Many thanks, Bernd, for the introduction and for having me here.

Since Christian Plessel is also working at PCSquare, has a longer talk thereafter and

will talk about the more PC, HPC-oriented topics.

We've worked together for quite a while.

I've slightly changed the topic and will focus more on a methodological stealth project,

which is ongoing for many years now in my group.

I basically wanted to start with some sort of what we are doing.

I was a little bit caught off guard by Peter Salfrank.

Apparently, in the meantime, he's doing very similar things.

But the cornerstones of the research I'm doing is not the reductionist approach, but really

to embrace the complexity of nature and to combine the dynamics of the nuclei, classical

or quantum mechanical by means of path integral simulation, but still assuming the Bonoppenheimer

approximation on preferably long length and time scales.

With the quantum mechanical nature of the electrons to model experimental reality as

possible.

So we want to do that and really drive that by methodological development to preferably

large length and time scales.

And as you have pointed out in the thing that typically requires the electronic structure

calculation should be taken about these 10 second barrier, which has been just referred

to.

So most of the research we have been doing is conducting using the CB2K program package,

this is a mixed Gaussian and plane wave basis set.

And first of all, allows to compute all your favorite static properties using any regular

quantum chemistry or solid state physics code.

But doing that with arbitrary boundary conditions or from the gas phase up to fully 3D periodic

boundary conditions.

The main appeal, however, is that we can combine a lot of different sampling techniques.

So Monte Carlo molecular dynamics, rare events sampling, Aaron Fisk dynamics and a couple

of others with all sorts of energy and force methods.

And they can be going up from fully classical MD simulations, over semi empirical techniques,

these new network potentials are now very new to pseudo potential based and very efficient

in the DFT calculation or electron calculations.

And very recently also a large variety of different post-hard refoc techniques, which

shares a lot of similarities in terms of what Christian talked about.

Just before we try to provide as much as possible forces.

And we want to do this also as much as possible using periodic boundary conditions.

And there's a very nice review article summarizing all the developments of the whole community,

which just came out last year.

So it's really a multifunctional, very general purpose tool, the 32K program package.

And today I will mainly focusing about very recent developments.

I've just listed here some of them.

Some of them are really very efficient post-hard refoc techniques.

The other one, and this is not so well known, there's a full plane wave based DFT code with

this full GPU support now integrated.

Linear scaling techniques, this is what I'm going to talk about today.

And approximate ab initium D simulations, mainly using based on the approximate computing

paradigm.

And if time permits, I will talk a little bit about FPGA support.

It's an open source code and it's really available at this specific URL for you to try.

Right.