Computational Physics degree: what do you guys think?

[Couperin]

Main Question or Discussion Point

On the surface, Computational Physics appeals to me a lot. I'm very interested in physics, computer programming, creating simulations of mechanical systems, and games programming.

However, I don't really know what Computational Physics is.

What would I learn?
How attractive would I be to employers with a degree in Computational Physics?
What fields of work could I get into?
Could I get into the videogames industry?
Are there any computational physicists on this forum?

 

[neurocomp2003]

yes you can get into the gaming industry...its a matter how what you have to show for them when you graduate. So i suggest trying to build as much demos.

There are two sides to Computational Science: Numerical Modelling and Computational Geometry/Spatial/3D rendering.

To get into the gaming industry you would require both. But perhaps the latter(especially comptuational geometry) is more important.

See if you can get your hands on Chris Heckers Works or a text by Christof Erricsson(sony entertianment,my spelling is bad) called collisions detection. I'm reading the text right now and am enjoying it (also for a beginner J.O'Rourke's Computational Geometry book but its less useful interms of physics stuff)

As for the physics side...depends on which physics field you go into but its more numerical crunching then 3D rendering. Just visit any profs page who does numerical work and they myal have some simple demos. Most Astro Sims are fun to watch. There's also a site from oregon that does education demos...sorry i can't remember it, but if you search google you might find it.

 

[las3rjock]

Let me start off by stating that there is very little overlap between video-game physics and computational physics.

Video-game physics is rarely more complicated than basic first-semester college physics, generally applying Newton's laws of motion to calculate the motion of game objects. These are rather simple calculations which can be done essentially in real time. The coding can be performed by a moderately-skilled programmer with a rudimentary understanding of basic physics.

Serious computational physics includes but is not limited to the following:

• N-body simulations (vs. the 2-body simulations used in video games)
• Molecular dynamics simulations of molecules or chemical reactions
• Density functional theory to determine the electronic structure and properties of materials
• Scattering simulations for acoustics, electromagnetics, optics, or particle physics
• Computational fluid dynamics

These are rather complicated calculations that can only performed by making a number of simplifying assumptions or approximations and by employing sophisticated algorithms, and even so these simulations may hours, days, or even weeks of computer time to simulate a process that elapses in a fraction of a second in physical experiments. The coding for these problems generally requires a fairly sophisticated mathematical background (at least through partial differential equations and numerical linear algebra) and a sophisticated physics background as well (in order to know what sort of approximations can be made in solving the problem at hand without overly compromising the simulation accuracy). There is definitely a job market for computational physicists, but it depends on one's specific expertise.

 

[chadat]

Urgent: frm elect eng to physics phd: have a chance?

well guys, i just found this forum and am hoping you can help me here.

i am intl student with bs in elect eng. but i realized when i entered that program that it wasnt for me n that shd have stuck with my math/physics. i ended up with some sloppy gpa, probably just above 3.

but now i want to get back to physics n have already applied to wesleyan university phd physics( yes, they have a phd in physics, pure math, comp sci, etc), (can any of you make suggestions here) and am considering applying to some math/applied math programs too. i am interested in computional physics, chaos theory, etc, and have some experience in programming and high performance computing.

my gre general are: 770q, 680v, 4.0aw. how do u guys evaluate me? i stand a chance w/out the gre physics subject test vis a vis my low gpa?

 

[cscott]

Serious computational physics includes but is not limited to the following:

• N-body simulations (vs. the 2-body simulations used in video games)
• Molecular dynamics simulations of molecules or chemical reactions
• Density functional theory to determine the electronic structure and properties of materials
• Scattering simulations for acoustics, electromagnetics, optics, or particle physics
• Computational fluid dynamics

Is there any of the "Computational Geometry/Spatial/3D rendering." mentioned by neurocomp in the list you made above?

What aspects of computational physics are used most in the industry at this moment?

 

[neurocomp2003]

CScott: for industry probably MD and CFD. (..abbrevs, see the list above) Nbody seems to be more academic. As for the CompGeom/3D stuff most people have other people implement the rendering portion(in industry/academia known as visualization) or use some already available package.

I would have to agree & disagree with las3rjock in terms of the difference between COmputational Physics & Video Game Pseudo Physics, especially in helping a student acheieve foundations for a career in one or the other(eg. providing a learning curve)

In terms of physics-based theory, he is right in saying that the theory is far apart ...but gaming physics of 3D FPS & vehicle simulators has gone past 1st year physics and requires alot of higher level Classical/Analytical Mechanics (esp Collision-based RigidBody Dynamics & Particle effects). Some good resources are David Wu's notes from Psuedo Interactive, Chris Hecker (...) and Christoff Ericssons book.

In terms of computing skills, I would disagree that CompPhys & GamePhys are far apart (though one could argue that with all the opensource and cheap licenses one wouldn't need to learn to much). There is much overlap in the programming skills required especially for NBody/MD/CFD stuff.

an undergraduate geared towards either field should have some implementation knowledge of Linear Algebra, Classical Mechanics, Numerical Analysis(eg LinearSolvers, DE solvers), Robustness & Optimization & Compiler stuff, 3D Spatial Geometry, SceneGraph Management(Tree Codes), and Networking or HPC.

Rendering is an unnecessary but beneficial step in both. My supervisor mentioned that its just one step away from his research in astrophys but he gets other people to do it or uses opensource.

 

[creepypasta13]

Serious computational physics includes but is not limited to the following:

• N-body simulations (vs. the 2-body simulations used in video games)
• Molecular dynamics simulations of molecules or chemical reactions
• Density functional theory to determine the electronic structure and properties of materials
• Scattering simulations for acoustics, electromagnetics, optics, or particle physics
• Computational fluid dynamics

.. The coding for these problems generally requires a fairly sophisticated mathematical background (at least through partial differential equations and numerical linear algebra) and a sophisticated physics background as well..

what about in EE? From what I've seen from professor's research in EE, there doesn't seem to be much computational work in solid-state, optics, and lasers

 

[Phyisab****]

Zombie thread!

I have to disagree that there is little computational work in optics and electromagnetism.
http://en.wikipedia.org/wiki/Computational_electromagnetics
These techniques are crucial for all sorts of applications including modeling antennae/cell phones, em fields in electronic devices, stealth aircraft, EM pulse protection, any optical device etc.

 

[creepypasta13]

Zombie thread!

I have to disagree that there is little computational work in optics and electromagnetism.
http://en.wikipedia.org/wiki/Computational_electromagnetics
These techniques are crucial for all sorts of applications including modeling antennae/cell phones, em fields in electronic devices, stealth aircraft, EM pulse protection, any optical device etc.

Oh yeah, I forgot about that. But are there computational methods used in lasers? I see that DFT, molecular dynamics, and N-body simulations are used in condensed matter. Are they also used in solid-state physics in EE depts?

 

[flemmyd]

Oh yeah, I forgot about that. But are there computational methods used in lasers? I see that DFT, molecular dynamics, and N-body simulations are used in condensed matter. Are they also used in solid-state physics in EE depts?

I figure there are computational methods in various spectroscopic fields. so if you do raman, someone in your lab group might make the actual machine and you make might simulated spectra and compare the two.