Viktor Zaverkin
@viktorzaverkin.bsky.social
Research Scientist @ NEC Labs Europe, Ph.D. in Theoretical Chemistry @ SimTech & @unistuttgart.bsky.social, ML/DL for Chemistry & Materials Science
For more details about the work on learning uniformly accurate interatomic potentials from scratch I'll present in B1.36:
📄 Paper: www.nature.com/articles/s41...
💻 Code: github.com/nec-research...
📄 Paper: www.nature.com/articles/s41...
💻 Code: github.com/nec-research...
Uncertainty-biased molecular dynamics for learning uniformly accurate interatomic potentials - npj Computational Materials
npj Computational Materials - Uncertainty-biased molecular dynamics for learning uniformly accurate interatomic potentials
www.nature.com
August 24, 2025 at 12:37 PM
For more details about the work on learning uniformly accurate interatomic potentials from scratch I'll present in B1.36:
📄 Paper: www.nature.com/articles/s41...
💻 Code: github.com/nec-research...
📄 Paper: www.nature.com/articles/s41...
💻 Code: github.com/nec-research...
August 24, 2025 at 12:37 PM
🧵 TL;DR:
✅ Benchmark metrics improve with model size and electrostatics
❌ These gains don't always translate to improved simulation outcomes
⚠️ Training data & evaluation practices remain key bottlenecks
📄Preprint: arxiv.org/abs/2508.10841
💻Code: github.com/nec-research...
-/
✅ Benchmark metrics improve with model size and electrostatics
❌ These gains don't always translate to improved simulation outcomes
⚠️ Training data & evaluation practices remain key bottlenecks
📄Preprint: arxiv.org/abs/2508.10841
💻Code: github.com/nec-research...
-/
Performance of universal machine-learned potentials with explicit long-range interactions in biomolecular simulations
Universal machine-learned potentials promise transferable accuracy across compositional and vibrational degrees of freedom, yet their application to biomolecular simulations remains underexplored. Thi...
arxiv.org
August 15, 2025 at 8:30 AM
🧵 TL;DR:
✅ Benchmark metrics improve with model size and electrostatics
❌ These gains don't always translate to improved simulation outcomes
⚠️ Training data & evaluation practices remain key bottlenecks
📄Preprint: arxiv.org/abs/2508.10841
💻Code: github.com/nec-research...
-/
✅ Benchmark metrics improve with model size and electrostatics
❌ These gains don't always translate to improved simulation outcomes
⚠️ Training data & evaluation practices remain key bottlenecks
📄Preprint: arxiv.org/abs/2508.10841
💻Code: github.com/nec-research...
-/
August 15, 2025 at 8:30 AM
Moreover, simulation results are sensitive to training data composition.
E.g., water density predictions depend on whether NaCl-water clusters were in the training set: compare ICTP-LR(M) vs. ICTP-LR(M)*.
Legend: Solid green - ICTP-LR(M); dashed green - ICTP-LR(M)*.
14/
E.g., water density predictions depend on whether NaCl-water clusters were in the training set: compare ICTP-LR(M) vs. ICTP-LR(M)*.
Legend: Solid green - ICTP-LR(M); dashed green - ICTP-LR(M)*.
14/
August 15, 2025 at 8:30 AM
Moreover, simulation results are sensitive to training data composition.
E.g., water density predictions depend on whether NaCl-water clusters were in the training set: compare ICTP-LR(M) vs. ICTP-LR(M)*.
Legend: Solid green - ICTP-LR(M); dashed green - ICTP-LR(M)*.
14/
E.g., water density predictions depend on whether NaCl-water clusters were in the training set: compare ICTP-LR(M) vs. ICTP-LR(M)*.
Legend: Solid green - ICTP-LR(M); dashed green - ICTP-LR(M)*.
14/
Without DFT-level simulations or other baselines, it is difficult to assess to what extent universal ML potentials improve on classical FFs in realistic biomolecular settings.
While their qualitative advantages are often evident, quantitative validation remains challenging.
13/
While their qualitative advantages are often evident, quantitative validation remains challenging.
13/
August 15, 2025 at 8:30 AM
Without DFT-level simulations or other baselines, it is difficult to assess to what extent universal ML potentials improve on classical FFs in realistic biomolecular settings.
While their qualitative advantages are often evident, quantitative validation remains challenging.
13/
While their qualitative advantages are often evident, quantitative validation remains challenging.
13/
These results highlight the limitations of current evaluation practices.
12/
12/
August 15, 2025 at 8:30 AM
These results highlight the limitations of current evaluation practices.
12/
12/
In Trp-cage, simulations with explicit long-range electrostatics exhibit greater conformational variability.
However, the origin of these effects remains unclear without DFT-level simulations.
11/
However, the origin of these effects remains unclear without DFT-level simulations.
11/
August 15, 2025 at 8:30 AM
In Trp-cage, simulations with explicit long-range electrostatics exhibit greater conformational variability.
However, the origin of these effects remains unclear without DFT-level simulations.
11/
However, the origin of these effects remains unclear without DFT-level simulations.
11/
For Crambin, no significant differences are observed for the vibrational spectrum.
10/
10/
August 15, 2025 at 8:30 AM
For Crambin, no significant differences are observed for the vibrational spectrum.
10/
10/
For Ala3, larger models better reproduce experimental J-couplings.
9/
9/
August 15, 2025 at 8:30 AM
For Ala3, larger models better reproduce experimental J-couplings.
9/
9/
For water and NaCl-water mixtures:
- Larger models don't consistently outperform smaller ones
- Increasing model size doesn't yield systematic convergence
- Explicit electrostatics shifts density predictions from overestimation to underestimation, without consistent gains.
8/
- Larger models don't consistently outperform smaller ones
- Increasing model size doesn't yield systematic convergence
- Explicit electrostatics shifts density predictions from overestimation to underestimation, without consistent gains.
8/
August 15, 2025 at 8:30 AM
For water and NaCl-water mixtures:
- Larger models don't consistently outperform smaller ones
- Increasing model size doesn't yield systematic convergence
- Explicit electrostatics shifts density predictions from overestimation to underestimation, without consistent gains.
8/
- Larger models don't consistently outperform smaller ones
- Increasing model size doesn't yield systematic convergence
- Explicit electrostatics shifts density predictions from overestimation to underestimation, without consistent gains.
8/
BUT: These improvements do not consistently translate into more accurate physical observables in simulations.
Densities, radial distribution functions, and conformational ensembles show inconsistent trends with model size and long-range electrostatics.
7/
Densities, radial distribution functions, and conformational ensembles show inconsistent trends with model size and long-range electrostatics.
7/
August 15, 2025 at 8:30 AM
BUT: These improvements do not consistently translate into more accurate physical observables in simulations.
Densities, radial distribution functions, and conformational ensembles show inconsistent trends with model size and long-range electrostatics.
7/
Densities, radial distribution functions, and conformational ensembles show inconsistent trends with model size and long-range electrostatics.
7/
As expected, benchmark metrics (e.g., energy & force RMSEs) systematically improve with increasing model size and the inclusion of explicit long-range interactions.
6/
6/
August 15, 2025 at 8:30 AM
As expected, benchmark metrics (e.g., energy & force RMSEs) systematically improve with increasing model size and the inclusion of explicit long-range interactions.
6/
6/
We use DIMOS for our simulations:
📄Preprint: arxiv.org/abs/2503.20541
💻Code: github.com/nec-research...
5/
📄Preprint: arxiv.org/abs/2503.20541
💻Code: github.com/nec-research...
5/
August 15, 2025 at 8:30 AM
We use DIMOS for our simulations:
📄Preprint: arxiv.org/abs/2503.20541
💻Code: github.com/nec-research...
5/
📄Preprint: arxiv.org/abs/2503.20541
💻Code: github.com/nec-research...
5/
We assess the impact of model size, dataset composition, and explicit long-range electrostatics across:
📊 Benchmark datasets
💧 Pure liquid water
🧂 NaCl-water mixtures
🧬 Small peptides (blocked and cationic Ala3)
🧪 Small proteins (Trp-cage, Crambin)
4/
📊 Benchmark datasets
💧 Pure liquid water
🧂 NaCl-water mixtures
🧬 Small peptides (blocked and cationic Ala3)
🧪 Small proteins (Trp-cage, Crambin)
4/
August 15, 2025 at 8:30 AM
We assess the impact of model size, dataset composition, and explicit long-range electrostatics across:
📊 Benchmark datasets
💧 Pure liquid water
🧂 NaCl-water mixtures
🧬 Small peptides (blocked and cationic Ala3)
🧪 Small proteins (Trp-cage, Crambin)
4/
📊 Benchmark datasets
💧 Pure liquid water
🧂 NaCl-water mixtures
🧬 Small peptides (blocked and cationic Ala3)
🧪 Small proteins (Trp-cage, Crambin)
4/
DFT-level simulations and other high-quality baselines are unavailable or infeasible for biomolecular systems.
A more reliable evaluation should consider how model expressivity (model size, explicit long-range interactions) affects prediction errors and simulation results.
3/
A more reliable evaluation should consider how model expressivity (model size, explicit long-range interactions) affects prediction errors and simulation results.
3/
August 15, 2025 at 8:30 AM
DFT-level simulations and other high-quality baselines are unavailable or infeasible for biomolecular systems.
A more reliable evaluation should consider how model expressivity (model size, explicit long-range interactions) affects prediction errors and simulation results.
3/
A more reliable evaluation should consider how model expressivity (model size, explicit long-range interactions) affects prediction errors and simulation results.
3/
📎Paper: pubs.acs.org/doi/abs/10.1... (but check also pubs.acs.org/doi/abs/10.1...)
💻Code(s): gitlab.com/zaverkin_v/g... (TensorFlow), github.com/nec-research... (PyTorch), and github.com/apax-hub/apax (JAX)
💻Code(s): gitlab.com/zaverkin_v/g... (TensorFlow), github.com/nec-research... (PyTorch), and github.com/apax-hub/apax (JAX)
Gaussian Moments as Physically Inspired Molecular Descriptors for Accurate and Scalable Machine Learning Potentials
Machine learning techniques allow a direct mapping of atomic positions and nuclear charges to the potential energy surface with almost ab initio accuracy and the computational efficiency of empirical ...
pubs.acs.org
June 25, 2025 at 2:42 PM
📎Paper: pubs.acs.org/doi/abs/10.1... (but check also pubs.acs.org/doi/abs/10.1...)
💻Code(s): gitlab.com/zaverkin_v/g... (TensorFlow), github.com/nec-research... (PyTorch), and github.com/apax-hub/apax (JAX)
💻Code(s): gitlab.com/zaverkin_v/g... (TensorFlow), github.com/nec-research... (PyTorch), and github.com/apax-hub/apax (JAX)
Many thanks to everyone who has read, cited, or built on it. I hope it continues to be helpful!
June 25, 2025 at 2:42 PM
Many thanks to everyone who has read, cited, or built on it. I hope it continues to be helpful!
We proposed using full tensor contractions to construct many-body features, thereby avoiding expensive sums over triplets, quadruplets, and so on. I am thrilled to see that similar ideas are now an integral part of state-of-the-art architectures, such as MACE, CACE, and so on. 💪
June 25, 2025 at 2:42 PM
We proposed using full tensor contractions to construct many-body features, thereby avoiding expensive sums over triplets, quadruplets, and so on. I am thrilled to see that similar ideas are now an integral part of state-of-the-art architectures, such as MACE, CACE, and so on. 💪