arXiv subjects

Abhishek Das

Publications and source records attributed to Abhishek Das.

At least 37 records · Page 2Linked to original sources

arXiv2023

PIRLNav: Pretraining with Imitation and RL Finetuning for ObjectNav

We study ObjectGoal Navigation -- where a virtual robot situated in a new environment is asked to navigate to an object. Prior work has shown that imitation learning (IL) using behavior cloning (BC) on a dataset of human demonstrations achieves promising results. However, this has limitations -- 1) BC policies generalize poorly to new states, since the training mimics actions not their consequences, and 2) collecting demonstrations is expensive. On the other hand, reinforcement learning (RL) is trivially scalable, but requires careful reward engineering to achieve desirable behavior. We present PIRLNav, a two-stage learning scheme for BC pretraining on human demonstrations followed by RL-finetuning. This leads to a policy that achieves a success rate of $65.0\%$ on ObjectNav ($+5.0\%$ absolute over previous state-of-the-art). Using this BC$\rightarrow$RL training recipe, we present a rigorous empirical analysis of design choices. First, we investigate whether human demonstrations can be replaced with `free' (automatically generated) sources of demonstrations, e.g. shortest paths (SP) or task-agnostic frontier exploration (FE) trajectories. We find that BC$\rightarrow$RL on human demonstrations outperforms BC$\rightarrow$RL on SP and FE trajectories, even when controlled for same BC-pretraining success on train, and even on a subset of val episodes where BC-pretraining success favors the SP or FE policies. Next, we study how RL-finetuning performance scales with the size of the BC pretraining dataset. We find that as we increase the size of BC-pretraining dataset and get to high BC accuracies, improvements from RL-finetuning are smaller, and that $90\%$ of the performance of our best BC$\rightarrow$RL policy can be achieved with less than half the number of BC demonstrations. Finally, we analyze failure modes of our ObjectNav policies, and present guidelines for further improving them.

cs.LG↗

arXiv2023

The Open Catalyst 2022 (OC22) Dataset and Challenges for Oxide Electrocatalysts

The development of machine learning models for electrocatalysts requires a broad set of training data to enable their use across a wide variety of materials. One class of materials that currently lacks sufficient training data is oxides, which are critical for the development of OER catalysts. To address this, we developed the OC22 dataset, consisting of 62,331 DFT relaxations (~9,854,504 single point calculations) across a range of oxide materials, coverages, and adsorbates. We define generalized total energy tasks that enable property prediction beyond adsorption energies; we test baseline performance of several graph neural networks; and we provide pre-defined dataset splits to establish clear benchmarks for future efforts. In the most general task, GemNet-OC sees a ~36% improvement in energy predictions when combining the chemically dissimilar OC20 and OC22 datasets via fine-tuning. Similarly, we achieved a ~19% improvement in total energy predictions on OC20 and a ~9% improvement in force predictions in OC22 when using joint training. We demonstrate the practical utility of a top performing model by capturing literature adsorption energies and important OER scaling relationships. We expect OC22 to provide an important benchmark for models seeking to incorporate intricate long-range electrostatic and magnetic interactions in oxide surfaces. Dataset and baseline models are open sourced, and a public leaderboard is available to encourage continued community developments on the total energy tasks and data.

cond-mat.mtrl-sci↗

arXiv2022

Explicit structure of the vanishing viscosity limits with initial data consisting of $δ$-distributions starting from two point sources

In this article, we consider the one-dimensional zero-pressure gas dynamics system \[ u_t + \left( {u^2}/{2} \right)_x = 0,\ ρ_t + (ρu)_x = 0 \] in the upper-half plane with a linear combination of two $δ$-distributions \[ u|_{t=0} = u_a\ δ_{x=a} + u_b\ δ_{x=b},\ ρ|_{t=0} = ρ_c\ δ_{x=c} + ρ_d\ δ_{x=d} \] as initial data. Here $a$, $b$, $c$, $d$ are distinct points on the real line ordered as $a < c < b < d$. Our objective is to provide a detailed analysis of the structure of the vanishing viscosity limits of this system utilizing the corresponding modified adhesion model \[ u^ε_t + \left({(u^ε)^2}/{2} \right)_x =\fracε{2} u^ε_{xx},\ ρ^ε_t + (ρ^εu^ε)_x = \fracε{2} ρ^ε_{xx}. \] For this purpose, we extensively use the various asymptotic properties of the function erfc$: z \longmapsto \int_{z}^{\infty} e^{-s^2}\ ds$ along with suitable Hopf-Cole transformations.

math.AP↗

arXiv2022

Explicit structure of the vanishing viscosity limits for the zero-pressure gas dynamics system initiated by the linear combination of a characteristic function and a $δ$-distribution

In this article, we consider the one-dimensional zero-pressure gas dynamics system \[ u_t + \left( {u^2}/{2} \right)_x = 0,\ ρ_t + (ρu)_x = 0 \] in the upper-half plane with a linear combination of a characteristic function and a $δ$-measure \[ u|_{t=0} = u_a\ χ_{ {}_{ \left( -\infty , a \right) } } + u_b\ δ_{x=b},\ ρ|_{t=0} = ρ_c\ χ_{ {}_{ \left( -\infty , c \right) } } + ρ_d\ δ_{x=d} \] as initial data, where $a$, $b$, $c$, $d$ are distinct points on the real line ordered as $a < c < b < d$, and provide a detailed analysis of the vanishing viscosity limits for the above system utilizing the corresponding modified adhesion model \[ u^ε_t + \left({(u^ε)^2}/{2} \right)_x =\fracε{2} u^ε_{xx},\ ρ^ε_t + (ρ^εu^ε)_x = \fracε{2} ρ^ε_{xx}. \] For this purpose, we use suitable Hopf-Cole transformations and various asymptotic properties of the function erfc$: z \longmapsto \int_{z}^{\infty} e^{-s^2}\ ds$.

math.AP↗

arXiv2022

Spherical Channels for Modeling Atomic Interactions

Modeling the energy and forces of atomic systems is a fundamental problem in computational chemistry with the potential to help address many of the world's most pressing problems, including those related to energy scarcity and climate change. These calculations are traditionally performed using Density Functional Theory, which is computationally very expensive. Machine learning has the potential to dramatically improve the efficiency of these calculations from days or hours to seconds. We propose the Spherical Channel Network (SCN) to model atomic energies and forces. The SCN is a graph neural network where nodes represent atoms and edges their neighboring atoms. The atom embeddings are a set of spherical functions, called spherical channels, represented using spherical harmonics. We demonstrate, that by rotating the embeddings based on the 3D edge orientation, more information may be utilized while maintaining the rotational equivariance of the messages. While equivariance is a desirable property, we find that by relaxing this constraint in both message passing and aggregation, improved accuracy may be achieved. We demonstrate state-of-the-art results on the large-scale Open Catalyst dataset in both energy and force prediction for numerous tasks and metrics.

physics.chem-ph↗

arXiv2022

GemNet-OC: Developing Graph Neural Networks for Large and Diverse Molecular Simulation Datasets

Recent years have seen the advent of molecular simulation datasets that are orders of magnitude larger and more diverse. These new datasets differ substantially in four aspects of complexity: 1. Chemical diversity (number of different elements), 2. system size (number of atoms per sample), 3. dataset size (number of data samples), and 4. domain shift (similarity of the training and test set). Despite these large differences, benchmarks on small and narrow datasets remain the predominant method of demonstrating progress in graph neural networks (GNNs) for molecular simulation, likely due to cheaper training compute requirements. This raises the question -- does GNN progress on small and narrow datasets translate to these more complex datasets? This work investigates this question by first developing the GemNet-OC model based on the large Open Catalyst 2020 (OC20) dataset. GemNet-OC outperforms the previous state-of-the-art on OC20 by 16% while reducing training time by a factor of 10. We then compare the impact of 18 model components and hyperparameter choices on performance in multiple datasets. We find that the resulting model would be drastically different depending on the dataset used for making model choices. To isolate the source of this discrepancy we study six subsets of the OC20 dataset that individually test each of the above-mentioned four dataset aspects. We find that results on the OC-2M subset correlate well with the full OC20 dataset while being substantially cheaper to train on. Our findings challenge the common practice of developing GNNs solely on small datasets, but highlight ways of achieving fast development cycles and generalizable results via moderately-sized, representative datasets such as OC-2M and efficient models such as GemNet-OC. Our code and pretrained model weights are open-sourced.

cs.LG↗

arXiv2022

Nonlinear mode-tide coupling in coalescing binary neutron stars with relativistic corrections

We compute the internal modes of a nonspinning neutron star and its tidal metric perturbation in general relativity, and determine the effect of relativistic corrections to the modes on mode coupling. Claims have been made that a new hydrodynamic instability can occur in a neutron star in a binary neutron star system triggered by the nonlinear coupling of the companion's tidal field to pairs of p-modes and g-modes in it as the binary inspirals toward merger. This "PG" instability may be significant since it can influence the binary's inspiral phase by extracting orbital energy, thereby potentially causing large deviations in their gravitational waveforms from those predicted by theoretical models that do not account for it. This can result in incorrect parameter estimation, at best, or mergers going undetected, at worst, owing to the use of deficient waveform models. On the other hand, better modeling of this instability and its effect on binary orbits can unravel a new phenomenon and shed light on stellar instabilities, via gravitational wave observations. So far, all mode-tide coupling instability studies have been formulated in Newtonian perturbation theory. Neutron stars are compact objects, so relativistic corrections might be important. We present and test a new code to calculate the relativistic eigenmodes of nonrotating relativistic stars. We use these relativistic tide and neutron star eigenmodes to compute the mode-tide coupling strength (MTCS) for a few selected equations of state. The MTCS thus calculated can be at most tens of percent different from its purely Newtonian value, but we confirm the dependencies on orbital separation and equation of state found by Newtonian calculations. For some equations of state, the MTCS is very sensitive to the neutron star crust region, demonstrating the importance of treating this region accurately.

gr-qc↗

arXiv2022

Open Challenges in Developing Generalizable Large Scale Machine Learning Models for Catalyst Discovery

The development of machine learned potentials for catalyst discovery has predominantly been focused on very specific chemistries and material compositions. While effective in interpolating between available materials, these approaches struggle to generalize across chemical space. The recent curation of large-scale catalyst datasets has offered the opportunity to build a universal machine learning potential, spanning chemical and composition space. If accomplished, said potential could accelerate the catalyst discovery process across a variety of applications (CO2 reduction, NH3 production, etc.) without additional specialized training efforts that are currently required. The release of the Open Catalyst 2020 (OC20) has begun just that, pushing the heterogeneous catalysis and machine learning communities towards building more accurate and robust models. In this perspective, we discuss some of the challenges and findings of recent developments on OC20. We examine the performance of current models across different materials and adsorbates to identify notably underperforming subsets. We then discuss some of the modeling efforts surrounding energy-conservation, approaches to finding and evaluating the local minima, and augmentation of off-equilibrium data. To complement the community's ongoing developments, we end with an outlook to some of the important challenges that have yet to be thoroughly explored for large-scale catalyst discovery.

physics.chem-ph↗

arXiv2022

Evolution of Curvature in Riemannian Geometry

In this paper shall we endeavour to substantiate that the evolution of the Riemann- Christoffel tensor or curvature tensor can be expressed entirely by an arbitrary timelike vector field and that the curvature tensor returns to its initial value with respect to change in a particular index. This implies that Poincare's recurrence theorem is valid in this cosmological scenario. Also, it has been shown that geodesics can diverge just as they can converge. As is ostensible, this result indicates the existence the of a point of exclusivity - the opposite of a singularity.

physics.gen-ph↗

arXiv2022

Habitat-Web: Learning Embodied Object-Search Strategies from Human Demonstrations at Scale

We present a large-scale study of imitating human demonstrations on tasks that require a virtual robot to search for objects in new environments -- (1) ObjectGoal Navigation (e.g. 'find & go to a chair') and (2) Pick&Place (e.g. 'find mug, pick mug, find counter, place mug on counter'). First, we develop a virtual teleoperation data-collection infrastructure -- connecting Habitat simulator running in a web browser to Amazon Mechanical Turk, allowing remote users to teleoperate virtual robots, safely and at scale. We collect 80k demonstrations for ObjectNav and 12k demonstrations for Pick&Place, which is an order of magnitude larger than existing human demonstration datasets in simulation or on real robots. Second, we attempt to answer the question -- how does large-scale imitation learning (IL) (which hasn't been hitherto possible) compare to reinforcement learning (RL) (which is the status quo)? On ObjectNav, we find that IL (with no bells or whistles) using 70k human demonstrations outperforms RL using 240k agent-gathered trajectories. The IL-trained agent demonstrates efficient object-search behavior -- it peeks into rooms, checks corners for small objects, turns in place to get a panoramic view -- none of these are exhibited as prominently by the RL agent, and to induce these behaviors via RL would require tedious reward engineering. Finally, accuracy vs. training data size plots show promising scaling behavior, suggesting that simply collecting more demonstrations is likely to advance the state of art further. On Pick&Place, the comparison is starker -- IL agents achieve ${\sim}$18% success on episodes with new object-receptacle locations when trained with 9.5k human demonstrations, while RL agents fail to get beyond 0%. Overall, our work provides compelling evidence for investing in large-scale imitation learning. Project page: https://ram81.github.io/projects/habitat-web.

cs.AI↗

arXiv2022

Numerical relativity simulations of prompt collapse mergers: threshold mass and phenomenological constraints on neutron star properties after GW170817

We determine the threshold mass for prompt (no bounce) black hole formation in equal-mass neutron star (NS) mergers using a new set of 227 numerical relativity simulations. We consider 23 phenomenological and microphysical finite temperature equations of state (EOS), including models with hyperons and first-order phase transitions to deconfined quarks. We confirm the existence of EOS-insensitive relations between the threshold mass, the binary tidal parameter at the threshold ($Λ_{th}$), the maximum mass of nonrotating NSs, and the radii of reference mass NSs. We correct the systematic errors in previously reported fitting coefficients that were obtained with approximate general-relativity simulations. We combine the EOS-insensitive relations, phenomenological constraints on NS properties and observational data from GW170817 to derive an improved lower limit on radii of maximum mass and 1.6 M$_\odot$ NS of 9.81 km and 10.90 km, respectively. We also constrain the radius and quadrupolar tidal deformability ($Λ$) of a 1.4 $M_\odot$ NS to be larger than 10.74 km and 172, respectively. We consider uncertainties in all independent parameters -- fitting coefficients as well as GW170817 masses while reporting the range of radii constraints. We introduce new methods to constrain the upper as well as lower limit of NS maximum mass using future BNS detections and their identification as prompt or delayed collapse. With future observations it will be possible to derive even tighter constraints on the properties of matter at and above nuclear density using the method proposed in this work.

astro-ph.HE↗

arXiv2022

Towards Training Billion Parameter Graph Neural Networks for Atomic Simulations

Recent progress in Graph Neural Networks (GNNs) for modeling atomic simulations has the potential to revolutionize catalyst discovery, which is a key step in making progress towards the energy breakthroughs needed to combat climate change. However, the GNNs that have proven most effective for this task are memory intensive as they model higher-order interactions in the graphs such as those between triplets or quadruplets of atoms, making it challenging to scale these models. In this paper, we introduce Graph Parallelism, a method to distribute input graphs across multiple GPUs, enabling us to train very large GNNs with hundreds of millions or billions of parameters. We empirically evaluate our method by scaling up the number of parameters of the recently proposed DimeNet++ and GemNet models by over an order of magnitude. On the large-scale Open Catalyst 2020 (OC20) dataset, these graph-parallelized models lead to relative improvements of 1) 15% on the force MAE metric for the S2EF task and 2) 21% on the AFbT metric for the IS2RS task, establishing new state-of-the-art results.

cs.LG↗

arXiv2022

Zitterbewegung field and the fifth force

In this paper, we investigate the {\it zitterbewegung region} based on a previous paper where we had derived equations for particles and antiparticles in the aforesaid region. Here, we shall introduce a field theoretic approach resorting to the seminal paper of 1954 by Yang-Mills.

physics.gen-ph↗

arXiv2022

NarrationBot and InfoBot: A Hybrid System for Automated Video Description

Video accessibility is crucial for blind and low vision users for equitable engagements in education, employment, and entertainment. Despite the availability of professional and amateur services and tools, most human-generated descriptions are expensive and time consuming. Moreover, the rate of human-generated descriptions cannot match the speed of video production. To overcome the increasing gaps in video accessibility, we developed a hybrid system of two tools to 1) automatically generate descriptions for videos and 2) provide answers or additional descriptions in response to user queries on a video. Results from a mixed-methods study with 26 blind and low vision individuals show that our system significantly improved user comprehension and enjoyment of selected videos when both tools were used in tandem. In addition, participants reported no significant difference in their ability to understand videos when presented with autogenerated descriptions versus human-revised autogenerated descriptions. Our results demonstrate user enthusiasm about the developed system and its promise for providing customized access to videos. We discuss the limitations of the current work and provide recommendations for the future development of automated video description tools.

cs.HC↗

arXiv2021

The Open Catalyst 2020 (OC20) Dataset and Community Challenges

Catalyst discovery and optimization is key to solving many societal and energy challenges including solar fuels synthesis, long-term energy storage, and renewable fertilizer production. Despite considerable effort by the catalysis community to apply machine learning models to the computational catalyst discovery process, it remains an open challenge to build models that can generalize across both elemental compositions of surfaces and adsorbate identity/configurations, perhaps because datasets have been smaller in catalysis than related fields. To address this we developed the OC20 dataset, consisting of 1,281,040 Density Functional Theory (DFT) relaxations (~264,890,000 single point evaluations) across a wide swath of materials, surfaces, and adsorbates (nitrogen, carbon, and oxygen chemistries). We supplemented this dataset with randomly perturbed structures, short timescale molecular dynamics, and electronic structure analyses. The dataset comprises three central tasks indicative of day-to-day catalyst modeling and comes with pre-defined train/validation/test splits to facilitate direct comparisons with future model development efforts. We applied three state-of-the-art graph neural network models (CGCNN, SchNet, Dimenet++) to each of these tasks as baseline demonstrations for the community to build on. In almost every task, no upper limit on model size was identified, suggesting that even larger models are likely to improve on initial results. The dataset and baseline models are both provided as open resources, as well as a public leader board to encourage community contributions to solve these important tasks.

cond-mat.mtrl-sci↗

arXiv2021

Auxiliary Tasks and Exploration Enable ObjectNav

ObjectGoal Navigation (ObjectNav) is an embodied task wherein agents are to navigate to an object instance in an unseen environment. Prior works have shown that end-to-end ObjectNav agents that use vanilla visual and recurrent modules, e.g. a CNN+RNN, perform poorly due to overfitting and sample inefficiency. This has motivated current state-of-the-art methods to mix analytic and learned components and operate on explicit spatial maps of the environment. We instead re-enable a generic learned agent by adding auxiliary learning tasks and an exploration reward. Our agents achieve 24.5% success and 8.1% SPL, a 37% and 8% relative improvement over prior state-of-the-art, respectively, on the Habitat ObjectNav Challenge. From our analysis, we propose that agents will act to simplify their visual inputs so as to smooth their RNN dynamics, and that auxiliary tasks reduce overfitting by minimizing effective RNN dimensionality; i.e. a performant ObjectNav agent that must maintain coherent plans over long horizons does so by learning smooth, low-dimensional recurrent dynamics. Site: https://joel99.github.io/objectnav/

cs.CV↗

arXiv2021

Rotation Invariant Graph Neural Networks using Spin Convolutions

Progress towards the energy breakthroughs needed to combat climate change can be significantly accelerated through the efficient simulation of atomic systems. Simulation techniques based on first principles, such as Density Functional Theory (DFT), are limited in their practical use due to their high computational expense. Machine learning approaches have the potential to approximate DFT in a computationally efficient manner, which could dramatically increase the impact of computational simulations on real-world problems. Approximating DFT poses several challenges. These include accurately modeling the subtle changes in the relative positions and angles between atoms, and enforcing constraints such as rotation invariance or energy conservation. We introduce a novel approach to modeling angular information between sets of neighboring atoms in a graph neural network. Rotation invariance is achieved for the network's edge messages through the use of a per-edge local coordinate frame and a novel spin convolution over the remaining degree of freedom. Two model variants are proposed for the applications of structure relaxation and molecular dynamics. State-of-the-art results are demonstrated on the large-scale Open Catalyst 2020 dataset. Comparisons are also performed on the MD17 and QM9 datasets.

cs.LG↗

arXiv2021

ForceNet: A Graph Neural Network for Large-Scale Quantum Calculations

With massive amounts of atomic simulation data available, there is a huge opportunity to develop fast and accurate machine learning models to approximate expensive physics-based calculations. The key quantity to estimate is atomic forces, where the state-of-the-art Graph Neural Networks (GNNs) explicitly enforce basic physical constraints such as rotation-covariance. However, to strictly satisfy the physical constraints, existing models have to make tradeoffs between computational efficiency and model expressiveness. Here we explore an alternative approach. By not imposing explicit physical constraints, we can flexibly design expressive models while maintaining their computational efficiency. Physical constraints are implicitly imposed by training the models using physics-based data augmentation. To evaluate the approach, we carefully design a scalable and expressive GNN model, ForceNet, and apply it to OC20 (Chanussot et al., 2020), an unprecedentedly-large dataset of quantum physics calculations. Our proposed ForceNet is able to predict atomic forces more accurately than state-of-the-art physics-based GNNs while being faster both in training and inference. Overall, our promising and counter-intuitive results open up an exciting avenue for future research.

cs.LG↗