Skip to content

Latest commit

 

History

History
32 lines (23 loc) · 1.89 KB

README.md

File metadata and controls

32 lines (23 loc) · 1.89 KB
Raw

EMSOL

EMSOL Primal simplex LP solver

EMSOL was written 1992-2013 by @jajhall, with rather ad-hoc versioning.

original

This is the last whole version of EMSOL source code.

  • It contains bug fixes for "EMSOL v2.16".
  • It lacks a driver program
  • It won't compile with modern f77 (specifically gcc version 9.4.0 - Ubuntu 9.4.0-1ubuntu1~20.04) due to some FORTRAN usage that's now illegal.

Specifically, the subroutines ems_cp_rl_a and ems_cp_i_a were written to operate in different modes: copying the values in a real (or integer) array; copying in a reversed loop; setting array entries equal to their index; setting all array entries to a fixed value. The second parameter (the origin of the values) was dimensioned in the subroutine as an arrray, but sometimes a scalar value was passed in when setting all array entries to a fixed value.

master

This includes

  • A driver program
  • Edits to allow it to be compiled with modern f77. Specifically the abuse of FORTRAN (above) has been side-stepped by creating special versions of ems_cp_rl_a and ems_cp_i_a for the cases where whole arrays are copied, and the original subroutines modified to deal with the case where an array is initialised to a scalar value.
  • A bug fix in dvx.f that may have been communicated to users
  • Full implementation of code to guard against "division by zero" exceptions in dan_r_d.f, sed_r_s.f, sed_r_d.f, dvx_r_s.f, dvx_r_d.f and dan_r_s.f.

As such, this represents the "best" version of EMSOL that @jajhall can come up with

Note

  • To compile EMSOL may require the-fno-range-check flag to cope with the mechanism used in EMSV.INC to set the 32nd bit in a 32-bit integer
  • EMSOL will only read files in EMS format. Instances for checking are provided
  • No version of EMSOL is supported in any way
  • Anyone using EMSOL is strongly advised to continue using the version that they have!