you also have to remember that it's still physics, and starting at a chalk board or with pen and paper can save you lots of computer time (that's probably just as well for a computer scientist, I guess).

Do you know any programming languages?

(jinx to Astronuc)

 

[Astronuc]

These days, folks program in C++ or Fortran2003/2008 or later, basically object-oriented programming (OOP). There is however a substantial population of legacy codes written in Fortran77, or Fortran90/95.

 

[Pythagorean]

In my ballpark it's been interpreted languages like MATLAB and python. They're higher level and have lots of function, so they reduce code-writing time (at a cost of computational time). You can do some slick things in python that aren't possible in other standard scientific languages.

 

[ParticleGrl]

I did less computational high energy stuff, most of the people I know who did computational physics phds now work for banks, insurance companies, and software companies. Physics can be pretty hard to market, but the computational skills provide lots of back up options so the computational guys I know had the least issue finding a first job post-phd.

The projects computational people worked on ranged from simulating DNA in various environments for biophysics stuff to simulating how dislocations move about in various metal alloy structures.

 

[Astronuc]

In my ballpark it's been interpreted languages like MATLAB and python. They're higher level and have lots of function, so they reduce code-writing time (at a cost of computational time). You can do some slick things in python that aren't possible in other standard scientific languages.

Python scripts are often used to do data manipulation. In some mesh programs, e.g., Cubit, python scripts can be used to automate mesh generation.

In other work we do, we use script files to process jobs and I/O.

 

[kinkmode]

Many of my friends from grad school work as computational physicists. Mostly at the weapons labs (LLNL, LANL, or Sandia). One is at NRL. Another is at General Atomics. I would imagine most of them are using Fortran (the guy at NRL is, and no, it's not a legacy code), while some of them probably use C.

This is in plasma physics/fusion research, so most of them are either working on inertial confinement, weapons, or tokamak related stuff. A few do straight up astrophysical modeling. They are all obviously career researchers and are not in industry.

If it is something you are interested in, there's a lot of opportunities in plasma physics/fusion for computational physicists, and probably quite a few, but lower paying, positions in astrophysics. There's a lot of crossover between plasma physics and astrophysics if you work on the right stuff.

 

[lasymphonie]

How much more difficult is it to obtain a position as a computational physicist than as an experimentalist (if at all)?

 

[kinkmode]

I don't think it is much more difficult.