Can an assignment cover just one variable?

Yes. It is partial if the problem contains additional unassigned variables.

Is every complete assignment a solution?

No. It must satisfy all required constraints as well as use permitted values.

Can a partial assignment have no immediate conflict yet still fail?

Yes. It may have no extension that satisfies the full problem.

AI Glossary / Graphs and symbolic AI / Beginner

Assignment

Definition

An assignment gives values to variables. In a constraint-satisfaction problem, it selects a value from the domain of each variable it covers. A partial assignment covers only some variables, while a complete assignment covers all of them. An assignment is a solution only when it also satisfies the problem's required constraints.

What it means

A variable represents something that can take different values, such as a room, time slot, or route choice. Its domain lists the permitted values. An assignment records a particular selection, for example Room = East and Time = Morning.

This is a foundational idea in symbolic AI: solving a problem often means finding assignments that make all the relevant conditions hold. A search procedure can build a solution by extending a partial assignment step by step.

The term also appears in programming and logic. Here the main focus is assigning values in constraint reasoning, not assigning work to a person or executing a programming-language update statement. The shared idea is a correspondence between variables and values.

How it works

Define the variables and their domains before constructing an assignment. A room variable might have domain {East, West}, while a slot variable has domain {Morning, Afternoon}. Assigning East to the slot variable would violate its domain even before other constraints are considered.

Next, check constraints among the assigned values. A complete assignment can still be invalid if, for example, it places two meetings in the same room at the same time.

A partial assignment can be locally consistent yet impossible to complete. Checking only constraints whose relevant values are already fixed may not reveal every future conflict. Constraint propagation and search can expose additional restrictions or dead ends.

Extending an assignment means adding compatible variable-value choices. It does not mean arbitrarily replacing a previous value while claiming the same partial assignment was preserved. Search algorithms may revise earlier choices, but that is a different step that should be tracked explicitly.

Three practical examples

1. A timetable solver assigns each session a room and slot, then checks conflicts and required resources.

2. A configuration system assigns values to component choices, accepting only combinations that meet compatibility constraints.

3. A propositional reasoning tool assigns true or false to proposition variables and tests whether a formula is satisfied by that truth assignment.

Think of it this way

Think of filling out a form whose fields have permitted choices and rules connecting them. Filling every field makes the form complete, but it does not necessarily make the combination acceptable.

The analogy has limits: a formal assignment is a mathematical mapping, and its validity depends on precisely stated domains and constraints. An ordinary form may allow missing information or informal interpretation that a solver cannot safely infer.

Worked example

Consider an original problem with variables Room in {East, West}, Slot in {Morning, Afternoon}, and Projector in {Yes, No}. There are initially 2 × 2 × 2 = 8 complete domain-respecting assignments.

The constraints require Projector = Yes, say that only East has a projector, and forbid using East in the morning. Projector = Yes fixes one variable. The equipment condition then requires Room = East. The availability condition forces Slot = Afternoon.

Exactly one solution remains: {Room = East, Slot = Afternoon, Projector = Yes}.

The partial assignment {Slot = Morning} does not itself violate any domain. However, it cannot be extended to a solution because the other requirements ultimately force East and forbid East in the morning. The example shows the difference between a permissible value, a complete assignment, and a globally satisfying solution.

If the projector requirement were removed, the solution set would change. An assignment cannot be evaluated meaningfully without the current constraints.

Illustrative case study

Fictional case: Elm Training builds a session planner and marks every filled schedule as complete. Reviewers discover that some complete schedules violate room-equipment requirements.

The team introduces separate status labels for partial, complete, and constraint-satisfying assignments. It also records which requirement rejects a candidate. This invented case illustrates clearer reasoning and communication; it reports no measured scheduling improvement or deployed system performance.

Common misconceptions and mistakes

Complete does not mean correct. A complete assignment may fail one or many constraints.

Partial does not mean useless. Partial assignments are central to search, explanation, and early detection of conflicts. But a partial assignment that has not yet exposed a conflict is not automatically extendable to a solution.

Do not confuse a variable's domain with its assigned value. The domain describes possibilities; the assignment selects a particular value. Also avoid combining incompatible assignments: {X = red} and {X = blue} cannot both specify one ordinary single-valued variable in the same assignment unless the model explicitly changes what X represents.

Business use

The concept supports clear descriptions of planning and configuration outputs. Business users can distinguish a draft that still needs choices, a fully specified candidate, and a verified feasible solution.

When constraints change, a previously valid assignment may need to be checked again. Preserve the data and rule version against which it was validated. This is especially useful when users revise availability, capacity, or preferences during planning. Feasibility and desirability should also be separated: several assignments may satisfy all mandatory constraints, while an objective function or human choice determines which is preferred.

Check your understanding

A problem requires X and Y to differ. Both have domain {1, 2}. Is {X = 1, Y = 1} complete and is it a solution?

Answer: It is complete because both variables have values, but it is not a solution because it violates the inequality constraint.

Related terms

Learn next

Next, enumerate all assignments for two small-domain variables and mark which satisfy a constraint. Then add a third variable and find a partial assignment that cannot be extended to a solution.

Sources

Poole and Mackworth, Variables and Constraints, Artificial Intelligence: Foundations of Computational Agents: https://artint.info/3e/html/ArtInt3e.Ch4.S1.html

Primary definitions read. Editorial scope is constraint and logical reasoning; programming assignment and organizational task allocation are distinguished. Examples and counts are original.

Previous: ASR-complete Next: Assignment problems