Value of simulation

[MusashiStyle]

Computer Simulation allows us to set up objects with various parameters that model some real physical system.
I'm curious how valuable this is as a research tool as compared to running an actual experiment. Obviously, certain fields like astrophysics this is very useful since certiain experiments are not possible.


I am curious about what people think the value of computer simulation is in term of testing theoretical models in physics?

To be more specific,

Suppose one develops a complex and detailed theory of black hole formation ( i.e. a heavy star --> passes critical radius). I know this theory is actually already worked out, but what I'm wondering if actual numerical computation of theories adds value to this theories and how it does so?

( I am interested in possibly working with simulation and definitely interested in theory)

[Borodog]

Yes; numerical simulation is critically important. As you've said, in some situations (astrophysics, nuclear weapons design) you simply can't conduct the experiments. In a wide variety of applications where you could conceivably conduct experiments, it is often far, far cheaper to conduct all or most of your experimentation via numerical simulation. This is how virtually all high end design work (automatoive, aerospace, nautical, etc) is done these days.

Take for example core collapse supernovae. We've had a general theory worked out for decades, with just one problem; no one can actually get one to explode in a computer simulation. My Ph.D. advisor is currently working with a team (and has been for 5+ years) to get one to blow up in the computer. His work has already shown that the rotation of the product neutron star is not due to conservation of angular momentum from a rotating progentitor; but rather from the fact that the blast wave creates its own rotation and spins up the neutron star in the opposite direction. The Navier-Stokes equations are far too complex to have discovered this analytically.

[MusashiStyle]

[ QUOTE ]
Yes; numerical simulation is critically important. As you've said, in some situations (astrophysics, nuclear weapons design) you simply can't conduct the experiments. In a wide variety of applications where you could conceivably conduct experiments, it is often far, far cheaper to conduct all or most of your experimentation via numerical simulation. This is how virtually all high end design work (automatoive, aerospace, nautical, etc) is done these days.

Take for example core collapse supernovae. We've had a general theory worked out for decades, with just one problem; no one can actually get one to explode in a computer simulation. My Ph.D. advisor is currently working with a team (and has been for 5+ years) to get one to blow up in the computer. His work has already shown that the rotation of the product neutron star is not due to conservation of angular momentum from a rotating progentitor; but rather from the fact that the blast wave creates its own rotation and spins up the neutron star in the opposite direction. The Navier-Stokes equations are far too complex to have discovered this analytically.

[/ QUOTE ]

fascinating, I'm curious as to what sort of coding/platforms you (or your advisor) work with.

My plan was to work independently with Java to start doing some simple models of Newtonian Systems and work my way up.
Hopefully I'll learn to program with Mathematica as it seems that software is really powerful.

[Borodog]

[ QUOTE ]
[ QUOTE ]
Yes; numerical simulation is critically important. As you've said, in some situations (astrophysics, nuclear weapons design) you simply can't conduct the experiments. In a wide variety of applications where you could conceivably conduct experiments, it is often far, far cheaper to conduct all or most of your experimentation via numerical simulation. This is how virtually all high end design work (automatoive, aerospace, nautical, etc) is done these days.

Take for example core collapse supernovae. We've had a general theory worked out for decades, with just one problem; no one can actually get one to explode in a computer simulation. My Ph.D. advisor is currently working with a team (and has been for 5+ years) to get one to blow up in the computer. His work has already shown that the rotation of the product neutron star is not due to conservation of angular momentum from a rotating progentitor; but rather from the fact that the blast wave creates its own rotation and spins up the neutron star in the opposite direction. The Navier-Stokes equations are far too complex to have discovered this analytically.

[/ QUOTE ]

fascinating, I'm curious as to what sort of coding/platforms you (or your advisor) work with.

My plan was to work independently with Java to start doing some simple models of Newtonian Systems and work my way up.
Hopefully I'll learn to program with Mathematica as it seems that software is really powerful.

[/ QUOTE ]

I rock FORTRAN, personally.

[garion888]

90...or 77?

[CrayZee]

Java is probably okay for "simpler simulations" but terrible for anything with tons of variables...unless they got that JVM bytecode super fast or perhaps Java with a different engine. Probably better languages to use anyway for this sort of thing, I dunno.

Yeah, sims are pretty useful...and interesting.