Thank you very much for the introduction and for the invitation.

So I will try to tell you a little bit how we use machine learning and optimization in

quantum chemistry.

Right from the start, I should point out I'm a quantum chemist, so my expertise is quantum

chemistry, certainly not machine learning and optimization.

We are using those as tools and the talk will be to show a little bit how these tools can

help you in an applied field here in quantum chemistry.

Let's start very basic.

What is a theoretical chemist or a quantum chemist doing?

Well, he's trying to investigate molecules or materials at the atomistic level, not by

experiment, rather by using basic physical theories, classical mechanics and quantum

mechanics.

I mean, here you have an example of a somewhat unusual molecule.

It has these five, six and seven rings and you can ask why does it form, what properties

it has and we usually do this in cooperation with experimental partners and then we provide,

for example, electronic properties of such materials or the other example is it's quite

fashionable now today, our two-dimensional materials.

What you see here is a sheet of an iron bromide, well, one dimension, I mean, it's just a few

atoms thick on a gold surface and if you look with an experiment, with a scanning tunneling

microscopy, you see these holes.

We can simulate this and then we can ask question why do we have such holes and they are interesting

because they have special electronic properties and you might use such materials for future

molecular electronics.

So these are typical questions we are asking.

Often you want to know also about the dynamics.

How do the atoms move or how stable is this?

And this is more what we are going to look here, but we are doing this using both quantum

mechanics and classical mechanics and then, of course, we will, the expensive steps we

want to avoid by using machine learning.

But we first have to look a little bit, what are we doing in classical mechanics?

So the classical example of classical mechanics is the movement of planets or generally you

have some particles in some entities here, it's planets and you know how this is done,

maybe from school.

You propagate Newton's equations, so first, what do we want to know from theory?

In the case of classical mechanics, we want to know the position and the momentum, this

mass times velocity of each particle or entity, here it's planets, at every time given a certain

starting point.

And we do this with Newton's law, you all know this from school, so mass times acceleration

is the force.

So this, you can also write it in the way, the derivative of the momentum, momentum is

mass times velocity, velocity is the first derivative of the position with respect to

the time, acceleration is the next one.

The second derivative and the force depends on the position of the quantities you are

interested, here the planets.

In this case, this is very easy, this is the gravitational law, it's always two particles,

the distance or the mass, the product of the mass is divided by the distance squared,

it's a force.

In general, we are looking at what's called conservative systems, so we consider the potential

energy of our system and, somehow, okay, it's a bit weak.

So we have to take the negative of the derivative of the potential energy with respect to the