Minos

DQQCleanup(tmpPath, originalDirectory)[source]

Performs cleanup after solving with the DQQ solver, removing temporary solver output and returning to the original working directory

USAGE:

DQQCleanup (tmpPath, originalDirectory)

INPUTS:

  • tmpPath – path to the temporary folder containing the results and MPS subfolders, and the fort.*/*.sol output files, to be deleted

  • originalDirectory – folder to cd back into once cleanup is complete

minosCleanUp(tmpPath, dataDirectory, modelName)[source]

Cleans up after the MINOS solver by deleting temporary run files from tmpPath (other than the spec/run files to keep) and removing the temporary data file for the given model

USAGE:

minosCleanUp (tmpPath, dataDirectory, modelName)

INPUTS:

  • tmpPath – path to the temporary MINOS working folder to clean; every file not in the keep list (lp1.spc, lp2.spc, qrunfba, runfba) is deleted

  • dataDirectory – path to the folder containing the temporary model data file to delete

  • modelName – name of the model, used to build the temporary data file name <modelName>.txt in dataDirectory

readMinosSolution(fname)[source]

Loads MINOS solution information from file fname. The file is created by MINOS with the run-time option Report file 81. Note that 81 is essential, and the gfortran compiler produces a file named fort.81 (which may be renamed by the script used to run MINOS). Other compilers may generate different generic names.

The optimization problem solved by MINOS is assumed to be

  • min osense*s(iobj)

  • st Ax - s = 0 + bounds on x and s,

where A has m rows and n columns.

USAGE:

sol = readMinosSolution (fname)

INPUT:

fname – File

OUTPUT:

sol – Structure

  • .inform - MINOS exit condition

  • .m - Number of rows in A

  • .n - Number of columns in A

  • .osense - osense

  • .objrow - Row of A containing a linear objective

  • .obj - Value of MINOS objective (linear + nonlinear)

  • .numinf - Number of infeasibilities in x and s.

  • .suminf - Sum of infeasibilities in x and s.

  • .xstate - n vector: state of each variable in x.

  • .sstate - m vector: state of each slack in s.

  • .x - n vector: value of each variable in x.

  • .s - m vector: value of each slack in s.

  • .rc - n vector: reduced gradients for x.

  • .y - m vector: dual variables for Ax - s = 0.