Recently, I have decided to change MLatom's license to Apache 2.0 to make it even more permissive than it was (MIT with the citation clause). In addition, installation has been simplified too, just a single command pulling all the most needed dependencies.

Aitomistic@aitomistic.com · 6d ago

MLatom is now Apache 2.0 — permissive, commercial-friendly, patent grant included, and the old citation clause is gone. One line on Python 3.9–3.11: pip install -U mlatom Source: github.com/dralgroup/mlatom Docs: aitomistic.com/mlatom #compchem #mlchem #opensource

just a small example of how all future #compchem calculations will look like. Not just the future, but the present is already agentic.

Aitomistic@aitomistic.com · 3w ago

Find a transition state — just by chatting. One prompt → an NEB path → the TS → its imaginary mode → IRC → the reaction energy diagram. Submitted, monitored & analyzed for you. No commercial QM; powered by AIQM3. Free aitomistic.xyz #compchem #mlchem #AIagents #AI4Science

Insane how far computational chemistry has jumped in a year. A year ago, we had Aitomia — standard simulations (geom opt, freq, …). Today: Protomia (aitomistic.com/protomia) does all that and drafts & edits manuscripts, writes code, and everything else you'd expect from a top agentic system.

Aitomistic@aitomistic.com · last mo.

Our framework, Aitomia, is now published in JCTC! "Aitomia: An Agentic Framework for AI-Driven Atomistic and Quantum Chemical Simulations" doi.org/10.1021/acs.jctc.6c00591 Want the live, ready-to-use system? Try Protomia on the Aitomistic Hub: aitomistic.com/protomia

Here is a reminder about Faraday Discussion poster abstract deadline. I will be giving an invited talk at Faraday’s discussions - might be a good chance to meet some of you in person! invt.io/1bxby0pyxoi

I've registered for Molecular excited states theory and experiment FD, join me

Register now

invt.io

OMNI-P2x is now out in Nat. Commun.! The first universal neural network potential for excited electronic states of small molecules. Approaching TD-DFT accuracy at lower cost. Lots of room to improve; fine-tuning helps. doi.org/10.1038/s414...

OMNI-P2x universal neural network potential for excited-state simulations - Nature Communications

OMNI-P2x is a universal neural network model that predicts molecular excited-state properties with near quantum-chemical accuracy at lower cost. It enables rapid screening and photodynamics simulation...

doi.org

Paper detailing FSSH implementation in MLatom and some cool functionality with hybrid QM/ML models, flexibility, versatile stop function, and interesting benchmark comparisons is now out in JCTC: doi.org/10.1021/acs..... #compchem #mlchem

Flexible Framework for Surface Hopping: From Hybrid Schemes for Machine Learning to Benchmarkable Nonadiabatic Dynamics

Nonadiabatic molecular dynamics is a key technique for investigating a broad range of photochemical and photophysical processes. Among the established approaches, surface hopping schemes are widely us...

doi.org

AIQM3 is out in JCTC: more elements, good performance for drug-design-related applications (better than tested universal MLIPs), and applicability to both charged and radical species, while preserving good performance for barriers. doi.org/10.1021/acs.... #compchem #mlchem @olexandr.bsky.social

Our novel approach of directly predicting nuclear positions for various molecules as a function of time, rather than doing stepwise propagations as in molecule dynamics, is finally published in JCTC: doi.org/10.1021/acs.... #compchem #mlchem #moleculardynamics

Our recent @ChemicalScience article ‘AIQM2: organic reaction simulations beyond DFT’ ( pubs.rsc.org/doi/D5SC02802G ) has been listed in the journal's 2025 most popular machine learning and automation articles collection. #ChemSciMostPopular #compchem #mlchem #aichem

@angewandtechemie.bsky.social with @savateevlab.bsky.social ! I love such collaborative studies, which allow us to look at the practical problems faced in experimental chemistry. Here, we deepen our understanding of the nature of chemical processes and sharpen our theoretical tools.

Our publication on enabling trajectory surface hopping with electronic structure methods without analytical gradients with #ML is out in @chemicalscience.rsc.org ! Read more about it: doi.org/10.1039/D5SC... Detailed tutorials, etc., are coming soon. Stay tuned! #compchem #mlchem

A great piece by @robinson-julia.bsky.social in @chemistryworld.com on how #AIagents will democratize #compchem. Soon, manual QC inputs will feel like building pyramids. Students already start by chatting with Aitomia. Gen-2 coming soon. Check out the older version online at aitomistic.xyz

Julia Robinson@robinson-julia.bsky.social · 7mo ago

LLMs are powering a new generation of #AI agents that could open up the field of computational #chemistry, speeding up research & materials discovery. In this piece for @chemistryworld.com I speak to the developers to find out their motivations & aspirations: www.chemistryworld.com/news/ai-agen...

Have you ever wanted to use cutting-edge #compchem methods to propagate NAMD simulations, such as QD-NEVPT2, but were stopped by the lack of available energy gradients? If so, check out our new preprint by M. Martyka, J. Jankowska, H. Lischka, and @pavlodral.bsky.social : doi.org/10.26434/che...

Gradients not needed: ML-driven propagation of nonadiabatic molecular dynamics without reference gradients

The recent development of machine learning (ML) methods for quantum chemistry has tremendously boosted the efficiency of molecular calculations. In this work, we use ML to enable nonadiabatic molecula...

doi.org

It is my great pleasure to be a subject chair of #RSCPoster 2026! Looking forward to learning about your great science through your digital posters!

Very humbled to see our research with @jakubmartinka.bsky.social, Lina, Mikolaj, Yi-Fan, Jiri, and @mbarbatti.bsky.social among the most-read recent articles in J Phys Chem Lett. Paper: doi.org/10.1021/acs.... My personal account of the study’s background: dr-dral.com/jpcl-a-descr...

I am looking forward to participating in Faraday Discussion's Molecular excited states theory and experiment, 14-16 September 2026, Cambridge, UK, rsc.li/excitedstate... Deadline for Oral abstract submissions is 15 December 2025. #compchem #aichem #mlchem

Just out in JPCL — our accurate ML approach for nonadiabatic coupling vectors! This work took years — from early ML-FSSH struggles to finding physics-based descriptors (energy gradient differences) and improving MLIPs #compchem @mbarbatti.bsky.social doi.org/10.1021/acs....

A Descriptor Is All You Need: Accurate Machine Learning of Nonadiabatic Coupling Vectors

Nonadiabatic couplings (NACs) play a crucial role in modeling photochemical and photophysical processes with methods such as the widely used fewest-switches surface hopping (FSSH). There is, therefore, a strong incentive to machine learn NACs for accelerating simulations. However, this is challenging due to NACs’ vectorial, double-valued character and the singularity near a conical intersection seam. For the first time, we design NAC-specific descriptors based on our domain expertise and show that they allow learning NACs with never-before-reported accuracy of R2 exceeding 0.99. The key to success is also our new ML phase-correction procedure. We demonstrate the efficiency and robustness of our approach on a prototypical example of fully ML-driven FSSH simulations of fulvene targeting the SA-2-CASSCF(6,6) electronic structure level. This ML-FSSH dynamics leads to an accurate description of S1 decay while reducing error bars by allowing the execution of a large ensemble of trajectories. Our approach is generalizable to more states as we demonstrate for a three-state ML-FSSH simulation of methylenimmonium cation. Our implementations are available in open-source MLatom.

doi.org

Delighted to present our AI-driven #compchem work at ICCOC 2025, Shenzhen. Huge congrats to my PhD student Xinxin for winning the Best Poster Prize on AIQM methods — she is surely one of the brightest up-and-coming scientists!

Nice work by my co-supervised PhD student Mateusz! You can read the work at doi.org/10.1021/acs.... .

Continuing the previous post, here is one of my favorite examples of how things can go wrong when you use universal #ML potentials - MD of H2. I love to show this example to my students, and it is in my online course (aitomistic.com/en/sub/course) at @aitomistic.com . #compchem

Pavlo O. Dral@pavlodral.bsky.social · 10mo ago

1/2Just came across this preprint discussing #ML potentials' failure even for H2. In my course, I have been showing this to my students already for many years, with both astonishing examples of failures of popular foundational ML models and tutorials on how to solve them. arxiv.org/abs/2509.26397

My talk at @Smlqc1Smlqc -2025 is now online. This is the third SMLQC edition ( www.smlqc2025.com )! Talk is covering #ML models for #compchem simulations, also available with #AIagents at the @aitomistic.com Hub ( aitomistic.xyz ). youtu.be/gIpE_pqF2e4

SMLQC 2025, Pavlo O. Dral's talk "Universal AI Models for Ground and Excited States"

YouTube video by Prof. Pavlo O. Dral

youtu.be

Poster on Aitomia presented by Hassan Nawaz at #MDMM25, where @pavlodral.bsky.social also gave a talk on Aitomia. Showcasing Aitomia's ability to autonomously design #compchem workflows with #AIagents, such as calculating reaction thermochemistry and spectra, on aitomistic.xyz

1/2Just came across this preprint discussing #ML potentials' failure even for H2. In my course, I have been showing this to my students already for many years, with both astonishing examples of failures of popular foundational ML models and tutorials on how to solve them. arxiv.org/abs/2509.26397

Are neural scaling laws leading quantum chemistry astray?

Neural scaling laws are driving the machine learning community toward training ever-larger foundation models across domains, assuring high accuracy and transferable representations for extrapolative t...

arxiv.org

1/2Just came across this preprint discussing #ML potentials' failure even for H2. In my course, I have been showing this to my students already for many years, with both astonishing examples of failures of popular foundational ML models and tutorials on how to solve them. arxiv.org/abs/2509.26397

Are neural scaling laws leading quantum chemistry astray?

Neural scaling laws are driving the machine learning community toward training ever-larger foundation models across domains, assuring high accuracy and transferable representations for extrapolative t...

arxiv.org

hard work by Xinxin (the first author), she has many more such models in her library! You can run #compchem simulations with AIQM2 as described in our tutorials: mlatom.com/docs/tutoria... Also, online via a web browser on @aitomistic.com Hub at aitomistic.xyz (free)

Chemical Science@chemicalscience.rsc.org · 11mo ago

Issue 35 is here! #ChemSciCovers Our front cover this week features Pavlo O. Dral et al 🤩 'AIQM2: organic reaction simulations beyond DFT' 🔗 doi.org/10.1039/D5SC... pavlodral.bsky.social aitomistic.com

All-in-one leaning is a very handy method to learning from multiple levels of theory (and data sources in general) simultaneously. Better than alternative transfer learning in many respects. Just out in JCTC: pubs.acs.org/doi/10.1021/...

It is always nice to see creative ways the users apply our methods and software (UAIQM & #MLatom) to solve their #compchem problems: www.sciencedirect.com/science/arti... You can use them online too at the @aitomistic.com Hub.

Back in 2021, I wrote about a future where computers could autonomously run & analyze #compchem simulations: shorturl.at/Vq4tq Now, I’m thrilled to be building #AIagents that make this vision real!

Artificial intelligence makes accurate quantum chemical simulations more affordable

We have developed artificial intelligence-enhanced quantum mechanical method 1 (AIQM1), which can be used out of the box for very fast quantum chemical calculations with the accuracy of the gold-stand...

shorturl.at

Aitomistic@aitomistic.com · 12mo ago

🚀 Autonomous #compchem by #AI is here, not future. In our demo, Aitomia on Aitomistic Hub computed the Diels–Alder reaction energy & thermodynamics from scratch in 5 min on 1 CPU—same accuracy as manual setup, but 4× faster & zero human cost. 🔗 aitomistic.xyz