Re: [Help-glpk] interpretation of 'glp_print_sol' and 'glp_print_ranges'

[Andrew Makhorin]

Hi Robbie,

> I am currently trying to decode the output from
> 'glp_print_sol' (also glpsol --output).  It looks
> like a subset of the output from
> ''glp_print_ranges' (also glpsol --ranges).  Just
> the second call is documented in detail, though.

The output produced by glp_print_sol is similar to the one used in the
IBM mps/360 package and described in the book "Modern Linear
Programming" by Bruce Murtagh. However, I altered it a bit, because glpk
uses auxiliary variables rather than slack/surplus ones. On the other
hand, the output produced by glp_print_ranges is exactly the same as in
mps/360.

> 
> It seems that the 'glp_print_sol' prints status
> "B " and 'glp_print_ranges' prints status "BS" for
> an inactive constraint or basic variable.
> 
> The ranges call is also described here:
> 
>   
> http://en.wikibooks.org/wiki/GLPK/Solution_information#Sensitivity_analysis_report
> 
> I looked at 'glpapi11.c' but could not see a
> compelling reason from these abbreviations not
> being both "BS".  Is there a semantic difference?
> 

No difference. Both "B" and "BS" mean "basic".

> If there is a difference, I will make a note on
> the GLPK wikibook.  If not, should they not both
> print "BS".

Okay. I will change glp_print_sol. (There was an idea to print "B*"
instead of "B " to mark infeasible basic variables.)


Best regards,

Andrew Makhorin