Should the 2022 cars be moved to 2023?

[PlatinumZealot]

Machine learning is used to a limited extent in F1. Amazon is actually leading the way with the AWS as much as fans like to laugh at them. But its a good exposure to bring teams into taking advantage of machine learning / AI. Machine learning and AI requires huge computing. It's application in F1 in think should be limited, because if it is allowed we would too much convergence in solutions. F1 would be almost a spec series if everybody knows the optimal solutions to things.

I would say Machine learning is used to an enormous extent in F1, because that term includes almost everything. As a term, It's also imprecise to the degree of being worthless. I mean any laptime simulation or error detection is going to be able to be called ML. You could push it and say monte-carlo is ML probably.

For a 10000 foot view here is a map of flavours of ML problems/solutions at least

https://docs.microsoft.com/en-us/azure/ ... heat-sheet

I would argue DL and the convolutional approaches should get their own sheet, and another for optimzation strategies per DL arch, but thats not needed now.

Machine learning is not used as much as we think it is. Google it. There is a recent interview on that.

[nzjrs]

Machine learning is not used as much as we think it is. Google it. There is a recent interview on that.

Thanks PZ my Google. I Tried. Could not find anything. Would depend on the definitions..... Which was what I'm trying to convey... Maybe they defined the terms in the interview? I'll see I guess.

Anyway if the interview was recent I guess they mean ML aka DL which I agree with - DL is oversold.

I hear people describe simulated annealing as AI. It's wild out there PZ.

[PlatinumZealot]

There is an F1 interview on it I saw but I cant find it. It could be Horner or Zoe chilton i dont remember.

It's not used as much as we think it is.

Here is a reference to the Horner one.

https://www.motorsport.com/f1/news/red- ... p/6271315/

[DChemTech]

Just going to copypasta the PM I sent JaF here since there's interest.

How does that tie into the CFD mesh?
Because the mesh nodes are what are culcated.

And as rays are straight lines how does that work say around a curve? Like the curves of the car?

Hard to see how that works for F1 CFD.

Two different techniques. Im not a programmer but I was a user of ray-tracing and CFD. So this is interesting but you have to explain this one.

It is point cloud, so no mesh. The conjecture was to use the rays to "erode" the point cloud and leave shapes behind that generate the desired flow-fields around the object.

The "bounce behavior" can be turbulent (curved) with added ray collision detection. The horsepower to calculate was the only issue. The bi-directional nature of the ray caster allowed you to pick where you want any ray to start and end. That was what gave this generative modeling concept power, as you could use the BiRT to bundle the rays by density, and generate a model to run in an actual CFD software to correlate.

He/we never explored much past that conversation, so I'm sure there are more devils in the details. It just seemed reasonable to speculate that someone has done this by now with the huge computing power available in a multi-GPU desktops and super-computers.

But the challenge with CFD as opposed to ray tracing is that fluid packages are continuously mutually interacting, while photons are not. It's not just bouncing off the surface - it's the continuous bouncing in the air itself that makes the problem much harder. At least, that would be my first impression. Please do correct me if you think otherwise.

Anyway, accounting for those fluid-fluid collisions in a rational way (not in a completely mesh-free environment, which is workable for rarified gases but seems a bit outlandish for dense ones), you effectively get to Lattice-Boltzmann simulations, where large numbers of mesoscale fluid particles are streamed between grid-nodes, and collision models account for the rest. And lo and behold, these type of simulations do work very well on GPUs, and are now quite competitive with finite volume (or substantially better) in several applications. But that's not really new and earth-shattering, it's a continuous development. I'm not sure how these simulations are leveraged by F1 teams though. Many seem to be partnering up with Siemens or ANSYS for the time being, and while ANSYS is working on introducing LB in Fluent, it's certainly not mature yet. And with LB being intrinsically dynamic, steady-state FV may still win on overall computation time, for the time being.

[Ethan Connor]

I think yes they have to move to 2023 as for the first time they change the rule.

[mzso]

It should be moved to 2017 and even then it would be too late.