Reverse statuses and unavailable variables

Read feasibility and goal achievement separately

A successful explicit-goal result must be interpreted with its goal tolerance and constraints. A closest feasible result satisfies the relevant feasibility conditions but misses the requested target; it is presented as such. A maximize/minimize result is the best feasible candidate found by the bounded search, not proof of a global optimum. No feasible result found can reflect bounds, discrete choices, conflicting constraints or search limitations.

During Solve, progress shows evaluations, iterations and elapsed time; optimization can also show the best feasible result found so far. Larger connected models or broader searches can take longer, and the search may finish before its limit. Cancel stops the search and marks it incomplete. A cancelled search supplies no candidate for Preview or Apply. The actual model remains unchanged; review the setup and run again when needed.

Read binding constraints and reached bounds

The result lists each constraint’s achieved value, comparison and status. Binding means the condition is satisfied at or close to its boundary under the numerical tolerance; a satisfied equality is also marked binding. It does not provide a shadow price or establish which assumption caused the limit. A variable can separately be marked as reaching its lower or upper bound. Review both before widening a range or changing a business condition.

Details report evaluations, iterations, elapsed time, model revision and the stopping reason. A search can find a valid candidate before exhausting its budget. The stopping reason and a satisfied goal describe different aspects of that result.

Keep the identity of a missing source

A configured variable can become unavailable after a line is deleted, a method changes or its context no longer permits that input. The row retains its available human-readable identity and offers Change or Remove. Solve stays blocked while required variables are unavailable. If the exact source becomes eligible again, its saved bounds can be recovered. Old configurations without a stored human-readable name cannot reconstruct the name of a deleted source.

Repair the question before increasing effort

Check source eligibility, units, lower/upper order, integer restrictions and the meaning of each constraint. Inspect the deterministic target and any N/M result first. Increase evaluations or starts only when the question is coherent and additional search is justified. Do not loosen a real business constraint merely to obtain a green status.