Difference between revisions of "PlanWise Performance"
Jump to navigation
Jump to search
(Created page with "In this article we discuss things you can do to minimize computation times in PlanWise. For very large problems (say at least 25000 treatment units) it may take hours, or even...") |
|||
| Line 1: | Line 1: | ||
| + | <!-- | ||
In this article we discuss things you can do to minimize computation times in PlanWise. For very large problems (say at least 25000 treatment units) it may take hours, or even days, to generate alternatives with the treatment program generator (TPG). These alternatives are used as input to the optimzation model. The optimization model involves a so called build step, in which a user-defined modul is interpreted, compiled, populated with data, and transformed to a format that a solver can read. This step can also be very time consuming, while the time for solving the problem can be suprisingly short, depending on the complexity of the model and whether it is an LP or MIP problem. However, this is a general optimization problem issue, not specific for PlanWise. When solved as ordinary LP-problem, solution times is usually very short when using a commercial solver such as Gurobi or Cplex. | In this article we discuss things you can do to minimize computation times in PlanWise. For very large problems (say at least 25000 treatment units) it may take hours, or even days, to generate alternatives with the treatment program generator (TPG). These alternatives are used as input to the optimzation model. The optimization model involves a so called build step, in which a user-defined modul is interpreted, compiled, populated with data, and transformed to a format that a solver can read. This step can also be very time consuming, while the time for solving the problem can be suprisingly short, depending on the complexity of the model and whether it is an LP or MIP problem. However, this is a general optimization problem issue, not specific for PlanWise. When solved as ordinary LP-problem, solution times is usually very short when using a commercial solver such as Gurobi or Cplex. | ||
| Line 9: | Line 10: | ||
For large problems, computation times can be longer than | For large problems, computation times can be longer than | ||
| + | --> | ||