NVIDIA
  • Santa Clara, CA, United States
Recent publications
The IceCube Neutrino Observatory is a cubic kilometer neutrino detector located at the geographic South Pole designed to detect high-energy astrophysical neutrinos. To thoroughly understand the detected neutrinos and their properties, the detector response to signal and background has to be modeled using Monte Carlo techniques. An integral part of these studies are the optical properties of the ice the observatory is built into. The simulated propagation of individual photons from particles produced by neutrino interactions in the ice can be greatly accelerated using graphics processing units (GPUs). In this paper, we (a collaboration between NVIDIA and IceCube) reduced the propagation time per photon by a factor of up to 3 on the same GPU. We achieved this by porting the OpenCL parts of the program to CUDA and optimizing the performance. This involved careful analysis and multiple changes to the algorithm. We also ported the code to NVIDIA OptiX to handle the collision detection. The hand-tuned CUDA algorithm turned out to be faster than OptiX. It exploits detector geometry and only a small fraction of photons ever travel close to one of the detectors.
Lead (Pb) halide perovskite solar cells (PSCs) have emerged as a highly promising next-generation photovoltaic (PV) technology that combines high device performance with ease of processing and low cost. However, the potential leaching of lead has been recognized as a major environmental concern for their large-scale commercialization, especially for application areas with significant overlap with human life. Herein, we report a quantitative kinetic analysis of the Pb leaching behavior of five types of benchmark PSCs, namely MAPbI3, FA0.95MA0.05Pb(I0.95Br0.05)3, Cs0.05(FA0.85MA0.15)0.95Pb(I0.85Br0.15)3, CsPbI3, and CsPbI2Br, under laboratory rainfall conditions. Strikingly, over 60% of the Pb contained in the unencapsulated perovskite devices was leached within the first 120 s under rainfall exposure, suggesting that very rapid leaching of Pb can occur when indoor and outdoor PV devices are subject to physical damage or failed encapsulation. The initial Pb leaching rate is found to be strongly dependent on the type of PSCs, pointing to a potential route toward Pb leaching reduction through further optimization of their materials design. Our findings offer kinetic insights into the Pb leaching behavior of PSCs upon aqueous exposure, highlighting the urgency to develop robust mitigation methods to avoid a potentially catastrophic impact on the environment for their large-scale deployment.
Boundary element methods produce dense linear systems that can be accelerated via multipole expansions. Solved with Krylov methods, this implies computing the matrix-vector products within each iteration with some error, at an accuracy controlled by the order of the expansion, $p$. We take advantage of a unique property of Krylov iterations that allow lower accuracy of the matrix-vector products as convergence proceeds, and propose a relaxation strategy based on progressively decreasing $p$. Via extensive numerical tests, we show that the relaxed Krylov iterations converge with speed-ups of between 2x and 4x for Laplace problems and between 3.5x and 4.5x for Stokes problems. We include an application to Stokes flow around red blood cells, computing with up to 64 cells and problem size up to 131k boundary elements and nearly 400k unknowns. The study was done with an in-house multi-threaded C++ code, on a quad-core CPU.
Model bias is one of the main obstacles to improved accuracy and reliability in numerical weather prediction conducted with state‐of‐the‐art atmospheric models. To deal with model bias, a modification of the standard four‐dimensional variational (4D‐Var) algorithm, called weak‐constraint 4D‐Var, has been developed where a forcing term is introduced into the model to correct for the bias that accumulates along the model trajectory. This approach reduced the temperature bias in the stratosphere by up to 50% and is implemented in the European Centre for Medium‐Range Weather Forecasts operational forecasting system. Despite different origins and applications, data assimilation (DA) and Deep Learning are both able to learn about the Earth system from observations. In this paper, a deep learning approach for model bias correction is developed using temperature retrievals from radio occultation (RO) measurements. Neural networks (NNs) require a large number of samples to properly capture the relationship between the temperature first‐guess trajectory and the model bias. As running the Integrated Forecasting System (IFS) DA system for extended periods of time with a fixed model version and at realistic resolutions is computationally very expensive, we have chosen to train, the initial NNs are trained using the ERA5 reanalysis before using transfer learning on 1 year of the current IFS model. Preliminary results show that convolutional NNs are adequate to estimate model bias from RO temperature retrievals. The different strengths and weaknesses of both deep learning and weak constraint 4D‐Var are discussed, highlighting the potential for each method to learn model biases effectively and adaptively.
Geomagnetically Induced Currents (GICs) arise from spatio‐temporal changes to Earth's magnetic field, which arise from the interaction of the solar wind with Earth's magnetosphere, and drive catastrophic destruction to our technologically dependent society. Hence, computational models to forecast GICs globally with large forecast horizon, high spatial resolution and temporal cadence are of increasing importance to perform prompt necessary mitigation. Since GIC data is proprietary, the time variability of the horizontal component of the magnetic field perturbation (dB/dt) is used as a proxy for GICs. In this work, we develop a fast, global dB/dt forecasting model, which forecasts 30 min into the future using only solar wind measurements as input. The model summarizes 2 hr of solar wind measurement using a Gated Recurrent Unit and generates forecasts of coefficients that are folded with a spherical harmonic basis to enable global forecasts. When deployed, our model produces results in under a second, and generates global forecasts for horizontal magnetic perturbation components at 1 min cadence. We evaluate our model across models in literature for two specific storms of 5 August 2011 and 17 March 2015, while having a self‐consistent benchmark model set. Our model outperforms, or has consistent performance with state‐of‐the‐practice high time cadence local and low time cadence global models, while also outperforming/having comparable performance with the benchmark models. Such quick inferences at high temporal cadence and arbitrary spatial resolutions may ultimately enable accurate forewarning of dB/dt for any place on Earth, resulting in precautionary measures to be taken in an informed manner.
We propose a machine learning approach for quickly solving Mixed Integer Programs (MIPs) by learning to prioritize sets of branching variables at the root node which result in faster solution times, which we call pseudo-backdoors. Learning-based approaches have seen success in combinatorial optimization by flexibly leveraging common structures in a given distribution of problems. Our approach takes inspiration from the concept of strong backdoors, which are small sets of variables such that only branching on these variables yields an optimal integral solution and a proof of optimality. Our notion of pseudo-backdoors corresponds to a small set of variables such that prioritizing branching on them when possible leads to faster solve time. A key advantage of pseudo-backdoors over strong backdoors is that they retain the solver’s optimality guarantees and are amenable to data-driven identification. Our proposed method learns to estimate the relative solver speed of a candidate pseudo-backdoor and determine whether or not to use it. This pipeline can be used to identify high-quality pseudo-backdoors on unseen MIP instances for a given MIP distribution. We evaluate our method on five problem distributions and find that our approach can efficiently identify high-quality pseudo-backdoors. In addition, we compare our learned approach against Gurobi, a state-of-the-art MIP solver, demonstrating that our method can be used to improve solver performance.
We develop a deep learning approach to extract ray directions at discrete locations by analyzing highly oscillatory wave fields. A deep neural network is trained on a set of local plane-wave fields to predict ray directions at discrete locations. The resulting deep neural network is then applied to a reduced-frequency Helmholtz solution to extract ray directions, which are further incorporated into a ray-based interior-penalty discontinuous Galerkin (IPDG) method to solve the corresponding Helmholtz equations at higher frequencies. In this way, we observe no apparent pollution effects in the resulting Helmholtz solutions in inhomogeneous media. Our 2D and 3D numerical results show that the proposed scheme is very efficient and yields highly accurate solutions.
Physics-informed neural networks allow models to be trained by physical laws described by general nonlinear partial differential equations. However, traditional architectures of this approach struggle to solve more challenging time-dependent problems. In this work, we present a novel physics-informed framework for solving time-dependent partial differential equations. Using only the governing differential equations and problem initial and boundary conditions, we generate a latent representation of the problem’s spatio-temporal dynamics. Our model utilizes discrete cosine transforms to encode spatial frequencies and re-current neural networks to process the time evolution. This efficiently and flexibly produces a compressed representation which is used for additional conditioning of physics-informed models. We show experimental results on the Taylor-Green vortex solution to the Navier-Stokes equations. Our proposed model achieves state-of-the-art performance on the Taylor-Green vortex relative to other physics-informed baseline models.
Let us recall that, from a pragmatic perspective, AI is nothing more or less than a tool for implementing a new leap in automation . This pragmatic perspective on AI is the one that matters in the current world economy, and therefore will necessarily receive primacy for development.
In many respects, machine learning’s current concerns are reminiscent of those which heralded the rejection of GOFAI.
The failure of GOFAI created a vacancy for a new guiding philosophy for AI.
The research field of AI is concerned with devising theories, methods, and workflows for producing software artifacts which behave as intelligent subjects. Evidently, intelligence, as the property of an agent, is not of necessity inherited from the methods used to construct it: that a car has been assembled by robots does not make it a robot. Unfortunately, even this obvious distinction can sometimes be erased in some prominent published work. To wit: the statement, “an agent that performs sufficiently well on a sufficiently wide range of tasks is classified as intelligent” was recently published by DeepMind [273] to give context to a paper claiming to have developed “the first deep RL agent that outperforms the standard human benchmark on all 57 Atari games” [14]. This invites the inference that the range of the tasks (57 games) that have been achieved warrants calling the advertised agent ‘intelligent’. However, careful reading of the paper reveals that the authors have in fact developed 57 different agents. Granted, this was achieved using the same development method and system architecture, but 57 agents were nonetheless trained, rather than the claimed single agent. Here is a prime example of distilled confusion: a property (applicability to 57 tasks) of one construction method (instantiating the Agent57 system architecture) has just been ‘magically’ transferred to some 57 artifacts produced by the method.
In this chapter, semantic closure meets system engineering: we describe how SCL systems can be constructed and controlled in practice, casting a developmental perspective on automation which we call ‘2nd order automation engineering’. Let us first give context to our objective, starting with a quote from Bundy and McNeil [40], who described in 2006 what they considered to be ‘a major goal of artificial intelligence research over the next 50 years’.
Machine learning excels at inducing mappings from data, but struggles to induce causal hierarchies. In contrast, symbolic reasoning (in particular, when considered as an expression language ) can represent any form of domain knowledge and can index into code or data via pattern matching.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
692 members
Stefan Jeschke
  • Department of Research
John Gunnels
  • Mathematical Libraries/Quantum Computing
Information
Address
2788 San Tomas Expressway, 95051, Santa Clara, CA, United States
Website
http://nvidia.com/
Phone
+1 (408) 486-2000