Abstract:Bessel functions are critical in scientific computing for applications such as machine learning, protein structure modeling, and robotics. However, currently, available routines lack precision or fail for certain input ranges, such as when the order $v$ is large, and GPU-specific implementations are limited. We address the precision limitations of current numerical implementations while dramatically improving the runtime. We propose two novel algorithms for computing the logarithm of modified Bessel functions of the first and second kinds by computing intermediate values on a logarithmic scale. Our algorithms are robust and never have issues with underflows or overflows while having relative errors on the order of machine precision, even for inputs where existing libraries fail. In C++/CUDA, our algorithms have median and maximum speedups of 45x and 6150x for GPU and 17x and 3403x for CPU, respectively, over the ranges of inputs and third-party libraries tested. Compared to SciPy, the algorithms have median and maximum speedups of 77x and 300x for GPU and 35x and 98x for CPU, respectively, over the tested inputs. The ability to robustly compute a solution and the low relative errors allow us to fit von Mises-Fisher, vMF, distributions to high-dimensional neural network features. This is, e.g., relevant for uncertainty quantification in metric learning. We obtain image feature data by processing CIFAR10 training images with the convolutional layers of a pre-trained ResNet50. We successfully fit vMF distributions to 2048-, 8192-, and 32768-dimensional image feature data using our algorithms. Our approach provides fast and accurate results while existing implementations in SciPy and mpmath fail to fit successfully. Our approach is readily implementable on GPUs, and we provide a fast open-source implementation alongside this paper.

On the Computation of the Logarithm of the Modified Bessel Function of the Second Kind

Accurate Computation of the Logarithm of Modified Bessel Functions on GPUs

An algorithm for the rapid numerical evaluation of Bessel functions of real orders and arguments

Two integral representations for the logarithm of the Glaisher-Kinkelin constant

Modified Bessel Functions in Analytic Number Theory

Inverse Applications of the Generalized Littlewood Theorem Concerning Integrals of the Logarithm of Analytic Functions

Logarithmic corrections for near-extremal black holes

A New Formula for the Natural Logarithm of a Natural Number

A note on the asymptotics of the modified Bessel functions on the Stokes lines

Approximations of the modified Bessel functions of the second kind $K_ν$. Applications in random field generation

[O III]/[N II] as an Abundance Indicator at High Redshift

Series representations for the logarithm of the Glaisher-Kinkelin constant

Linear Forms in Logarithms of Rational Numbers

Non-global Logarithms at 3 Loops, 4 Loops, 5 Loops and Beyond

Logarithms of Catalan generating functions: A combinatorial approach

On the Numerical Computation of the Mittag–Leffler Function

A note on trigonometric approximations of Bessel functions of the first kind and trigonometric power sums

On the Numerical Evaluation of Loop Integrals With Mellin-Barnes Representations

A novel method with error analysis for the numerical solution of a logarithmic singular Fredholm integral equation

Representations of the Glaisher-Kinkelin constant deduced from integrals due to Féaux and Kummer