CPLEX solve()返回false但状态显示Optimal的原因及疑问

Core Contradiction Breakdown #

• The cplex.solve() method returns false not only for infeasible or unbounded models, but also when invalid model data triggers an error during solve initialization or presolve.
• Even though cplex.getCplexStatus() returns CPX_STAT_OPTIMAL and cplex.getStatus() returns Optimal, this status refers to the trivially empty model left after presolve eliminated all rows and columns—not a valid solved model with variables.

Root Cause: NaN Coefficients #

• Your model contained NaN values in coefficients, which are invalid for CPLEX. During presolve, CPLEX detects these invalid entries and removes all associated rows/columns to "resolve" the invalidity, resulting in an empty model.
• The empty model is technically optimal (no constraints to violate, no objective to optimize), hence the status codes. However, the presence of NaNs is an error condition that causes solve() to return false as a signal that the original model was invalid.

Why getValues() Triggers a Runtime Exception #

• After presolve eliminates all variables, there are no variable values to retrieve. Calling cplex.getValues() in this state attempts to access non-existent data, leading to the runtime error.

Fix Recap #

• Removing the NaN coefficients from your model ensures CPLEX can process a valid, non-empty model. Presolve no longer eliminates all rows/columns, solve() returns true on successful solution, and cplex.getValues() works as expected.

内容的提问来源于stack exchange，提问作者thewoz