The future is now: a neural network (AI) simulates fluid dynamics faster than any other algorithm before (video with simulations + the overall idea)

The new frontier

The AI and machine learning conquer new frontiers. Lately scientists used neural network, which is a processing unit that works similar to human brain, to simulate dynamics of fluids. The neural network not only learned how those fluids behave, but was able to simulate them much faster than traditional algorithms.

---

The video showcasing this amazing new approach to physics simulation (cool animations inside):

---

The traditional, old approach

In order to simulate the shape of the fluid, the processor has to compute a lot of complicated algorithms - for example the Euler equations:

this is just a small fragment

Those equations “describe” how water, or any other fluid (including smoke) behave in given situation. This is some pretty complicated math and even very fast PCs needs a lot of time to solve this. Using modern GPU (graphics card) helps a lot, but it is effective only for small, less complex scenarios:

fluid simulation using NVIDIA CUDA technology

The neural network approach

So basically instead of solving all those complicated equations, we use neural network. The network learns what happens with the fluid in given situation and after the learning period, it can determinate the fluids behavior in any situation - not just those from learning session.

If we would like to present this change in a algorythm, the old way would look like this (please excuse huge simplification):

a =  5
b = 12
x = 1
y  = ax + b
show me the value of y

With the new way, we create a neural network and teach it how the fluid beheaves on a large number of examples:

when x = 1 -> y = 21
when x = 2  -> y = 431

Than the network tries to find the pattern (the parameters) to match all the equations. After the learning is done we can start to ask questions;)

show me y when x = 3.

(please keep in mind that this is a huge simplification in order to show the overall idea)

---

Wow!

Now, the amazing part is that performing computations on (trained) neural network is much faster than computing traditional algorithms. Where's the old method needs at least several minutes, the AI can give result in milliseconds !

This is where the similarity of neural network to human brain comes in - if you drop a glass of water, you don't take pencil and paper and start to solve complicated equations. You just know what's gonna happen - just like artificial neural network. And this is just...wow:)