Why Choose Computational Physics?

[Arij]

Hello,

I've been wondering lately why would one change study path from theoretical or applied physics to computational physics or become a simulation expert?
I am a physics graduate myself and based on the job market where I live I might need to pursue material sciences or computational physics. I am trying to figure out which path I prefer, but I don't feel like I understand what does a computational physicist do? Can't anyone use the computational/simulation programs with enough training? how would a background if physics be especially helpful for modeling and prediction?

 

[Vanadium 50]

Presumably there is someone at your university doing this, otherwise it wouldn't be offered as an option. What did they say?

 

[Arij]

Presumably there is someone at your university doing this, otherwise it wouldn't be offered as an option. What did they say?

I have graduated and left the united states. I have not questioned this before, that's why I'm asking the community at PF.

 

[phyzguy]

Someone with a degree in computational physics would typically not just be using the simulation programs, but writing and/or modifying them. This requires a knowledge of the physics involved as well as a knowledge of how to do the required calculations in a computationally efficient manner.

 

[Crass_Oscillator]

Hello,

I've been wondering lately why would one change study path from theoretical or applied physics to computational physics or become a simulation expert?
I am a physics graduate myself and based on the job market where I live I might need to pursue material sciences or computational physics. I am trying to figure out which path I prefer, but I don't feel like I understand what does a computational physicist do? Can't anyone use the computational/simulation programs with enough training? how would a background if physics be especially helpful for modeling and prediction?

Computational physicists and theorists are extremely closely related; the distinction between the two is often artificial, although real.

I have worked in computational physics for 6 years, working in molecular biology, materials science, and device physics. In some cases, much of what I did was just computer science, high performance computing, and data analysis. Statistical modeling was involved in the latter. In other cases, I have done theoretical modeling in tandem with simulations, writing codes and developing effective models from high dimensional simulation data.

One project hybridized the two. An ab initio model of a protein would be sampled on a distributed system, as the configuration space was so ludicrously sophisticated that months of simulation would be required to compute useful statistics. A theoretical model in collective variables was used to strategically guide the sampling of the high dimensional model across many cores, providing a performance boost in a "Bayesian" fashion. The sampling of the ab initio model refined the theoretical model in a variational fashion. I was a theorist (statistical mechanics/non-equilibrium statistical physics), software designer, simulation programmer, high performance computing specialist, and data scientist, all in one.

The field is honestly pretty diverse. A PhD is a necessity for much of the work, as one must thoroughly understand all of the theory underlying the codes, even canned, plug and play codes. Some work can be done by engineers with a masters, but in device physics for instance they pretty much exclusively hire PhD's.

 

[ZapperZ]

Hello,

I've been wondering lately why would one change study path from theoretical or applied physics to computational physics or become a simulation expert?
I am a physics graduate myself and based on the job market where I live I might need to pursue material sciences or computational physics. I am trying to figure out which path I prefer, but I don't feel like I understand what does a computational physicist do? Can't anyone use the computational/simulation programs with enough training? how would a background if physics be especially helpful for modeling and prediction?

In addition to the responses that you have received, I strongly suggest that you look the topics being published in computational physics. Starting at arXiv will be a good start:

https://arxiv.org/list/physics.comp-ph/recent

Edit: In addition to that, please have a look at the Computational Physics Division page within the APS. You should have plenty of information on what the people in this area do.

https://www.aps.org/units/dcomp/index.cfm

Zz.

 

[alantheastronomer]

Numerical modeling in physics is the ONLY way to solve non-linear partial differential equations and obtain solutions that cannot be solved analytically. Your knowledge of physics is essential in guiding the programming.