Thorben Frank

PhD Student @ml_tuberlin and @bifoldberlin | currently @GoogleDeepmind | interested in deep learning & quantum chemistry

[marcel ⊙ @marceldotsci]

Our (@spozdn & @MicheleCeriotti) recent work on testing the impacts of using approximate, rather than exact, rotational invariance in a machine-learning interatomic potential of water has been published by @MLSTjournal! We find that, for bulk water, approximate invariance is ok..

[Thorben Frank @FrankThorben]

It was great fun! Check it out!! 🧬

[Adil Kabylda @kabylda_]

Couldn't have asked for better timing #NobelPrize

Our new work on 'Molecular Simulations with a Pretrained Neural Network and Universal Pairwise Force Fields' is now out on @ChemRxiv!

#MachineLearning #compchem

[ML Group, TU Berlin @ml_tuberlin]

🔊

Check out our latest pre-print! Congrats to the whole team! 👏 #Neuroscience #AI

[ML Group, TU Berlin @ml_tuberlin]

🔊 New publication in @TmlrOrg

Decoupling Pixel Flipping and Occlusion Strategy for Consistent #XAI Benchmarks, by @sbluecher @jvielhaben and @nstrodt together with @bifoldberlin and @FraunhoferHHI 🙏🏼

#MachineLearning #AI

[Alexandre Duval @ADuvalinho]

(Big) life update 😊:

With @vict0rsch & @m_galtier, we are co-founding @Entalpic_ai, an AI startup for chemistry and materials discovery. Our mission is to develop science and its applications for a fair and ecological transition.

In stealth but hiring
https://t.co/idCEredN8B

[Florian Knoop @flokno_phys]

I am happy to announce that the Temperature Dependent Effective Potentials (TDEP) code is officially released and published in JOSS! 🎉

TDEP is a code for advanced phonon simulations in materials with a focus on anharmonic properties. More details below:

https://t.co/HBbrp5i3HN

[ML Group, TU Berlin @ml_tuberlin]

🚨 We have 2 open #PhD positions!

Come join us @TUBerlin. We offer cutting edge #research ranging from #XAI over #Probabilistic #Modelling to #AI4Science 🧬

Check out our research profile 👉 https://t.co/5SZKwWrUIh

Apply here 👉 https://t.co/Y4XCY0kqgx

#AI #MachineLearning

[Thorben Frank @FrankThorben]

🚀

[marcel ⊙ @marceldotsci]

✨ Out in JCP today: https://t.co/mPDbhx2zaV! Everything you ever needed to know about using AD to compute forces, stress, and heat flux. Everything is implemented in the glp library at https://t.co/JqtIl7HngX.

Data and code specific to the paper at https://t.co/W13VCIc7UM!

[Lars Schaaf @lars__schaaf]

❓Struggling to capture non-local effects using message passing? We present Matrix Function Networks, a new architecture that captures non-local effects in a structured and topology aware manner.

https://t.co/9VYEHTNN05 🧵👇(1/10)

[Lars Holdijk @HoldijkLars]

Excited to share our latest work at the intersection of machine learning and computational chemistry; Path Integral Stochastic Optimal Control for Sampling Transition Paths between molecular conformations. With @YuanqiD*, @priyankjaini, Ferry Hooft, @BerndEnsing and @wellingmax

[ML Group, TU Berlin @ml_tuberlin]

🔊💾

Interested in learning @JuliaLanguage for #MachineLearning?

Our PhD student @adr_hill created a beautiful course and all material is available online.

Lectures and homework are interactive using @PlutoJL, giving you instant feedback while working on it.

#JuliaLang

[ML Group, TU Berlin @ml_tuberlin]

This is the #MachineLearning Group of Prof. Klaus-Robert Müller at the @TUBerlin. Let’s connect! #ScienceTwitter

[Florian Knoop @flokno_phys]

New paper day! Ab initio Green-Kubo thermal conductivities for MgO and CuI. Thread will follow

https://t.co/UziJ9GxshG

[Stefan Blücher @sbluecher]

Happy to share that our recent work on scalable kernel solvers is now published in @JCIM_JCTC.
#cheminformatics #compchem #machinelearning @bifoldberlin @TUBerlin

https://t.co/Gv2lq6EJz5

[Thorben Frank @FrankThorben]

Check out this new work w/ @marceldotsci and @flokno_phys which tells you all you need to know about implementing stress and heat flux for machine learning potentials with AD.

Featuring implementations in JAX that allow calculating thermal conductivities in <1d 🚀🔥

[marcel ⊙ @marceldotsci]

🚀 New work: "Stress and heat flux via automatic differentiation" w/ @frankthorben and @flokno_phys. We give an overview of using AD for stress and heat flux, and provide an implementation in JAX. See https://t.co/A1omuOgLUV; code at https://t.co/JqtIl7HngX. 🧵⤵️

[Florian Knoop @flokno_phys]

New Preprint: @marceldotsci and me joined expertises to enable thermal transport simulations with semi-local machine-learning potentials that use message passing, where one has to be careful with the definition of the heat flux. Check out!
https://t.co/VOdGUat0Ou

[marcel ⊙ @marceldotsci]

let's go!!

[Itai Yanai @ItaiYanai]

New Nature paper provides evidence that science has become less innovative since the 1950s. The authors suggest reversing the trend by:
1. reading widely,
2. focusing less on quantity of papers, & more on research quality,
3. taking year-long sabbaticals.
https://t.co/oYV5JCXRl8

[Prof. Anima Anandkumar @AnimaAnandkumar]

I believe @AI_for_Science is a true test of #AI capabilities. While generating realistic text or images can be fun to play with, more controlled tests of #AI is easier with scientific domains where we have more concrete metrics and objectives to measure

[David Pfau @pfau]

OK, this website still seems to be working…so…time to share our latest preprint!

Very pleased to be able to share this one: is attention all you need to solve the Schrödinger equation? https://t.co/Ya4aiFT3l7