EquiFiLM: Charge-Conditioned Equivariant Force Fields via Feature-wise Linear Modulation

Authors: Samuel Sahel-Schackis, Ken-ichi Nomura, Aiichiro Nakano, Matthias F. Kling, Thomas Linker

Abstract: Foundation machine learning force fields (MLFFs) such as MACE-MP-0 and UMA cover broad chemical space at near density functional theory (DFT) accuracy. However, they assume equilibrium ground-state physics and do not natively handle externally induced changes to the electronic state, such as charging, applied fields, or electronic excitation, which limits their use for driven processes such as pho… ▽ More Foundation machine learning force fields (MLFFs) such as MACE-MP-0 and UMA cover broad chemical space at near density functional theory (DFT) accuracy. However, they assume equilibrium ground-state physics and do not natively handle externally induced changes to the electronic state, such as charging, applied fields, or electronic excitation, which limits their use for driven processes such as photoexcitation and charge injection. We propose EquiFiLM, a lightweight extension that adds continuous external conditioning to any equivariant foundation MLFF via a per-layer Feature-wise Linear Modulation (FiLM) block, learning externally driven changes to the potential energy surface from minimal training data. The block modulates only scalar channels and preserves E(3)-equivariance exactly. We demonstrate the recipe on charged liquid water with the foundation model MACE-MatPES as the backbone, yielding E-MACE. On the four training charges, E-MACE delivers a $3.1\times$ reduction in force RMSE ($21.3$ to $6.96$ meV/$\mathring{A}$) and a $61\times$ reduction in per-atom energy RMSE ($6.1$ to $0.1$ meV/atom) over a baseline without EquiFiLM trained on the same data, at indistinguishable inference cost. Across seven held-out interpolation and extrapolation charges, force RMSE stays within $18-61$ meV/$\mathring{A}$ and energy RMSE within $0.7-5.4$ meV/atom. The model runs stable molecular dynamics across the full range tested and predicts the charge-dependent first-shell response of the reduced pair distribution function probed by ultrafast electron diffraction. Adding this conditioning axis to the foundation requires only a few thousand DFT-labeled frames, against the $\approx 10^8$ structures of a charge-aware foundation trained from scratch. The recipe is backbone- and conditioning-agnostic: it applies without architectural change to any equivariant MLFF with scalar interaction-layer channels. △ Less

Submitted 6 July, 2026; originally announced July 2026.

Toward Exascale AI for Science: A Scalable AI Skill for Autonomous Microkinetics Discovery

Authors: Ken-ichi Nomura, William Dawson, Nabankur Dasgupta, Taufeq Mohammed Razakh, Thomas Linker, Kai Ito, Aiichiro Nakano

Abstract: We present a scalable AI-driven framework that advances autonomous scientific discovery by combining agentic workflow automation, high-performance computing, and scientific surrogate models. Using microkinetics discovery as a testbed, the work demonstrates how AI can reduce expert intervention, recover from failed simulations, and systematically evaluate surrogate model reliability. This study sho… ▽ More We present a scalable AI-driven framework that advances autonomous scientific discovery by combining agentic workflow automation, high-performance computing, and scientific surrogate models. Using microkinetics discovery as a testbed, the work demonstrates how AI can reduce expert intervention, recover from failed simulations, and systematically evaluate surrogate model reliability. This study shows how AI skills can transform complex domain workflows into robust, scalable capabilities for next-generation materials research. △ Less

Submitted 3 July, 2026; v1 submitted 27 June, 2026; originally announced June 2026.

When Object-Centric World Models Meet Policy Learning: From Pixels to Policies, and Where It Breaks

Authors: Stefano Ferraro, Akihiro Nakano, Masahiro Suzuki, Yutaka Matsuo

Abstract: Object-centric world models (OCWM) aim to decompose visual scenes into object-level representations, providing structured abstractions that could improve compositional generalization and data efficiency in reinforcement learning. We hypothesize that explicitly disentangled object-level representations, by localizing task-relevant information, can enhance policy performance across novel feature com… ▽ More Object-centric world models (OCWM) aim to decompose visual scenes into object-level representations, providing structured abstractions that could improve compositional generalization and data efficiency in reinforcement learning. We hypothesize that explicitly disentangled object-level representations, by localizing task-relevant information, can enhance policy performance across novel feature combinations. To test this hypothesis, we introduce DLPWM, a fully unsupervised, disentangled object-centric world model that learns object-level latents directly from pixels. DLPWM achieves strong reconstruction and prediction performance, including robustness to several out-of-distribution (OOD) visual variations. However, when used for downstream model-based control, policies trained on DLPWM latents underperform compared to DreamerV3. Through latent-trajectory analyses, we identify representation shift during multi-object interactions as a key driver of unstable policy learning. Our results suggest that, although object-centric perception supports robust visual modeling, achieving stable control requires mitigating latent drift. △ Less

Submitted 11 November, 2025; v1 submitted 8 November, 2025; originally announced November 2025.

Object-Centric Temporal Consistency via Conditional Autoregressive Inductive Biases

Authors: Cristian Meo, Akihiro Nakano, Mircea Lică, Aniket Didolkar, Masahiro Suzuki, Anirudh Goyal, Mengmi Zhang, Justin Dauwels, Yutaka Matsuo, Yoshua Bengio

Abstract: Unsupervised object-centric learning from videos is a promising approach towards learning compositional representations that can be applied to various downstream tasks, such as prediction and reasoning. Recently, it was shown that pretrained Vision Transformers (ViTs) can be useful to learn object-centric representations on real-world video datasets. However, while these approaches succeed at extr… ▽ More Unsupervised object-centric learning from videos is a promising approach towards learning compositional representations that can be applied to various downstream tasks, such as prediction and reasoning. Recently, it was shown that pretrained Vision Transformers (ViTs) can be useful to learn object-centric representations on real-world video datasets. However, while these approaches succeed at extracting objects from the scenes, the slot-based representations fail to maintain temporal consistency across consecutive frames in a video, i.e. the mapping of objects to slots changes across the video. To address this, we introduce Conditional Autoregressive Slot Attention (CA-SA), a framework that enhances the temporal consistency of extracted object-centric representations in video-centric vision tasks. Leveraging an autoregressive prior network to condition representations on previous timesteps and a novel consistency loss function, CA-SA predicts future slot representations and imposes consistency across frames. We present qualitative and quantitative results showing that our proposed method outperforms the considered baselines on downstream tasks, such as video prediction and visual question-answering tasks. △ Less

Submitted 21 October, 2024; originally announced October 2024.

Masked Generative Priors Improve World Models Sequence Modelling Capabilities

Authors: Cristian Meo, Mircea Lica, Zarif Ikram, Akihiro Nakano, Vedant Shah, Aniket Rajiv Didolkar, Dianbo Liu, Anirudh Goyal, Justin Dauwels

Abstract: Deep Reinforcement Learning (RL) has become the leading approach for creating artificial agents in complex environments. Model-based approaches, which are RL methods with world models that predict environment dynamics, are among the most promising directions for improving data efficiency, forming a critical step toward bridging the gap between research and real-world deployment. In particular, wor… ▽ More Deep Reinforcement Learning (RL) has become the leading approach for creating artificial agents in complex environments. Model-based approaches, which are RL methods with world models that predict environment dynamics, are among the most promising directions for improving data efficiency, forming a critical step toward bridging the gap between research and real-world deployment. In particular, world models enhance sample efficiency by learning in imagination, which involves training a generative sequence model of the environment in a self-supervised manner. Recently, Masked Generative Modelling has emerged as a more efficient and superior inductive bias for modelling and generating token sequences. Building on the Efficient Stochastic Transformer-based World Models (STORM) architecture, we replace the traditional MLP prior with a Masked Generative Prior (e.g., MaskGIT Prior) and introduce GIT-STORM. We evaluate our model on two downstream tasks: reinforcement learning and video prediction. GIT-STORM demonstrates substantial performance gains in RL tasks on the Atari 100k benchmark. Moreover, we apply Transformer-based World Models to continuous action environments for the first time, addressing a significant gap in prior research. To achieve this, we employ a state mixer function that integrates latent state representations with actions, enabling our model to handle continuous control tasks. We validate this approach through qualitative and quantitative analyses on the DeepMind Control Suite, showcasing the effectiveness of Transformer-based World Models in this new domain. Our results highlight the versatility and efficacy of the MaskGIT dynamics prior, paving the way for more accurate world models and effective RL policies. △ Less

Submitted 22 October, 2025; v1 submitted 10 October, 2024; originally announced October 2024.

Generative AI-driven forecasting of oil production

Authors: Yash Gandhi, Kexin Zheng, Birendra Jha, Ken-ichi Nomura, Aiichiro Nakano, Priya Vashishta, Rajiv K. Kalia

Abstract: Forecasting oil production from oilfields with multiple wells is an important problem in petroleum and geothermal energy extraction, as well as energy storage technologies. The accuracy of oil forecasts is a critical determinant of economic projections, hydrocarbon reserves estimation, construction of fluid processing facilities, and energy price fluctuations. Leveraging generative AI techniques,… ▽ More Forecasting oil production from oilfields with multiple wells is an important problem in petroleum and geothermal energy extraction, as well as energy storage technologies. The accuracy of oil forecasts is a critical determinant of economic projections, hydrocarbon reserves estimation, construction of fluid processing facilities, and energy price fluctuations. Leveraging generative AI techniques, we model time series forecasting of oil and water productions across four multi-well sites spanning four decades. Our goal is to effectively model uncertainties and make precise forecasts to inform decision-making processes at the field scale. We utilize an autoregressive model known as TimeGrad and a variant of a transformer architecture named Informer, tailored specifically for forecasting long sequence time series data. Predictions from both TimeGrad and Informer closely align with the ground truth data. The overall performance of the Informer stands out, demonstrating greater efficiency compared to TimeGrad in forecasting oil production rates across all sites. △ Less

Submitted 24 September, 2024; originally announced September 2024.

Allegro-Legato: Scalable, Fast, and Robust Neural-Network Quantum Molecular Dynamics via Sharpness-Aware Minimization

Authors: Hikaru Ibayashi, Taufeq Mohammed Razakh, Liqiu Yang, Thomas Linker, Marco Olguin, Shinnosuke Hattori, Ye Luo, Rajiv K. Kalia, Aiichiro Nakano, Ken-ichi Nomura, Priya Vashishta

Abstract: Neural-network quantum molecular dynamics (NNQMD) simulations based on machine learning are revolutionizing atomistic simulations of materials by providing quantum-mechanical accuracy but orders-of-magnitude faster, illustrated by ACM Gordon Bell prize (2020) and finalist (2021). State-of-the-art (SOTA) NNQMD model founded on group theory featuring rotational equivariance and local descriptors has… ▽ More Neural-network quantum molecular dynamics (NNQMD) simulations based on machine learning are revolutionizing atomistic simulations of materials by providing quantum-mechanical accuracy but orders-of-magnitude faster, illustrated by ACM Gordon Bell prize (2020) and finalist (2021). State-of-the-art (SOTA) NNQMD model founded on group theory featuring rotational equivariance and local descriptors has provided much higher accuracy and speed than those models, thus named Allegro (meaning fast). On massively parallel supercomputers, however, it suffers a fidelity-scaling problem, where growing number of unphysical predictions of interatomic forces prohibits simulations involving larger numbers of atoms for longer times. Here, we solve this problem by combining the Allegro model with sharpness aware minimization (SAM) for enhancing the robustness of model through improved smoothness of the loss landscape. The resulting Allegro-Legato (meaning fast and "smooth") model was shown to elongate the time-to-failure $t_\textrm{failure}$, without sacrificing computational speed or accuracy. Specifically, Allegro-Legato exhibits much weaker dependence of timei-to-failure on the problem size, $t_{\textrm{failure}} \propto N^{-0.14}$ ($N$ is the number of atoms) compared to the SOTA Allegro model $\left(t_{\textrm{failure}} \propto N^{-0.29}\right)$, i.e., systematically delayed time-to-failure, thus allowing much larger and longer NNQMD simulations without failure. The model also exhibits excellent computational scalability and GPU acceleration on the Polaris supercomputer at Argonne Leadership Computing Facility. Such scalable, accurate, fast and robust NNQMD models will likely find broad applications in NNQMD simulations on emerging exaflop/s computers, with a specific example of accounting for nuclear quantum effects in the dynamics of ammonia. △ Less

Submitted 14 March, 2023; originally announced March 2023.

Multiscale Graph Neural Networks for Protein Residue Contact Map Prediction

Authors: Kuang Liu, Rajiv K. Kalia, Xinlian Liu, Aiichiro Nakano, Ken-ichi Nomura, Priya Vashishta, Rafael Zamora-Resendizc

Abstract: Machine learning (ML) is revolutionizing protein structural analysis, including an important subproblem of predicting protein residue contact maps, i.e., which amino-acid residues are in close spatial proximity given the amino-acid sequence of a protein. Despite recent progresses in ML-based protein contact prediction, predicting contacts with a wide range of distances (commonly classified into sh… ▽ More Machine learning (ML) is revolutionizing protein structural analysis, including an important subproblem of predicting protein residue contact maps, i.e., which amino-acid residues are in close spatial proximity given the amino-acid sequence of a protein. Despite recent progresses in ML-based protein contact prediction, predicting contacts with a wide range of distances (commonly classified into short-, medium- and long-range contacts) remains a challenge. Here, we propose a multiscale graph neural network (GNN) based approach taking a cue from multiscale physics simulations, in which a standard pipeline involving a recurrent neural network (RNN) is augmented with three GNNs to refine predictive capability for short-, medium- and long-range residue contacts, respectively. Test results on the ProteinNet dataset show improved accuracy for contacts of all ranges using the proposed multiscale RNN+GNN approach over the conventional approach, including the most challenging case of long-range contact prediction. △ Less

Submitted 22 December, 2022; v1 submitted 2 December, 2022; originally announced December 2022.

Cross-Task Consistency Learning Framework for Multi-Task Learning

Authors: Akihiro Nakano, Shi Chen, Kazuyuki Demachi

Abstract: Multi-task learning (MTL) is an active field in deep learning in which we train a model to jointly learn multiple tasks by exploiting relationships between the tasks. It has been shown that MTL helps the model share the learned features between tasks and enhance predictions compared to when learning each task independently. We propose a new learning framework for 2-task MTL problem that uses the p… ▽ More Multi-task learning (MTL) is an active field in deep learning in which we train a model to jointly learn multiple tasks by exploiting relationships between the tasks. It has been shown that MTL helps the model share the learned features between tasks and enhance predictions compared to when learning each task independently. We propose a new learning framework for 2-task MTL problem that uses the predictions of one task as inputs to another network to predict the other task. We define two new loss terms inspired by cycle-consistency loss and contrastive learning, alignment loss and cross-task consistency loss. Both losses are designed to enforce the model to align the predictions of multiple tasks so that the model predicts consistently. We theoretically prove that both losses help the model learn more efficiently and that cross-task consistency loss is better in terms of alignment with the straight-forward predictions. Experimental results also show that our proposed model achieves significant performance on the benchmark Cityscapes and NYU dataset. △ Less

Submitted 28 November, 2021; originally announced November 2021.

Graph Signal Recovery Using Restricted Boltzmann Machines

Authors: Ankith Mohan, Aiichiro Nakano, Emilio Ferrara

Abstract: We propose a model-agnostic pipeline to recover graph signals from an expert system by exploiting the content addressable memory property of restricted Boltzmann machine and the representational ability of a neural network. The proposed pipeline requires the deep neural network that is trained on a downward machine learning task with clean data, data which is free from any form of corruption or in… ▽ More We propose a model-agnostic pipeline to recover graph signals from an expert system by exploiting the content addressable memory property of restricted Boltzmann machine and the representational ability of a neural network. The proposed pipeline requires the deep neural network that is trained on a downward machine learning task with clean data, data which is free from any form of corruption or incompletion. We show that denoising the representations learned by the deep neural networks is usually more effective than denoising the data itself. Although this pipeline can deal with noise in any dataset, it is particularly effective for graph-structured datasets. △ Less

Submitted 20 November, 2020; originally announced November 2020.

Physics-informed Neural-Network Software for Molecular Dynamics Applications

Authors: Taufeq Mohammed Razakh, Beibei Wang, Shane Jackson, Rajiv K. Kalia, Aiichiro Nakano, Ken-ichi Nomura, Priya Vashishta

Abstract: We have developed a novel differential equation solver software called PND based on the physics-informed neural network for molecular dynamics simulators. Based on automatic differentiation technique provided by Pytorch, our software allows users to flexibly implement equation of atom motions, initial and boundary conditions, and conservation laws as loss function to train the network. PND comes w… ▽ More We have developed a novel differential equation solver software called PND based on the physics-informed neural network for molecular dynamics simulators. Based on automatic differentiation technique provided by Pytorch, our software allows users to flexibly implement equation of atom motions, initial and boundary conditions, and conservation laws as loss function to train the network. PND comes with a parallel molecular dynamics (MD) engine in order for users to examine and optimize loss function design, and different conservation laws and boundary conditions, and hyperparameters, thereby accelerate the PINN-based development for molecular applications. △ Less

Submitted 21 November, 2020; v1 submitted 6 November, 2020; originally announced November 2020.

Predictive Synthesis of Quantum Materials by Probabilistic Reinforcement Learning

Authors: Pankaj Rajak, Aravind Krishnamoorthy, Ankit Mishra, Rajiv K. Kalia, Aiichiro Nakano, Priya Vashishta

Abstract: Predictive materials synthesis is the primary bottleneck in realizing new functional and quantum materials. Strategies for synthesis of promising materials are currently identified by time-consuming trial and error approaches and there are no known predictive schemes to design synthesis parameters for new materials. We use reinforcement learning to predict optimal synthesis schedules, i.e. a time-… ▽ More Predictive materials synthesis is the primary bottleneck in realizing new functional and quantum materials. Strategies for synthesis of promising materials are currently identified by time-consuming trial and error approaches and there are no known predictive schemes to design synthesis parameters for new materials. We use reinforcement learning to predict optimal synthesis schedules, i.e. a time-sequence of reaction conditions like temperatures and reactant concentrations, for the synthesis of a prototypical quantum material, semiconducting monolayer MoS$_{2}$, using chemical vapor deposition. The predictive reinforcement leaning agent is coupled to a deep generative model to capture the crystallinity and phase-composition of synthesized MoS$_{2}$ during CVD synthesis as a function of time-dependent synthesis conditions. This model, trained on 10000 computational synthesis simulations, successfully learned threshold temperatures and chemical potentials for the onset of chemical reactions and predicted new synthesis schedules for producing well-sulfidized crystalline and phase-pure MoS$_{2}$, which were validated by computational synthesis simulations. The model can be extended to predict profiles for synthesis of complex structures including multi-phase heterostructures and can also predict long-time behavior of reacting systems, far beyond the domain of the MD simulations used to train the model, making these predictions directly relevant to experimental synthesis. △ Less

Submitted 14 September, 2020; originally announced September 2020.

DNA Sequencing via Quantum Mechanics and Machine Learning

Authors: Henry Yuen, Fuyuki Shimojo, Kevin J. Zhang, Ken-ichi Nomura, Rajiv K. Kalia, Aiichiro Nakano, Priya Vashishta

Abstract: Rapid sequencing of individual human genome is prerequisite to genomic medicine, where diseases will be prevented by preemptive cures. Quantum-mechanical tunneling through single-stranded DNA in a solid-state nanopore has been proposed for rapid DNA sequencing, but unfortunately the tunneling current alone cannot distinguish the four nucleotides due to large fluctuations in molecular conformation… ▽ More Rapid sequencing of individual human genome is prerequisite to genomic medicine, where diseases will be prevented by preemptive cures. Quantum-mechanical tunneling through single-stranded DNA in a solid-state nanopore has been proposed for rapid DNA sequencing, but unfortunately the tunneling current alone cannot distinguish the four nucleotides due to large fluctuations in molecular conformation and solvent. Here, we propose a machine-learning approach applied to the tunneling current-voltage (I-V) characteristic for efficient discrimination between the four nucleotides. We first combine principal component analysis (PCA) and fuzzy c-means (FCM) clustering to learn the "fingerprints" of the electronic density-of-states (DOS) of the four nucleotides, which can be derived from the I-V data. We then apply the hidden Markov model and the Viterbi algorithm to sequence a time series of DOS data (i.e., to solve the sequencing problem). Numerical experiments show that the PCA-FCM approach can classify unlabeled DOS data with 91% accuracy. Furthermore, the classification is found to be robust against moderate levels of noise, i.e., 70% accuracy is retained with a signal-to-noise ratio of 26 dB. The PCA-FCM-Viterbi approach provides a 4-fold increase in accuracy for the sequencing problem compared with PCA alone. In conjunction with recent developments in nanotechnology, this machine-learning method may pave the way to the much-awaited rapid, low-cost genome sequencer. △ Less

Submitted 4 December, 2010; originally announced December 2010.