Minos is a solver based on MINOS versions 5.51 that can be used
either "stand-alone" or with AMPL's -ob and -og options to solve
linear and nonlinear problems expressed in AMPL.
To use minos with AMPL, you have several options. You can invoke
it within an AMPL session by saying
solve;
or, if $solver is not already minos,
option solver minos;
solve;
Minos understands AMPL's -ob and -og output formats; you can thus
use stand-alone invocations like
ampl -obfoo foo.mod foo.dat
minos foo [assignments]
With no stub argument, minos tries to read a SPECS file on stdin,
followed (unless the SPECS files dictates otherwise) by an MPS file,
so you also can invoke
cat foo.specs foo.mps | minos [assignment ...]
Invocation, in general, is
minos [options] [stub [-AMPL]] [assignment ...]
where stub is from `ampl -obstub` or `ampl -ogstub`.
Assignments have the form spec_phrase=value or n=filename
(where n is a 1 or 2 digit Fortran unit number, presumably one
mentioned in a spec_phrase=value assignment or the SPECS file).
No spaces may appear in either form of assignment; spec_phrases
are phrases that can appear in a SPECS file, with _ (underscore)
substituted for blank. An assignment n=filename attaches filename
to Fortran unit n. Assignments can also appear in the environment
variable $minos_options; minos first reads the SPECS file (if any),
then $minos_options, then the command-line assignments. For example
minos foo backup_basis_file=2 2=zip new_basis_file=3 3=zap
will make minos behave as though it read a SPECS file containing
BACKUP BASIS FILE 2
NEW BASIS FILE 3
and will connect files zip and zap with the Fortran units 2 and 3.
(It's unfortunate that you must explicitly mention Fortran unit
numbers, but this is not onerous.) The default file name for Fortran
unit u is fort.u .
Specifying objno=n in $minos_options or the command line is the
same as specifying problem_number=n-1 (i.e., objno=2 is the
same as problem_number=1), and objno=0 means "ignore the objectve;
just seek a feasible point".
Similar comments apply to a.out's you create to solve nonlinear
either "stand-alone" or with AMPL's -ob and -og options to solve
linear and nonlinear problems expressed in AMPL.
To use minos with AMPL, you have several options. You can invoke
it within an AMPL session by saying
solve;
or, if $solver is not already minos,
option solver minos;
solve;
Minos understands AMPL's -ob and -og output formats; you can thus
use stand-alone invocations like
ampl -obfoo foo.mod foo.dat
minos foo [assignments]
With no stub argument, minos tries to read a SPECS file on stdin,
followed (unless the SPECS files dictates otherwise) by an MPS file,
so you also can invoke
cat foo.specs foo.mps | minos [assignment ...]
Invocation, in general, is
minos [options] [stub [-AMPL]] [assignment ...]
where stub is from `ampl -obstub` or `ampl -ogstub`.
Assignments have the form spec_phrase=value or n=filename
(where n is a 1 or 2 digit Fortran unit number, presumably one
mentioned in a spec_phrase=value assignment or the SPECS file).
No spaces may appear in either form of assignment; spec_phrases
are phrases that can appear in a SPECS file, with _ (underscore)
substituted for blank. An assignment n=filename attaches filename
to Fortran unit n. Assignments can also appear in the environment
variable $minos_options; minos first reads the SPECS file (if any),
then $minos_options, then the command-line assignments. For example
minos foo backup_basis_file=2 2=zip new_basis_file=3 3=zap
will make minos behave as though it read a SPECS file containing
BACKUP BASIS FILE 2
NEW BASIS FILE 3
and will connect files zip and zap with the Fortran units 2 and 3.
(It's unfortunate that you must explicitly mention Fortran unit
numbers, but this is not onerous.) The default file name for Fortran
unit u is fort.u .
Specifying objno=n in $minos_options or the command line is the
same as specifying problem_number=n-1 (i.e., objno=2 is the
same as problem_number=1), and objno=0 means "ignore the objectve;
just seek a feasible point".
Similar comments apply to a.out's you create to solve nonlinear
