Where do physics and programming overlap?

[jamesbrewer]

I'm taking a calculus-based introduction to physics this semester and I'm finding it fascinating so far. The idea that there is a reason and a method for everything really resonates with me and that has lead to me really enjoying this course.

I'm also really into programming. I've spent quite a bit of time over the last year programming with Python in different domains and I want to get into physics programming now, but I'm not exactly sure what 'physics programming' is.

What are some ways the I can integrate these two disciplines? Where do the domains of physics and programming overlap? Game physics sounds like it would be interesting, but as I understand it, physics programming jobs at game companies are becoming more and more rare as dedicated physics processing units are starting to gain a following. Can anyone (reliably) confirm or deny this?

 

[rcgldr]

For games that simulate aircraft or land based vehicles, there can be a lot of physics. There's some type of mathematical model used to determine forces for a given set of conditions (orientation, position, velocity, control inputs, ...) , which are then translated into acceleration. Then some form of numerical integration is used to predict new velocity, position, ... for the calculated acceleration. A common method is Runge Kutta 4:

wiki_runge_kutta_4.htm

 

[AVReidy]

Games, simulators, and calculators. Maybe they also overlap in advanced vehicle systems where airplanes, for example, use a computer and gyroscopes to keep them flying level. I would assume a programmer chooses the degrees the control surfaces would have to move to sustain level flight.

 

[jamesbrewer]

I was doing a problem set earlier today, using WolframAlpha to check my work when I realized that WolframAlpha is a pretty cool site and definitely something I'd be interested in working on. The company seems to have internships, although the link just sends you to the regular jobs page.

What I'd be really interested in working on is something similar to WolframAlpha, or WolframAlpha itself. What else is out there that is similar?

 

[DavidSnider]

Physics processing units do not replace the demand for people who understand how to apply physics to computer games. They just make doing parallel processing of physics calculations easier.

The main problem is that computer game physics just isn't that complicated. Very simple high-school level physics is more than enough for most games to be fun and a graphics programmer could probably come up with a good-enough physics engine or easily understand one of the many that have already been built.

There is also a lot of diminishing returns for how realistic the physics is. It's almost instantly obvious when doing graphics rendering that something is 'off', but with games they often have to tweak the physics to be slightly off because the real world just isn't that fun to play in.

 

[rcgldr]

In the case of simulation type racing games or radio control aircraft simulators, only a small portion of a development team is directly involved with the hard core physics. For the racing games, one difficult aspect is trying to model how tires behave when at the limits of grip, which is compounded by trying to model the suspension as well, and as mentioned, only a tiny percentage of gaming programmers ever get involved with this stuff.

It's a similar situation for programmers that try to model aircraft, including the people that develop real aircraft and/or models. Only a small percentage of physicists are involved in airfoil design. High end contest type radio control gliders are made in large enough quantities that significant research has been done on optimizing airfoils and the overall design of such models, but it's only a handfull of people doing this. Most full scale and model aircraft are made using existing air foil designs (with some modifications) as opposed to development of new airfoils.

 

[Negatron]

The main problem is that computer game physics just isn't that complicated. Very simple high-school level physics is more than enough for most games to be fun

I find that to be misleading. Nothing from vehicle physics to collision detection to parallel programming of the implementation is simple. You can come up with a few garbagety side-scrollers as examples to the contrary but 3D game physics for any respectable game would have top engineers crying tears of agony, forget high school students.