r - Intentional Infeasible Linear Program in Rsymphony -
i using rsymphony
package in r mixed integer program. able iteratively relax 1 of constraints until problem becomes feasible. so, i'd able to:
send milp's constraint matrix, objective function, etc. symphony solver via
rsymphony_solve_lp
if
rsymphony_solve_lp
returns feasible solution (or states there is feasible solution) terminate solve, rebuild constraint matrix tighter constraint, re-send solver viarsymphony_solve_lp
if
rsymphony_solve_lp
says problem infeasible, backtrack last feasible constraint matrix , solve there.
any tips/tricks on how without manually watching solver verbosity on?
a poster on different board gave approach. reformulated milp constraint want iterate, multiplied 'tuning parameter' continuous decision variable. put penalty term associated decision variable objective function relax constraint little possible. effect should constraint relaxed enough feasible, not beyond that.
Comments
Post a Comment