Which physics simulations require several days of computing time?

A guy in a forum said that a physics simulation can keep a Threadripper running at full capacity for several days.

What kind of simulations are/could these be? Even FEM can be done in a few minutes with a CPU like this.

(2 votes)
Loading...

Similar Posts

Subscribe
Notify of
3 Answers
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Picus48
7 months ago

The definitive answer to the meaning of life. "42" has already been suggested.

But seriously. The simulation of nuclear fusion in plasmas, as pursued in fusion reactors (e.g., tokamaks), requires the calculation of extremely complex processes. This involves the behavior of ionized gases (plasma) under extreme conditions, the interaction of electromagnetic fields, and the turbulence in the plasma. Cosmological simulations model the large-scale structure of the universe, including the formation of galaxies, stars, and black holes. Such simulations cover vast periods of time, from the Big Bang to the present, and take dark matter and dark energy into account. Our computers are far from adequate for this. Infinity—if it exists—will probably not be computable with even the most powerful quantum computers.

Kelec
7 months ago

It depends on the size of the problem. Try calculating the sun's magnetic field using a FEM.

Or try to calculate the response of the magnetic field of an array to billions of variations.

Wave propagation in built-up areas can also be very challenging if you want to calculate thousands of buildings and reflective surfaces with different reflectivity levels.

This can easily take several days.

Simulating the formation of galaxies or solar systems also takes an extremely long time, because relatively small time steps are required for precise simulations. This generally applies to all chaotic systems.

The computational effort often increases exponentially as the problem size increases.

StormstarZeus
7 months ago

Weather simulations still can't judge correctly 😂👀.

There's even a specific term for chaotic systems. It's pretty interesting.

A characteristic sign of chaotic behavior of systems is that their temporal progression can only be described by a continuum of periodic motions. The equations of motion are purely deterministic and without random elements, but they are nonlinear. Chaos requires only a few degrees of freedom.

Here is an article:

https://www.scinexx.de/news/technik/ki-oder-supercomputer-wer-macht-bessere-wettervorhersagen/#:~:text=macht%20bessere%20Forecasts%3F-,K%C3%BCnstliche%20Intelligenz%20gewerbe%2010%2DTage s%2DWeather forecasts%20also%20without%20physical%20knowledge,2023%2C%20Reading time%3A%204%20Min.&text=Neural%20Network%20instead of%20Supercomputer%3A%20One, one%20physical%20equation%20to%20know.