Debugging the FUMILIM minimization package

Abstract

The suggested package FUMILIM, based on famous FUMILI minimization package, has the following advantages.Unlimited number of parameters (multi-set tasks).Ability to work with multidimensional experimental points, described by a vector function.The preliminary scan is envisaged for complicated tasks.For heavy user's functions the parallel fit is envisaged with the help of OpenMP service.Multi-optional user interface, including the option to ignore wrong experimental points.The package contains popular intrinsic user's functions. All of them can be used without the definition of the parameter initial values.There is a package of fast track reconstruction programs for working with detectors, including drift chambers. The capacity of these programs is about of 106 tracks per second (at 2.8 GHz).In the corrected version a number of bugs is fixed.The service for multi-set tasks has been essentially expanded.Information support of poorly converged tasks and removing of wrong experimental points is expanded.All programs are written in FORTRAN-90. New version program summaryProgram Title: FUMILIM v4CPC Library link to program files:https://doi.org/10.17632/xphc8ctxr3.3Licensing provisions: GPLv3Programming language: Fortran-90Journal reference of previous version: Comput. Phys. Commun. 251 (2020) 107202Does the new version supersede the previous version?: Yes.Reasons for the new version: To extend the class of solvable tasks, fixing a number of bugs.Summary of revisions:• A number of bugs is fixed.• The service for multi-set tasks has been essentially expanded.• Information support of poorly converged task and removing WP is expanded.• Fast programs are accelerated by ≃15%.• Organization of output options is slightly revised, which provided better comfort for users.• Settings of the package through the redefinition of numbers in common-blocks are replaced by calls to short subroutines. This makes the main program more compact. The old way of setting still works.Nature of problem: To minimize χ2-functional.Solution method: Least square method.