Search

This paper explores two condensed-space interior-point methods to efficiently solve large-scale nonlinear programs on graphics processing units (GPUs). The interior-point method solves a sequence of symmetric indefinite linear systems, or Karush-Kuhn-Tucker (KKT) systems, which become increasingly ill-conditioned as we approach the solution. Solving a KKT system with traditional sparse factorization methods involve numerical pivoting, making parallelization difficult. A solution is to condense the KKT system into a symmetric positive-definite matrix and solve it with a Cholesky factorization, stable without pivoting. Although condensed KKT systems are more prone to ill-conditioning than the original ones, they exhibit structured ill-conditioning that mitigates the loss of accuracy. This paper compares the benefits of two recent condensed-space interior-point methods, HyKKT and LiftedKKT. We implement the two methods on GPUs using MadNLP.jl, an optimization solver interfaced with the NVIDIA sparse linear solver cuDSS and with the GPU-accelerated modeler ExaModels.jl. Our experiments on the PGLIB and the COPS benchmarks reveal that GPUs can attain up to a tenfold speed increase compared to CPUs when solving large-scale instances.

This paper demonstrates the scalability of open-source GPU-accelerated nonlinear programming (NLP) frameworks -- ExaModels.jl and MadNLP.jl -- for solving multi-period alternating current (AC) optimal power flow (OPF) problems on GPUs with high memory capacities (e.g., NVIDIA GH200 with 480 GB of unified memory). There has been a growing interest in solving multi-period AC OPF problems, as the increasingly fluctuating electricity market requires operation planning over multiple periods. These problems, formerly deemed intractable, are now becoming technologically feasible to solve thanks to the advent of high-memory GPU hardware and accelerated NLP tools. This study evaluates the capability of these tools to tackle previously unsolvable multi-period AC OPF instances. Our numerical experiments, run on an NVIDIA GH200, demonstrate that we can solve a multi-period OPF instance with more than 10 million variables up to $10^{-4}$ precision in less than 10 minutes. These results demonstrate the efficacy of the GPU-accelerated NLP frameworks for the solution of extreme-scale multi-period OPF. We provide ExaModelsPower.jl, an open-source modeling tool for multi-period AC OPF models for GPUs.

Development of small pixels for high resolution image sensors implies a lot of challenges. A high level of performance should be guaranteed whereas the overall size must be reduced and so the degree of freedom in design and process. One key parameter of this constant improvement is the knowledge and the control of the crosstalk between pixels. In this paper, we present an advance in crosstalk characterization method based on the design of specific color patterns and the measurement of quantum efficiency. In a first part, we describe the color patterns designed to isolate one pixel or to simulate un-patterned colored pixels. These patterns have been implemented on test-chip and characterized. The second part deals with the characterization setup for quantum efficiency. Indeed, the use of spectral measurements allows us to discriminate pixels based on the color filter placed on top of them and to probe the crosstalk as a function of the depth in silicon, thanks to the photon absorption length variation with the wavelength. In the last part, results are presented showing the impact of color filters patterning, i.e. pixels in a Bayer pattern versus un-patterned pixels. The crosstalk directions and amplitudes are also analyzed in relation to pixel layout.

Sequential quadratic programming (SQP) is widely used in solving nonlinear optimization problem, with advantages of warm-starting solutions, as well as finding high-accurate solution and converging quadratically using second-order information, such as the Hessian matrix. In this study we develop a scalable SQP algorithm for solving the alternate current optimal power flow problem (ACOPF), leveraging the parallel computing capabilities of graphics processing units (GPUs). Our methodology incorporates the alternating direction method of multipliers (ADMM) to initialize and decompose the quadratic programming subproblems within each SQP iteration into independent small subproblems for each electric grid component. We have implemented the proposed SQP algorithm using our portable Julia package ExaAdmm.jl, which solves the ADMM subproblems in parallel on all major GPU architectures. For numerical experiments, we compared three solution approaches: (i) the SQP algorithm with a GPU-based ADMM subproblem solver, (ii) a CPU-based ADMM solver, and (iii) the QP solver Ipopt (the state-of-the-art interior point solver) and observed that for larger instances our GPU-based SQP solver efficiently leverages the GPU many-core architecture, dramatically reducing the solution time.

The B phase of superfluid 3He can be cooled into the pure superfluid regime, where the thermal quasiparticle density is negligible. The bulk superfluid is surrounded by a quantum well at the boundaries of the container, confining a sea of quasiparticles with energies below that of those in the bulk. We can create a non-equilibrium distribution of these states within the quantum well and observe the dynamics of their motion indirectly. Here we show that the induced quasiparticle currents flow diffusively in the two-dimensional system. Combining this with a direct measurement of energy conservation, we conclude that the bulk superfluid 3He is effectively surrounded by an independent two-dimensional superfluid, which is isolated from the bulk superfluid but which readily interacts with mechanical probes. Our work shows that this two-dimensional quantum condensate and the dynamics of the surface bound states are experimentally accessible, opening the possibility of engineering two-dimensional quantum condensates of arbitrary topology., Comment: 11 pages, 9 figures

We investigate how to port the standard interior-point method to new exascale architectures for block-structured nonlinear programs with state equations. Computationally, we decompose the interior-point algorithm into two successive operations: the evaluation of the derivatives and the solution of the associated Karush-Kuhn-Tucker (KKT) linear system. Our method accelerates both operations using two levels of parallelism. First, we distribute the computations on multiple processes using coarse parallelism. Second, each process uses a SIMD/GPU accelerator locally to accelerate the operations using fine-grained parallelism. The KKT system is reduced by eliminating the inequalities and the state variables from the corresponding equations, to a dense matrix encoding the sensitivities of the problem's degrees of freedom, drastically minimizing the memory exchange. We demonstrate the method's capability on the supercomputer Polaris, a testbed for the future exascale Aurora system. Each node is equipped with four GPUs, a setup amenable to our two-level approach. Our experiments on the stochastic optimal power flow problem show that the method can achieve a 50x speed-up compared to the state-of-the-art method., Comment: 23 pages, 8 figures

We present the working prototype of a levitation system designed for investigation of flows in cryogenic helium fluids. The current device allows the levitation of a superconducting sphere and has several provisions made for allowing precise control over its motion. We report on progress in the detection and control systems of the prototype and demonstrate how uniform circular motion can be implemented., Comment: 13 pages, 7 figures

The interior-point method (IPM) has become the workhorse method for nonlinear programming. The performance of IPM is directly related to the linear solver employed to factorize the Karush--Kuhn--Tucker (KKT) system at each iteration of the algorithm. When solving large-scale nonlinear problems, state-of-the art IPM solvers rely on efficient sparse linear solvers to solve the KKT system. Instead, we propose a novel reduced-space IPM algorithm that condenses the KKT system into a dense matrix whose size is proportional to the number of degrees of freedom in the problem. Depending on where the reduction occurs we derive two variants of the reduced-space method: linearize-then-reduce and reduce-then-linearize. We adapt their workflow so that the vast majority of computations are accelerated on GPUs. We provide extensive numerical results on the optimal power flow problem, comparing our GPU-accelerated reduced space IPM with Knitro and a hybrid full space IPM algorithm. By evaluating the derivatives on the GPU and solving the KKT system on the CPU, the hybrid solution is already significantly faster than the CPU-only solutions. The two reduced-space algorithms go one step further by solving the KKT system entirely on the GPU. As expected, the performance of the two reduction algorithms depends intrinsically on the number of available degrees of freedom: their performance is poor when the problem has many degrees of freedom, but the two algorithms are up to 3 times faster than Knitro as soon as the relative number of degrees of freedom becomes smaller., Comment: 21 pages, 3 figures, 2 tables

We describe a new type of torsional oscillator, suitable for studies of quantum fluids at frequencies of $\sim$ $100$ Hz, but capable of reaching high velocities of up to several cm\,s$^{-1}$. This requires the oscillator amplitude to exceed $100$ $\mu m$, which is much too large for a conventional capacitor-driven device. We describe the new geometry for the oscillator, discuss its design, and report our initial tests of its performance., Comment: 7 pages, 7 figures,first autuor deceased October 30th 2019

To target challenges in differentiable optimization we analyze and propose strategies for derivatives of the Mat\'ern kernel with respect to the smoothness parameter. This problem is of high interest in Gaussian processes modelling due to the lack of robust derivatives of the modified Bessel function of second kind with respect to order. In the current work we focus on newly identified series expansions for the modified Bessel function of second kind valid for complex orders. Using these expansions we obtain highly accurate results using the complex step method. Furthermore, we show that the evaluations using the recommended expansions are also more efficient than finite differences.

This paper presents an efficient method for extracting the second-order sensitivities from a system of implicit nonlinear equations on upcoming graphical processing units (GPU) dominated computer systems. We design a custom automatic differentiation (AutoDiff) backend that targets highly parallel architectures by extracting the second-order information in batch. When the nonlinear equations are associated to a reduced space optimization problem, we leverage the parallel reverse-mode accumulation in a batched adjoint-adjoint algorithm to compute efficiently the reduced Hessian of the problem. We apply the method to extract the reduced Hessian associated to the balance equations of a power network, and show on the largest instances that a parallel GPU implementation is 30 times faster than a sequential CPU reference based on UMFPACK., Comment: SIAM-PP22

The Mat\'ern covariance function is ubiquitous in the application of Gaussian processes to spatial statistics and beyond. Perhaps the most important reason for this is that the smoothness parameter $\nu$ gives complete control over the mean-square differentiability of the process, which has significant implications for the behavior of estimated quantities such as interpolants and forecasts. Unfortunately, derivatives of the Mat\'ern covariance function with respect to $\nu$ require derivatives of the modified second-kind Bessel function $\mathcal{K}_\nu$ with respect to $\nu$. While closed form expressions of these derivatives do exist, they are prohibitively difficult and expensive to compute. For this reason, many software packages require fixing $\nu$ as opposed to estimating it, and all existing software packages that attempt to offer the functionality of estimating $\nu$ use finite difference estimates for $\partial_\nu \mathcal{K}_\nu$. In this work, we introduce a new implementation of $\mathcal{K}_\nu$ that has been designed to provide derivatives via automatic differentiation (AD), and whose resulting derivatives are significantly faster and more accurate than those computed using finite differences. We provide comprehensive testing for both speed and accuracy and show that our AD solution can be used to build accurate Hessian matrices for second-order maximum likelihood estimation in settings where Hessians built with finite difference approximations completely fail.

Motivated by the increasing availability of high-performance parallel computing, we design a distributed parallel algorithm for linearly-coupled block-structured nonconvex constrained optimization problems. Our algorithm performs Jacobi-type proximal updates of the augmented Lagrangian function, requiring only local solutions of separable block nonlinear programming (NLP) problems. We provide a cheap and explicitly computable Lyapunov function that allows us to establish global and local sublinear convergence of our algorithm, its iteration complexity, as well as simple, practical and theoretically convergent rules for automatically tuning its parameters. This in contrast to existing algorithms for nonconvex constrained optimization based on the alternating direction method of multipliers that rely on at least one of the following: Gauss-Seidel or sequential updates, global solutions of NLP problems, non-computable Lyapunov functions, and hand-tuning of parameters. Numerical experiments showcase its advantages for large-scale problems, including the multi-period optimization of a 9000-bus AC optimal power flow test case over 168 time periods, solved on the Summit supercomputer using an open-source Julia code.

We propose a novel feasible-path algorithm to solve the optimal power flow (OPF) problem for real-time use cases. The method augments the seminal work of Dommel and Tinney with second-order derivatives to work directly in the reduced space induced by the power flow equations. In the reduced space, the optimization problem includes only inequality constraints corresponding to the operational constraints. While the reduced formulation directly enforces the physical constraints, the operational constraints are softly enforced through Augmented Lagrangian penalty terms. In contrast to interior-point algorithms (state-of-the art for solving OPF), our algorithm maintains feasibility at each iteration, which makes it suitable for real-time application. By exploiting accelerator hardware (Graphic Processing Units) to compute the reduced Hessian, we show that the second-order method is numerically tractable and is effective to solve both static and real-time OPF problems., Comment: 7 pages, 3 figures, conference proceeding

Understanding the temperature dependence of thermal boundary resistance, or Kapitza resistance, between liquid helium and sintered metal has posed a problem in low temperature physics for decades. In the ballistic regime of superfluid $^{3}$He-B, we find the Kapitza resistance can be described via scattering of thermal excitations (quasiparticles) with a macroscopic geometric area, rather than the sintered metal's microscopic area. We estimate that a quasiparticle needs on the order of 1000 collisions to successfully thermalise with the sinter. Finally, we find that the Kapitza resistance is approximately doubled with the addition of two mono-layers of solid $^{4}$He on the sinter surface, which we attribute to an extra magnetic channel of heat transfer being closed as the non-magnetic solid $^{4}$He replaces the magnetic solid $^{3}$He., Comment: 7 pages, 4 figures

The ground state of a fermionic condensate is well protected against perturbations in the presence of an isotropic gap. Regions of gap suppression, surfaces and vortex cores which host Andreev-bound states, seemingly lift that strict protection. Here we show that the role of bound states is more subtle: when a macroscopic object moves in superfluid $^3$He at velocities exceeding the Landau critical velocity, little to no bulk pair breaking takes place, while the damping observed originates from the bound states covering the moving object. We identify two separate timescales that govern the bound state dynamics, one of them much longer than theoretically anticipated, and show that the bound states do not interact with bulk excitations., Comment: 10 pages, 4 figures

Superfluids, such as superfluid $^3\mathrm{He}$ and $^4\mathrm{He}$, exhibit a broad range of quantum phenomena and excitations which are unique to these systems. Nanoscale mechanical resonators are sensitive and versatile force detectors with the ability to operate over many orders of magnitude in damping. Using nanomechanical-doubly clamped beams of extremely high quality factors ($Q>10^6$), we probe superfluid $^4\mathrm{He}$ from the superfluid transition temperature down to $\mathrm{mK}$ temperatures at frequencies up to $11.6 \, \mathrm{MHz}$. Our studies show that nanobeam damping is dominated by hydrodynamic viscosity of the normal component of $^4\mathrm{He}$ above $1\,\mathrm{K}$. In the temperature range $0.3-0.8\,\mathrm{K}$, the ballistic quasiparticles (phonons and rotons) determine the beams' behavior. At lower temperatures, damping saturates and is determined either by magnetomotive losses or acoustic emission into helium. It is remarkable that all these distinct regimes can be extracted with just a single device, despite damping changing over six orders of magnitude., Comment: 4 pages, 2 figures. arXiv admin note: text overlap with arXiv:1810.10129

Microelectromechanical (MEMS) and nanoelectromechanical systems (NEMS) are ideal candidates for exploring quantum fluids since they can be manufactured reproducibly, cover the frequency range from hundreds of kilohertz up to gigahertz and usually have very low power dissipation. Their small size offers the possibility of probing the superfluid on scales comparable to, and below, the coherence length. That said, there have been hitherto no successful measurements of NEMS resonators in the liquid phases of helium. Here we report the operation of doubly-clamped aluminum nanobeams in superfluid $^4$He at temperatures spanning the superfluid transition. The devices are shown to be very sensitive detectors of the superfluid density and the normal fluid damping. However, a further and very important outcome of this work is the knowledge that now we have demonstrated that these devices can be successfully operated in superfluid $^4$He, it is straightforward to apply them in superfluid $^3$He which can be routinely cooled to below 100\,$\mu$K. This brings us into the regime where nanomechanical devices operating at a few MHz frequencies may enter their mechanical quantum ground state.

The present work is targeted at performing a strong scaling study of the high-order spectral element fluid dynamics solver Nek5000. Prior studies indicated a recommendable metric for strong scalability from a theoretical viewpoint, which we test here extensively on three parallel machines with different performance characteristics and interconnect networks, namely Mira (IBM Blue Gene/Q), Beskow (Cray XC40) and Titan (Cray XK7). The test cases considered for the simulations correspond to a turbulent flow in a straight pipe at four different friction Reynolds numbers $Re_{\tau}$ = 180, 360, 550 and 1000. Considering the linear model for parallel communication we quantify the machine characteristics in order to better assess the scaling behaviors of the code. Subsequently sampling and profiling tools are used to measure the computation and communication times over a large range of compute cores. We also study the effect of the two coarse grid solvers XXT and AMG on the computational time. Super-linear scaling due to a reduction in cache misses is observed on each computer. The strong scaling limit is attained for roughly 5000 - 10,000 degrees of freedom per core on Mira, 30,000 - 50,0000 on Beskow, with only a small impact of the problem size for both machines, and ranges between 10,000 and 220,000 depending on the problem size on Titan. This work aims at being a reference for Nek5000 users and also serves as a basis for potential issues to address as the community heads towards exascale supercomputers., Comment: 10 pages, 9 figures, Proceedings of the Exascale Applications and Software Conference 2016 (EASC '16, Stockholm)

Coherent condensates appear as emergent phenomena in many systems, sharing the characteristic feature of an energy gap separating the lowest excitations from the condensate ground state. This implies that a scattering object, moving through the system with high enough velocity for the excitation spectrum in the scatter frame to become gapless, can create excitations at no energy cost, initiating the breakdown of the condensate. This limit is the well-known Landau velocity. While, for the neutral Fermionic superfluid 3He-B in the T=0 limit, flow around an oscillating body displays a very clear critical velocity for the onset of dissipation, here we show that for uniform linear motion there is no discontinuity whatsoever in the dissipation as the Landau critical velocity is passed and exceeded. Since the Landau velocity is such a pillar of our understanding of superfluidity, this is a considerable surprise, with implications for the understanding of the dissipative effects of moving objects in all coherent condensate systems., Comment: 25 pages - article and figures plus supplementary methods