arXiv subjects

Alan Edelman

Publications and source records attributed to Alan Edelman.

At least 37 records · Page 2Linked to original sources

arXiv2025

Efficient Symbolic Computation via Hash Consing

Symbolic computation systems suffer from memory inefficiencies due to redundant storage of structurally identical subexpressions, commonly known as expression swell, which degrades performance in both classical computer algebra and emerging AI-driven mathematical reasoning tools. In this paper, we present the first integration of hash consing into JuliaSymbolics, a high-performance symbolic toolkit in Julia, by employing a global weak-reference hash table that canonicalizes expressions and eliminates duplication. This approach reduces memory consumption and accelerates key operations such as differentiation, simplification, and code generation, while seamlessly integrating with Julia's metaprogramming and just-in-time compilation infrastructure. Benchmark evaluations across different computational domains reveal substantial improvements: symbolic computations are accelerated by up to 3.2 times, memory usage is reduced by up to 2 times, code generation is up to 5 times faster, function compilation up to 10 times faster, and numerical evaluation up to 100 times faster for larger models. While certain workloads with fewer duplicate unknown-variable expressions show more modest gains or even slight overhead in initial computation stages, downstream processing consistently benefits significantly. These findings underscore the importance of hash consing in scaling symbolic computation and pave the way for future work integrating hash consing with e-graphs for enhanced equivalence-aware expression sharing in AI-driven pipelines.

cs.PL↗

arXiv2025

Performant Unified GPU Kernels for Portable Singular Value Computation Across Hardware and Precision

This paper presents a portable, GPU-accelerated implementation of a QR-based singular value computation algorithm in Julia. The singular value ecomposition (SVD) is a fundamental numerical tool in scientific computing and machine learning, providing optimal low-rank matrix approximations. Its importance has increased even more in large-scale machine learning pipelines, including large language models (LLMs), where it enables low-rank adaptation (LoRA). The implemented algorithm is based on the classic two-stage QR reduction, consisting of successive matrix reduction to band form and bidiagonal form. Our implementation leverages Julia's multiple dispatch and metaprogramming capabilities, integrating with the GPUArrays and KernelAbstractions frameworks to provide a unified type and hardware-agnostic function. It supports diverse GPU architectures and data types, and is, to our knowledge, the first GPU-accelerated singular value implementation to support Apple Metal GPUs and half precision. Performance results on multiple GPU backends and data types demonstrate that portability does not require sacrificing performance: the unified function outperforms most linear algebra libraries (MAGMA, SLATE, rocSOLVER, oneMKL) for matrix sizes larger than 1024x1024, and achieves 80%-90% of the performance of cuSOLVER for large matrices.

cs.DC↗

arXiv2025

Semi-Explicit Neural DAEs: Learning Long-Horizon Dynamical Systems with Algebraic Constraints

Despite the promise of scientific machine learning (SciML) in combining data-driven techniques with mechanistic modeling, existing approaches for incorporating hard constraints in neural differential equations (NDEs) face significant limitations. Scalability issues and poor numerical properties prevent these neural models from being used for modeling physical systems with complicated conservation laws. We propose Manifold-Projected Neural ODEs (PNODEs), a method that explicitly enforces algebraic constraints by projecting each ODE step onto the constraint manifold. This framework arises naturally from semi-explicit differential-algebraic equations (DAEs), and includes both a robust iterative variant and a fast approximation requiring a single Jacobian factorization. We further demonstrate that prior works on relaxation methods are special cases of our approach. PNODEs consistently outperform baselines across six benchmark problems achieving a mean constraint violation error below $10^{-10}$. Additionally, PNODEs consistently achieve lower runtime compared to other methods for a given level of error tolerance. These results show that constraint projection offers a simple strategy for learning physically consistent long-horizon dynamics.

cs.LG↗

arXiv2025

NonlinearSolve.jl: High-Performance and Robust Solvers for Systems of Nonlinear Equations in Julia

Efficiently solving nonlinear equations underpins numerous scientific and engineering disciplines, yet scaling these solutions for challenging system models remains a challenge. This paper presents NonlinearSolve.jl -- a suite of high-performance open-source nonlinear equation solvers implemented natively in the Julia programming language. NonlinearSolve.jl distinguishes itself by offering a unified API that accommodates a diverse range of solver specifications alongside features such as automatic algorithm selection based on runtime analysis, support for GPU-accelerated computation through static array kernels, and the utilization of sparse automatic differentiation and Jacobian-free Krylov methods for large-scale problem-solving. Through rigorous comparison with established tools such as PETSc SNES, Sundials KINSOL, and MINPACK, NonlinearSolve.jl demonstrates robustness and efficiency, achieving significant advancements in solving nonlinear equations while being implemented in a high-level programming language. The capabilities of NonlinearSolve.jl unlock new potentials in modeling and simulation across various domains, making it a valuable addition to the computational toolkit of researchers and practitioners alike.

math.NA↗

arXiv2025

Toward Portable GPU Performance: Julia Recursive Implementation of TRMM and TRSM

This paper presents a performant and portable recursive implementation of triangular matrix-matrix multiplication (TRMM) and triangular solve (TRSM) in Julia for GPUs, two kernels that underlie many linear-algebra algorithms. We restructure TRMM and TRSM so that most work is executed as general matrix-matrix multiplication (GEMM), improving use of the GPU memory hierarchy and reducing latency. Exploiting Julia's multiple dispatch and metaprogramming together with the GPUArrays and KernelAbstractions frameworks, we expose a single hardware-agnostic API that runs on NVIDIA, AMD, and Apple Silicon GPUs. For large matrices the recursive code reaches throughput comparable to vendor libraries such as cuBLAS and rocBLAS, while providing these routines on Apple Silicon for the first time. The entire implementation is only a few hundred lines of code, showing that unified Julia programs can deliver near-vendor performance across heterogeneous architectures.

cs.MS↗

arXiv2025

On a perturbation analysis of Higham squared maximum Gaussian elimination growth matrices

Gaussian elimination is the most popular technique for solving a dense linear system. Large errors in this procedure can occur in floating point arithmetic when the matrix's growth factor is large. In the study of numerical linear algebra, it is often valuable to study and characterize the worst case examples. To this end, in their 1989 paper, Higham and Higham characterized the complete set of real n by n matrices that achieves the maximum growth factor under partial pivoting. Left undone is a sensitivity analysis for these matrices under perturbations. The growth factor of these and nearby matrices is the subject of this work. Through theoretical insights and empirical results, we illustrate the high sensitivity of the growth factor of these matrices to perturbations and show how subtle changes can be strategically applied to matrix entries to significantly reduce the growth.

math.NA↗

arXiv2025

A New Upper Bound For the Growth Factor in Gaussian Elimination with Complete Pivoting

The growth factor in Gaussian elimination measures how large the entries of an LU factorization can be relative to the entries of the original matrix. It is a key parameter in error estimates, and one of the most fundamental topics in numerical analysis. We produce an upper bound of $n^{0.2079 \ln n +0.91}$ for the growth factor in Gaussian elimination with complete pivoting -- the first improvement upon Wilkinson's original 1961 bound of $2 \, n ^{0.25\ln n +0.5}$.

math.NA↗

arXiv2025

Scalable higher-order nonlinear solvers via higher-order automatic differentiation

This paper demonstrates new methods and implementations of nonlinear solvers with higher-order of convergence, which is achieved by efficiently computing higher-order derivatives. Instead of computing full derivatives, which could be expensive, we compute directional derivatives with Taylor-mode automatic differentiation. We first implement Householder's method with arbitrary order for one variable, and investigate the trade-off between computational cost and convergence order. We find that the second-order variant, i.e., Halley's method, to be the most valuable, and further generalize Halley's method to systems of nonlinear equations and demonstrate that it can scale efficiently to large-scale problems. We further apply Halley's method on solving large-scale ill-conditioned nonlinear problems, as well as solving nonlinear equations inside stiff ODE solvers, and demonstrate that it could outperform Newton's method.

math.NA↗

arXiv2025

Stochastic Optimal Control via Local Occupation Measures

Viewing stochastic processes through the lens of occupation measures has proved to be a powerful angle of attack for the theoretical and computational analysis of stochastic optimal control problems. We present a simple modification of the traditional occupation measure framework derived from resolving the occupation measures locally on a partition of the control problem's space-time domain. This notion of local occupation measures provides fine-grained control over the construction of structured semidefinite programming relaxations for a rich class of stochastic optimal control problems with embedded diffusion and jump processes via the moment-sum-of-squares hierarchy. As such, it bridges the gap between discretization-based approximations to the Hamilton-Jacobi-Bellmann equations and occupation measure relaxations. We demonstrate with examples that this approach enables the computation of high quality bounds for the optimal value of a large class of stochastic optimal control problems with significant performance gains relative to the traditional occupation measure framework.

math.OC↗

arXiv2025

Matrix Calculus (for Machine Learning and Beyond)

This course, intended for undergraduates familiar with elementary calculus and linear algebra, introduces the extension of differential calculus to functions on more general vector spaces, such as functions that take as input a matrix and return a matrix inverse or factorization, derivatives of ODE solutions, and even stochastic derivatives of random functions. It emphasizes practical computational applications, such as large-scale optimization and machine learning, where derivatives must be re-imagined in order to be propagated through complicated calculations. The class also discusses efficiency concerns leading to "adjoint" or "reverse-mode" differentiation (a.k.a. "backpropagation"), and gives a gentle introduction to modern automatic differentiation (AD) techniques.

math.HO↗

arXiv2024

Performance Bounds for Quantum Feedback Control

The limits of quantum feedback control have immediate consequences for quantum information science at large, yet remain largely unexplored. Here, we combine quantum filtering theory and moment-sum-of-squares techniques to construct a hierarchy of convex optimization problems that furnish monotonically improving, computable bounds on the best attainable performance for a broad class of quantum feedback control problems. These bounds may serve as witnesses of fundamental limitations, optimality certificates, or performance targets. We prove convergence of the bounds to the optimal control performance under technical conditions and demonstrate the practical utility of our approach by designing certifiably near-optimal controllers for a qubit in a cavity subjected to photon counting and homodyne detection measurements.

quant-ph↗

arXiv2024

Oceananigans.jl: A Julia library that achieves breakthrough resolution, memory and energy efficiency in global ocean simulations

Climate models must simulate hundreds of future scenarios for hundreds of years at coarse resolutions, and a handful of high-resolution decadal simulations to resolve localized extreme events. Using Oceananigans.jl, written from scratch in Julia, we report several achievements: First, a global ocean simulation with breakthrough horizontal resolution -- 488m -- reaching 15 simulated days per day (0.04 simulated years per day; SYPD). Second, Oceananigans simulates the global ocean at 488m with breakthrough memory efficiency on just 768 Nvidia A100 GPUs, a fraction of the resources available on current and upcoming exascale supercomputers. Third, and arguably most significant for climate modeling, Oceananigans achieves breakthrough energy efficiency reaching 0.95 SYPD at 1.7 km on 576 A100s and 9.9 SYPD at 10 km on 68 A100s -- the latter representing the highest horizontal resolutions employed by current IPCC-class ocean models. Routine climate simulations with 10 km ocean components are within reach.

physics.ao-ph↗

arXiv2024

Convex Network Flows

We introduce a general framework for flow problems over hypergraphs. In our problem formulation, which we call the convex flow problem, we have a concave utility function for the net flow at every node and a concave utility function for each edge flow. The objective is to maximize the sum of these utilities, subject to constraints on the flows allowed at each edge, which we only assume to be a convex set. This framework not only includes many classic problems in network optimization, such as max flow, min-cost flow, and multi-commodity flows, but also generalizes these problems to allow, for example, concave edge gain functions. In addition, our framework includes applications spanning a number of fields: optimal power flow over lossy networks, routing and resource allocation in ad-hoc wireless networks, Arrow-Debreu Nash bargaining, and order routing through financial exchanges, among others. We show that the convex flow problem has a dual with a number of interesting interpretations, and that this dual decomposes over the edges of the hypergraph. Using this decomposition, we propose a fast solution algorithm that parallelizes over the edges and admits a clean problem interface. We provide an open source implementation of this algorithm in the Julia programming language, which we show is significantly faster than the state-of-the-art commercial convex solver Mosek.

math.OC↗

arXiv2024

Some New Results on the Maximum Growth Factor in Gaussian Elimination

This paper combines modern numerical computation with theoretical results to improve our understanding of the growth factor problem for Gaussian elimination. On the computational side we obtain lower bounds for the maximum growth for complete pivoting for $n=1:75$ and $n=100$ using the Julia JuMP optimization package. At $n=100$ we obtain a growth factor bigger than $3n$. The numerical evidence suggests that the maximum growth factor is bigger than $n$ if and only if $n \ge 11$. We also present a number of theoretical results. We show that the maximum growth factor over matrices with entries restricted to a subset of the reals is nearly equal to the maximum growth factor over all real matrices. We also show that the growth factors under floating point arithmetic and exact arithmetic are nearly identical. Finally, through numerical search, and stability and extrapolation results, we provide improved lower bounds for the maximum growth factor. Specifically, we find that the largest growth factor is bigger than $1.0045n$ for $n>10$, and the lim sup of the ratio with $n$ is greater than or equal to $3.317$. In contrast to the old conjecture that growth might never be bigger than $n$, it seems likely that the maximum growth divided by $n$ goes to infinity as $n \rightarrow \infty$.

math.NA↗

arXiv2023

Automated Translation and Accelerated Solving of Differential Equations on Multiple GPU Platforms

We demonstrate a high-performance vendor-agnostic method for massively parallel solving of ensembles of ordinary differential equations (ODEs) and stochastic differential equations (SDEs) on GPUs. The method is integrated with a widely used differential equation solver library in a high-level language (Julia's DifferentialEquations.jl) and enables GPU acceleration without requiring code changes by the user. Our approach achieves state-of-the-art performance compared to hand-optimized CUDA-C++ kernels while performing 20--100$\times$ faster than the vectorizing map (vmap) approach implemented in JAX and PyTorch. Performance evaluation on NVIDIA, AMD, Intel, and Apple GPUs demonstrates performance portability and vendor-agnosticism. We show composability with MPI to enable distributed multi-GPU workflows. The implemented solvers are fully featured -- supporting event handling, automatic differentiation, and incorporation of datasets via the GPU's texture memory -- allowing scientists to take advantage of GPU acceleration on all major current architectures without changing their model code and without loss of performance. We distribute the software as an open-source library https://github.com/SciML/DiffEqGPU.jl

cs.DC↗

arXiv2023

The conditional DPP approach to random matrix distributions

We present the conditional determinantal point process (DPP) approach to obtain new (mostly Fredholm determinantal) expressions for various eigenvalue statistics in random matrix theory. It is well-known that many (especially $β=2$) eigenvalue $n$-point correlation functions are given in terms of $n\times n$ determinants, i.e., they are continuous DPPs. We exploit a derived kernel of the conditional DPP which gives the $n$-point correlation function conditioned on the event of some eigenvalues already existing at fixed locations. Using such kernels we obtain new determinantal expressions for the joint densities of the $k$ largest eigenvalues, probability density functions of the $k^\text{th}$ largest eigenvalue, density of the first eigenvalue spacing, and more. Our formulae are highly amenable to numerical computations and we provide various numerical experiments. Several numerical values that required hours of computing time could now be computed in seconds with our expressions, which proves the effectiveness of our approach. We also demonstrate that our technique can be applied to an efficient sampling of DR paths of the Aztec diamond domino tiling. Further extending the conditional DPP sampling technique, we sample Airy processes from the extended Airy kernel. Additionally we propose a sampling method for non-Hermitian projection DPPs.

math-ph↗

arXiv2023

Mapping out phase diagrams with generative classifiers

One of the central tasks in many-body physics is the determination of phase diagrams. However, mapping out a phase diagram generally requires a great deal of human intuition and understanding. To automate this process, one can frame it as a classification task. Typically, classification problems are tackled using discriminative classifiers that explicitly model the probability of the labels for a given sample. Here we show that phase-classification problems are naturally suitable to be solved using generative classifiers based on probabilistic models of the measurement statistics underlying the physical system. Such a generative approach benefits from modeling concepts native to the realm of statistical and quantum physics, as well as recent advances in machine learning. This leads to a powerful framework for the autonomous determination of phase diagrams with little to no human supervision that we showcase in applications to classical equilibrium systems and quantum ground states.

quant-ph↗

arXiv2023

Backpropagation through Back Substitution with a Backslash

We present a linear algebra formulation of backpropagation which allows the calculation of gradients by using a generically written ``backslash'' or Gaussian elimination on triangular systems of equations. Generally, the matrix elements are operators. This paper has three contributions: (i) it is of intellectual value to replace traditional treatments of automatic differentiation with a (left acting) operator theoretic, graph-based approach; (ii) operators can be readily placed in matrices in software in programming languages such as Julia as an implementation option; (iii) we introduce a novel notation, ``transpose dot'' operator ``$\{\}^{T_\bullet}$'' that allows for the reversal of operators. We further demonstrate the elegance of the operators approach in a suitable programming language consisting of generic linear algebra operators such as Julia \cite{bezanson2017julia}, and that it is possible to realize this abstraction in code. Our implementation shows how generic linear algebra can allow operators as elements of matrices. In contrast to ``operator overloading,'' where backslash would normally have to be rewritten to take advantage of operators, with ``generic programming'' there is no such need.

math.NA↗