Peter Wirnsberger


I am a Research Scientist at DeepMind with a strong background in physics, computational chemistry and high performance computing.

My research interests include atomistic simulations, machine learning and statistical modelling. Here you can find my CV, my latest research updates and a collection of links to my other public profiles. For a complete summary of my publications, please visit my Google scholar page.

Budapest, © Wolfgang Brandst├Ątter

Latest news

Microscopic analysis of thermo-orientation in systems of off-centre Lennard-Jones particles

Take a look at my latest paper on the arXiv.

I received my PhD from the University of Cambridge!

Here is a link to my PhD Thesis in case you are interested.

I started my new position at DeepMind in London!

I am very excited about joining DeepMind as a Research Scientist.

I won a Microsoft Azure Research Award!

Microsoft will support my research for an entire year by providing me with access to their cloud computing service Azure.

We won the Microsoft Hackathon!

Our team is one of the global winners of this year's Microsoft Hackathon with more than 18,000 participants worldwide.

Machine Learning internship at Microsoft

I am very excited about starting my 3-month research internship at Microsoft Research Cambridge today.

Numerical Evidence for Thermally Induced Monopoles

My latest paper is finally out! The animation below illustrates the basic idea behind 'thermally induced monopoles'.

arXiv | PNAS

See the Pen HEATED PARTICLE by noname (@al-ro) on CodePen.


Non-equilibrium simulations of thermally induced electric fields in water.

In this paper, we use atomistic simulations to study the 'thermo-polarisation effect' in water with particular focus on the treatment of electrostatic interactions.

arXiv | JCP


An enhanced version of the heat exchange algorithm with excellent energy conservation properties.

Our new algorithm for imposing a temperature gradient in atomistic simulations conserves the total energy more than a 100 times better than the original one. The article was selected as a 2015 Editors' Choice article!

arXiv | JCP