Philipp Marquetand

@marquetand
264 Followers
245 Following
19 Posts
Theoretical chemist (he/him) at University of Vienna. Machine learning, nonadiabatic dynamics, computational chemistry.
#compchem #ML #MachineLearning #AI #theochem #chemiverse #chemistry
Our NAFFS method is out! Speed up your excited-state dynamics no matter which potential you want to use (ab initio, #ML, …) Yes, you can speed up #machinelearning based dynamics! And overcome excited-state barriers!
Work with Madlen, Brigitta, Max, Sebastian, @leti_gonzalez @CHHDellago #compchem #openpathsampling #univienna #virapid https://doi.org/10.1021/acs.jctc.2c01088
San Valentin ❤️ here I go !
This is my first post in mastodon.

And here our first contribution of the day: #SHARC to unravel mechanistic pathways in a TADF OLED, just published in Angewandte_Chemie

https://doi.org/10.1002/anie.202217620

Gravity batteries in abandoned mines could power the whole planet, scientists say
https://www.techspot.com/news/97306-gravity-batteries-abandoned-mines-could-power-whole-planet.html
Gravity batteries in abandoned mines could power the whole planet, scientists say

A study from the International Institute for Applied Systems Analysis (IIASA) proposes that decommissioned mines could be repurposed to operate gravity batteries. Converting old mines could provide...

TechSpot
#MachineLearning for #Quantum: Guiseppe Carleo @gppcarleo is now on Mastodon, pioneer of deep learning applied to quantum physics, and professor at EPFL Lausanne !

First paper announcement here !
Very nice paper accepted just now on PCCP, first of FA PhD thesis (so we need to come back to some celebration): detailed study on how nuclear quantum effects can be modeled in thermal unimolecular fragmentation.
A preprint is available here:

https://chemrxiv.org/engage/chemrxiv/article-details/63595357311072ffa5f0c3d1

stay tuned for the final version #compchem #science

Quantum versus Classical Unimolecular Fragmentation Rate Constants and Activation Energies at Finite Temperature from Direct Dynamics Simulations

In the present work, we investigate how nuclear quantum effects modify the temperature dependent rate constants and, consequently, the activation energies in unimolecular reactions. In the reactions under study, nuclear quantum effects mainly stem from the presence of a large zero point energy. Thus, we investigate the behavior of methods compatible with direct dynamics simulations, the Quantum Thermal Bath (QTB) and Ring Polymer Molecular Dynamics (RPMD). To this end, we first compare them with quantum reaction theory for a model Morse potential before extending this comparison to molecular models. Our results show that, in particular in the temperature range comparable with or lower than the zero point energy of the system, the RPMD method is able to correctly capture nuclear quantum effects on rate constants and activation energies. On the other hand, although the QTB provides a good description of equilibrium properties including zero-point energy effects, it largely overestimates the rate constants. The origin of the different behaviours is in the different distance distributions provided by the two methods and in particular how they differently describe the tails of such distributions.

ChemRxiv

Hey #ChemiVerse

Nature Chemistry is recruiting for a full-time editor!

The ideal candidate will have expertise in chemical biology or biological chemistry.

Closing date for applications: 12th December

Locations: London, Berlin, New York, Shanghai

#Editorial #Publishing #Chemistry #Jobs #ScienceJobs #ChemJobs

Boosts appreciated!

https://careers.springernature.com/job/London-Associate-or-Senior-Editor%2C-Nature-Chemistry/871414101/

Associate or Senior Editor, Nature Chemistry

Associate or Senior Editor, Nature Chemistry

A little game: guess how many pearls are in the bowl? Please boost for better statistics. #statistics #guessinggame
I had the honor to write a highlight for an @ACSChemRev article by H. Mai, T. Le, D. Chen, D. Winkler, R. Caruso about #ML for electro- and photocatalyst discovery. See https://doi.org/10.1021/acs.chemrev.2c00703 Find the original article at https://doi.org/10.1021/acs.chemrev.2c00061 #compchem #machinelearning #ChemiVerse #theochem #photochemistry

Here's my (currently) favourite figure I made. It's mainly about the colours but the science is also really neat.

I implemented a simple machine learning method based on an XOR (yes!) and a small MLP that outperforms DFT- and transformer-based methods when predicting reaction yields.

You can actually use it right now to predict the yield of Buchwald-Hartwig reactions: https://colab.research.google.com/drive/1VvzSBwCkMi7vi653hTlzZV1YRt_8mk5n?usp=sharing

or just read the paper ;-)
https://pubs.rsc.org/en/content/articlelanding/2022/dd/d1dd00006c

#Science #Chemistry #Work

Google Colaboratory