Very high accuracy and sensitivity have become attainable by modern instruments for experimental measurements of physical quantities in various scientific fields. Yet it is still impossible to completely eliminate the influence of instrumental effects on the result. The measured values of a physical quantity inevitably differ, sometimes significantly, from the true ones. The question therefore arises of restoring the true distributions from the measured ones, taking the specific features of the experiment and the characteristics of scientific instruments into account. Different approaches are in use based on a mathematical model of the instrument and the formulation of the deconvolution problem. We describe this problem, key ideas and methods for its solution, and features and implementation details using the example of elementary particle physics and space physics experiments.

Keywords: unfolding, deconvolution, spectrum restoration methods, statistical estimates, Bayesian methods, regularization, binning-free methods, machine learning PACS: 02.50.−r, 02.60.−x, 96.50.S−, 96.50.sb (all) DOI: 10.3367/UFNe.2022.05.039189 URL: https://ufn.ru/en/articles/2023/6/d/ 001112624000004 2-s2.0-85163239655 2023PhyU...66..628B Citation: Bogomolov Yu V, Alekseev V V, Levanova O A, Mayorov A G, Malakhov V V "Review of unfolding methods" Phys. Usp. 66 628–642 (2023) BibTexBibNote ® (generic)BibNote ® (RIS)MedlineRefWorks Received: 23rd, January 2022, revised: 7th, April 2022, accepted: 2nd, May 2022 Оригинал: Богомолов Ю В, Алексеев В В, Леванова О А, Майоров А Г, Малахов В В «Обзор методов обратной свёртки» УФН 193 669–685 (2023); DOI: 10.3367/UFNr.2022.05.039189