Thank you very much for the invitation.

Thank you very much for the organization.

I can say for personal reasons that the town is great.

Anyhow, it's a town.

I will not improvise on my computer with regard to Google Maps now,

but Rostock maybe has double the size,

but I would say the inner city,

what is more important for us,

is I think from the size almost identical.

It's located at East Sea,

so more or less at the place where you arrive,

let's say with the burst connections.

It's in Mecklenburg-Vorpommern,

and that's a place where you have not so many inhabitants.

There are also some infrastructure,

at least fast trains, etc. is not as well-built.

That means if you take the train from here seven hours, if it works.

If it doesn't then it's nine, but of course everyone knows that.

Thank you very much.

I'm talking about optimal control of non-local conservation laws

and the so-called singular limit.

It's also not a machine learning talk,

but at least it has something which is common in machine learning.

We have optimization.

Okay, so that's great.

Okay.

What is non-local and why do we do this?

This is a typical slide I always show,

so I will not explain the slide.

The people who know this guy can find this funny.

The rest just says,

well, this is an external question now raised,

and I will try to answer the question shortly.

Okay, so Paula already mentioned that in her talk,

so that there are different types of traffic flow models,

and a rather recent development in traffic flow models maybe is particular,

I think also due to her,

that you don't take local models anymore,

but you take some kind of non-locality,

and the non-locality is usually an integral operator.

Now you can imagine you can have very, very different integral operators.

We will tackle very specific integral operators,

basically just L1 integrals.

What you see here is a conservation law in the value Q.

The quantity here is conserved,

and you see you have a velocity function V.

It's a given velocity function,

and you have this quantity W of Q and a gamma,

and the gamma is more or less this non-local kernel.

So basically that is important in the integration,