Finite-Value Constraint¶
Definition¶
A linear expression can be restricted to one of the listed values by introducing binary variables and writing
Interpretation¶
A finite-value constraint lets a linear expression take only a few listed values, such as , which linear inequalities alone cannot express. Exactly one indicator selects the allowed value, so no separate alternative constraint is needed for every possible value.
The red points are the three allowed totals, each produced by one selector: gives , gives and gives , as in the first Example. The light band from to shows what the right-hand side could reach if the selectors were fractions; , for instance, gives .
Each part of the formulation has a role. Because the are binary and sum to , they are mutually exclusive decisions of the exactly-one type: one equals , the others are , and collapses to the single selected value . The values are fixed data, which keeps every term linear.
A common slip is to write instead. Then all selectors may be , and the expression may equal , a value that is not on the list. The parity constraint can be read as a finite-value constraint with listed values and , written more compactly with one auxiliary variable.
Key Results¶
- The sum-to-one equality ensures that a value is actually selected.
- The are fixed data, so is linear.
- Relaxing the indicators to continuous fractions would generally allow values between the listed values.
Examples¶
The light band shows intermediate totals made possible by fractional selector values; the binary selectors permit only the listed totals.
- For allowed totals 3, 5, or 8, write with and .
- An even number of three yes decisions can take only the values 0 and 2.
Prerequisites¶
Related Concepts¶
Sources¶
- OR Modelling · Week 3
Symbols: element of, sum
Last updated: 27 October 2026 11:52
Finite-Value ConstraintBinary VariableMutually Exclusive DecisionsParity Constraint