Tim Duignan
@timothyduignan.bsky.social
Researcher at Orbital Materials. Working on molecular simulation with ML for chemical engineering applications.
I actually observed these pairs in water a while ago with an NNP and worried a lot that they were a hallucination as it is very counter intuitive that two such small cations would pair like that. But looks like it’s real, and really important! (iopscience.iop.org/article/10.1...)
October 9, 2025 at 1:00 PM
I actually observed these pairs in water a while ago with an NNP and worried a lot that they were a hallucination as it is very counter intuitive that two such small cations would pair like that. But looks like it’s real, and really important! (iopscience.iop.org/article/10.1...)
Couple of very nice new papers on understanding the SEI formation in lithium ion batteries using neural network/machine learning interatomic potentials (NNPs/MLIPs).
October 9, 2025 at 1:00 PM
Couple of very nice new papers on understanding the SEI formation in lithium ion batteries using neural network/machine learning interatomic potentials (NNPs/MLIPs).
Universal machine learning forcefields beating tailor made classical potentials for zeolites quite convincingly. Great to see all these benchmarking papers! Again demonstrates accurate training data + speed should be key focus now.
arxiv.org/abs/2509.07417
arxiv.org/abs/2509.07417
September 18, 2025 at 11:14 AM
Universal machine learning forcefields beating tailor made classical potentials for zeolites quite convincingly. Great to see all these benchmarking papers! Again demonstrates accurate training data + speed should be key focus now.
arxiv.org/abs/2509.07417
arxiv.org/abs/2509.07417
arxiv.org/pdf/2508.15614 This is right and it's a big deal. Been waiting my whole career for this point. So many things to simulate!
September 8, 2025 at 11:19 AM
arxiv.org/pdf/2508.15614 This is right and it's a big deal. Been waiting my whole career for this point. So many things to simulate!
Another very interesting benchmarking paper on NNPs. lnkd.in/gWbcTQw8 It seems the models are pretty much there. Very exciting times as these new large datasets continue to be built. Always need more though!
September 2, 2025 at 11:55 AM
Another very interesting benchmarking paper on NNPs. lnkd.in/gWbcTQw8 It seems the models are pretty much there. Very exciting times as these new large datasets continue to be built. Always need more though!
I'm particularly excited about how close the structure stays to experiment, with no constraints, even though the training data doesn't contain a single protein.
August 28, 2025 at 12:18 PM
I'm particularly excited about how close the structure stays to experiment, with no constraints, even though the training data doesn't contain a single protein.
And you can look at huge systems with them. Like the 20,000 atoms solvated carbonic anhydrase enzyme with many different complex interactions going on. You can do hundreds of thousands of calculations on it with a single GPU in a few days, with no unphysical behaviour.
August 28, 2025 at 12:18 PM
And you can look at huge systems with them. Like the 20,000 atoms solvated carbonic anhydrase enzyme with many different complex interactions going on. You can do hundreds of thousands of calculations on it with a single GPU in a few days, with no unphysical behaviour.
It's amazing to me that you can just pick general purpose DFT validation sets and benchmark them like they are a DFT functional and they will normally do a great job out of the box. Often similar to a dispersion corrected GGA or better but orders of magnitude faster.
August 28, 2025 at 12:18 PM
It's amazing to me that you can just pick general purpose DFT validation sets and benchmark them like they are a DFT functional and they will normally do a great job out of the box. Often similar to a dispersion corrected GGA or better but orders of magnitude faster.
Another remarkable jump in accuracy with these new OrbMol models for simulating chemistry. For example, they now quantitatively reproduce the structure of water. But they should be just as applicable for studying a vast range of different liquids.
August 28, 2025 at 12:18 PM
Another remarkable jump in accuracy with these new OrbMol models for simulating chemistry. For example, they now quantitatively reproduce the structure of water. But they should be just as applicable for studying a vast range of different liquids.
Another nice benchmarking paper highlighting the rapid exciting progress of universal MLIPS/NNPs: www.arxiv.org/abs/2507.11806
July 23, 2025 at 11:57 PM
Another nice benchmarking paper highlighting the rapid exciting progress of universal MLIPS/NNPs: www.arxiv.org/abs/2507.11806
Love this combination of LLMs and NNPs, a powerful pair of tools. www.sciencedirect.com/science/arti... Also wonderful to see people picking up Orb so quickly and getting good results!
June 23, 2025 at 6:36 AM
Love this combination of LLMs and NNPs, a powerful pair of tools. www.sciencedirect.com/science/arti... Also wonderful to see people picking up Orb so quickly and getting good results!
This is excellent! arxiv.org/abs/2506.14492
June 19, 2025 at 12:10 PM
This is excellent! arxiv.org/abs/2506.14492
So many nice NNP papers coming out now it is impossible to stay on top of them. Four very cool recent ones:
June 18, 2025 at 12:07 PM
So many nice NNP papers coming out now it is impossible to stay on top of them. Four very cool recent ones:
So Orb has blown me away again. I simulated the carbonic anhydrase enzyme with it: one of the most important and well studied enzymes in biology. (It converts CO2 to bicarbonate and is involved in many diseases and could also be useful for carbon capture.)
March 19, 2025 at 10:29 AM
So Orb has blown me away again. I simulated the carbonic anhydrase enzyme with it: one of the most important and well studied enzymes in biology. (It converts CO2 to bicarbonate and is involved in many diseases and could also be useful for carbon capture.)
Running out of memory used to be a common headache when running molecular simulations with neural network potentials. Not any more. Here Orb is simulating over half a million atoms on a single GPU (H200). This is a fully solvated COVID spike protein.
Models here: github.com/orbital-mate...
Models here: github.com/orbital-mate...
February 25, 2025 at 12:02 PM
Running out of memory used to be a common headache when running molecular simulations with neural network potentials. Not any more. Here Orb is simulating over half a million atoms on a single GPU (H200). This is a fully solvated COVID spike protein.
Models here: github.com/orbital-mate...
Models here: github.com/orbital-mate...
Uranium is one of the hardest elements to simulate due to the large number of electrons, so I thought I would try it with Orb and remarkably it seems to behaves well even getting the melting point roughly correct. I think looking at this systematically for many metals would be a great project.
February 13, 2025 at 11:33 AM
Uranium is one of the hardest elements to simulate due to the large number of electrons, so I thought I would try it with Orb and remarkably it seems to behaves well even getting the melting point roughly correct. I think looking at this systematically for many metals would be a great project.
Had a lot of fun outlining how I think AI accelerated simulation with tools like Orb are going to be profoundly useful for Chemical Engineering in this article for The Chemical Engineer:
www.thechemicalengineer.com/features/vie...
www.thechemicalengineer.com/features/vie...
February 12, 2025 at 12:05 PM
Had a lot of fun outlining how I think AI accelerated simulation with tools like Orb are going to be profoundly useful for Chemical Engineering in this article for The Chemical Engineer:
www.thechemicalengineer.com/features/vie...
www.thechemicalengineer.com/features/vie...
Here's a fun one you can't do in the lab: diamond melting at thousands of degrees simulated with Orb. Simulate anything you want with it here: colab.research.google.com/github/timdu...
February 10, 2025 at 12:44 PM
Here's a fun one you can't do in the lab: diamond melting at thousands of degrees simulated with Orb. Simulate anything you want with it here: colab.research.google.com/github/timdu...
Simulating over 10,000 atoms for 10 ps a day on my Macbook with close to quantum chemical accuracy using Orb. I can do this for almost any element from the periodic table I want. Just a few years ago this would have been totally inconceivable with even the world's largest supercomputers.
January 8, 2025 at 12:21 PM
Simulating over 10,000 atoms for 10 ps a day on my Macbook with close to quantum chemical accuracy using Orb. I can do this for almost any element from the periodic table I want. Just a few years ago this would have been totally inconceivable with even the world's largest supercomputers.
Lots of useful and interesting benchmarking of universal force fields in this paper. This fields really picking up now. Good to see all them doing well on silicon now. arxiv.org/abs/2412.10516
January 7, 2025 at 12:21 PM
Lots of useful and interesting benchmarking of universal force fields in this paper. This fields really picking up now. Good to see all them doing well on silicon now. arxiv.org/abs/2412.10516
Excellent perspective on using machine learning for coarse grained simulations. This is an incredibly promising field in my opinion. Lots of low hanging fruit to implement that will be really impactful, e.g, this is a great one: www.sciencedirect.com/science/arti...
January 6, 2025 at 11:39 AM
Excellent perspective on using machine learning for coarse grained simulations. This is an incredibly promising field in my opinion. Lots of low hanging fruit to implement that will be really impactful, e.g, this is a great one: www.sciencedirect.com/science/arti...
This is incredible, Orb can predict phase diagrams of metal alloys remarkably well. This is a very important class of materials that we now have a powerful new tool to study computationally starting from nothing but quantum mechanics. 1/2
December 17, 2024 at 1:52 PM
This is incredible, Orb can predict phase diagrams of metal alloys remarkably well. This is a very important class of materials that we now have a powerful new tool to study computationally starting from nothing but quantum mechanics. 1/2
So I think I've found another pretty incredible example of the generalisability of neural network potentials: this is a problem I've been dreaming of tackling for a decade but never felt I had the the tools to get at until now: How do potassium ion channels work. 1/n
December 2, 2024 at 1:31 PM
So I think I've found another pretty incredible example of the generalisability of neural network potentials: this is a problem I've been dreaming of tackling for a decade but never felt I had the the tools to get at until now: How do potassium ion channels work. 1/n
Yeah good point I should say ‘PDB and sequence information’. I like the interpretation that the structural info is key to the energy function but the sequence data is needed to optimise the search space. From here: journals.aps.org/prl/pdf/10.1...
December 2, 2024 at 12:51 AM
Yeah good point I should say ‘PDB and sequence information’. I like the interpretation that the structural info is key to the energy function but the sequence data is needed to optimise the search space. From here: journals.aps.org/prl/pdf/10.1...