Re: [Bug-glpk] Re: [Help-glpk] Problems with TSPSOL

[Andrew Makhorin]

> tspsol prints the correct objective function in the .out-file, but the
> solution provided seems to be the earlier obtained solution.
> 
> lpx_print_sol: writing LP problem solution to `example1_0_1.out'...
> +   262:   ip_obj =   4.630000000e+08 >=     tree is empty (0; 7)
> INTEGER OPTIMAL SOLUTION FOUND
> 
> When I check this solution it has a cost of 465.
> 

Oscar, looks like you sum up edge lengths marked by '*' while you
should use edges whose activity is 1, i.e. those ones which are included
in the tour. Edges marked by '*' correspond to the initial tour, not to
the optimal tour.

For example, edge '9-28' is *not* included in the optimal tour (its
activity is 0), however, it is presented in your .out file:

27 8 13000000