The final version of the FUMILIM minimization package

Abstract

The suggested package FUMILIM, based on the famous FUMILI minimization package, has the following advantages: multi-optional user interface; speed advantage when the number of parameters is high enough; there are options to ignore wrong experimental points and correct experimental errors. The preliminary scan is envisaged for complicated tasks.The next version of FUMILIM is capable of working efficiently with multidimensional experimental points, described by a vector function.For heavy user’s functions the parallel fit is envisaged by means of OpenMP service.The package contains popular intrinsic user’s functions. All of them can be used without the definition of the parameter initial values.The program of plane track reconstruction via drift chambers and straws, which can be combined with single detectors, is also included in the package. The capacity of this program is about of 0.5⋅106 tracks per second (at 2.8 GHz).In the final version a number of new intrinsic and extrinsic subroutines are incorporated into the package, and some of the existing procedures have been upgraded. The list of acceptable formats of experimental points has been expanded. The user interface is revised, which is coordinated with a number of users. Expanded the ability to control the program via common blocks.All programs are written in FORTRAN-90.The investigation has been performed at the Veksler and Baldin Laboratory of High Energy Physics, JINR. Program summaryProgram Title: FUMILIM v3Program Files doi:http://dx.doi.org/10.17632/xphc8ctxr3.2Licensing provisions: GPLv3Programming language: Fortran-90Journal reference of previous version: Comput. Phys. Comm. 209 (2016) 199Does 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 new intrinsic and extrinsic programs are added to the package.•Now the program can work 2-column experimental data arrays.•The program call has changed because the dimension of experimental points is now determined automatically.•User’s interface is improved.•A number of subsidiary programs is included to facilitate the creation of a file of initial parameter values for multi-set tasks.Nature of problem: To minimize χ2-functional.Solution method: Least square method.