Xin@QuantumChem
@xinquantum.bsky.social
Computational Chemist @NJ/NYC Github: https://github.com/XinChenQC
Testing three LLM models (Opus 4.7, Sonnet 4.6, Gemma 4) on LogP prediction. The pairwise ranking is the most important thing, 76% accuracy is acceptable. Accurate numerical prediction still relies on dedicated models, this test is more about how well LLMs understand chemistry. #compchemsky
Catalytic surface editing is pretty easy on LightCone interface. Good for heterogeneous cat. system building. #compchemsky #compchem
Two rescued cats. The first is named Isolated Belt (as I found it in an isolated belt bush). The second is Huahua (at least 9 yro). #compchem #biology #chemistry #compchemsky
Extreme compression test on C60. Even under very aggressive deformation, xTB stays stable and gives a reasonable energy trend. (Without clear bond breaking). @myzzzz.bsky.social would like to share the detailed files.
Phenomenal
Nicolas @nicolasn.bsky.social · 7mo agoI was able to do a longer time scale simulation of this system. There are still some problems and things I need to figure out, but here is the video showing the trajectory, with the input file at the end #chemsky #compchem #aimnet2 #packmol #orca youtu.be/JocFvLFY_Kk
When it comes to real production level chemistry code, both Gemini 3 Pro and Claude Opus became clumsy. It took them around 40 minutes and multiple rounds of conversation just to finish the project. They’re clearly not great with RDKit or chemistry APIs. #CompChemsky #compchem #Chemistry
In one sentence: Orb-v3 is accurate in describing F-Li-Be Molten Salt *without fine-tuning*. AIMD results came from: doi.org/10.1103/PRXE... They also compared with Neutron Diffraction. @myzzzz.bsky.social will benchmark the S(Q) with experimental data. #compchemsky #compchem
Wetlab is a different world. A wetlab biologist just told me she uses a visualization website where you input a protein sequence and click, the 3D structures show instantly. The visualization thing is called AlphaFold. #Biology #AlphaFold #compchemsky
Yesterday I posted an incorrect comparison. I accidentally copied the TS3 structure into TS1 for UMA and Orb-V3, so the MLIP results for TS1 were wrong. Really sorry about that misleading! In the next post I will share IRC of DFT calculation and wB97M-V/def2-TZVPD results. #compchem #compchemsky
Xin@QuantumChem@xinquantum.bsky.social · 9mo agoA failure case for deep-learning potential: the TS1 geometry is so weird that UMA and Orb-v3 underestimate the barrier by ~15 kcal/mol, likely because it far from the training set. Handling chemical reactions is extremely hard for general deep learning models. #compchem #compchemsky
A failure case for deep-learning potential: the TS1 geometry is so weird that UMA and Orb-v3 underestimate the barrier by ~15 kcal/mol, likely because it far from the training set. Handling chemical reactions is extremely hard for general deep learning models. #compchem #compchemsky
UMA (small model) is very close to DFT M06-2X. Maybe I can deploy nanoreactors using UMA model for coarse reaction space sampling. This is a reactive trajectory of a compound with 24 Carbon atoms. #CompChem #ChemChemSky
playground.tracto.ai/playground/n... The newly released open-source model from OpenAI can be tried on TractoAI playground for *FREE*.
TractoAI | Cloud platform for AI and Data teams
Run AI and Data workloads in the cloud at scale
playground.tracto.ai
This old theory is revitalized by deep learning. Building accurate kinetic energy functionals (< 0.5 kcal/mol) may be harder than climbing to the top of exchange-correlation Jacob's ladder.🪜
IOPP Machine Learning and AI@iopp-mlresearch.bsky.social · last yr.#MachineLearning (ML) for orbital-free density functional theory (OF-DFT) and OF-DFT for ML. Manzhos and coworkers tackle the kinetic energy functional in this new #MachineLearningScienceandTechnology article. #compchem #CompChemSky Read it here: bit.ly/45eGM0f
Cunning ChatGPT bypassing 3D depth analysis, and calling the rdkit function to generate the coordinate of an Ibuprofen. Grok3 just failed this test and misrecognized it as glucose, as expected.
Boltz2 can be run on a $430 RTX5060 Ti (16G) card, although it's much slower than ~$30,000 H100 (RTX 5060 Ti: 1min 47s, H100: 18s), but it doesn't matter, because the cold-start time is about 1 minute for both cards. #CompChemSky #CompChem #GPU
This is creepy, but lots of fun. Gemini Veo has a perception of spatial depth and can rotate the molecules. Inspired by www.youtube.com/watch?v=mDyt... #compchem #compchemsky
I ran a benchmark on a protein with ~3450 atoms. AIMnet2 takes 4.6 seconds, and UMA(small) takes 7.9 seconds. CUDA 12.4 on a Tesla T4. Additionally, the AIMnet2 checkpoint file is only 11.1 MB. The testing notebook link is shared. #compchemsky #compchem colab.research.google.com/drive/1g1PiW...
Google Colab
colab.research.google.com
This is a very interesting package. I will try this on colab notebook.
Alister Goodfellow@aligoodfellow.bsky.social · last yr.Wrote a bit of code to quickly analyse vibrational modes from QM calculations! #compchem github.com/aligfellow/v... Returns a list of internal coordinates associated with the vibrational mode, can be used from command line on xyz trjfiles or orca or gaussian outputs etc.
UMA (uma-sm) combined with NEB successfully locates the transition state and path for a Ni-catalyzed reaction. UMA prediction lies between two DFT functional, and the transition state closely matches the result from a literature: www.nature.com/articles/nat... #compchemsky #chemsky #compchem
The relationship between Ki and IC₅₀ is described by the Cheng–Prusoff equation. And William L. Jorgensen et al. did an effective calculation for predicting potency in his paper during pandemic. pubs.acs.org/doi/10.1021/...
2nd virtual lecture in the DFT&Application series starts today at 1500hrs CET (in 15 minutes!). Topics are Hartree Fock, SCF, and their connection to DFT. Join us in Zoom: us06web.zoom.us/j/8607128794... #compchem #dft
Join our Cloud HD Video Meeting
Zoom is the leader in modern enterprise cloud communications.
us06web.zoom.us