Comparing machine learning and interpolation methods for loop-level calculations

Abstract

The need to approximate functions is ubiquitous in science, either due to empirical constraints or high computational cost of accessing the function. In high-energy physics, the precise computation of the scattering cross-section of a process requires the evaluation of computationally intensive integrals. A wide variety of methods in machine learning have been used to tackle this problem, but often the motivation of using one method over another is lacking. Comparing these methods is typically highly dependent on the problem at hand, so we specify to the case where we can evaluate the function a large number of times, after which quick and accurate evaluation can take place. We consider four interpolation and three machine learning techniques and compare their performance on three toy functions, the four-point scalar Passarino-Veltman D_0D0 function, and the two-loop self-energy master integral M. We find that in low dimensions (d = 3), traditional interpolation techniques like the Radial Basis Function perform very well, but in higher dimensions (d=5, 6, 9) we find that multi-layer perceptrons (a.k.a neural networks) do not suffer as much from the curse of dimensionality and provide the fastest and most accurate predictions.

Status of the ClaRa Library: Detailed Transient Simulation of Complex Energy Systems

“…This combination offers new perspectives for system simulation that go beyond 'traditional' characteristic fields (feed forward neural networks) or response surfaces (recurrent neural networks): complex correlations inside data can be directly derived from the data and encoded by the network parameters. Especially for high dimensional data fields this can be much more memory efficient than by using conventional interpolation methods (Chahrour and Wells 2022). Moreover neural networks can be applied in order to increase the level of detail offered by system models at low performance costs, e.g.…”

Calomplification — the power of generative calorimeter models

“…GANplification arises, intuitively, from the fact that neural networks work like classical parametric fits [4,5], and they are particularly effective when we want to interpolate in many dimensions. This feature is behind the success of the NNPDF parton densities [6] as the first mainstream ML-application in particle theory.…”