Verified Aug 18, 20264 min read

Requirements, preferences, and objectives

Choose the correct rule type, scope, target, weight, and penalty function.

Choose the rule category

Category	Result behavior	Use it for
Requirement	Every returned schedule must satisfy it	A rule whose violation makes the schedule unusable
Preference	A miss is allowed and adds weighted cost	A target that may trade off against other targets
Objective	Optimizes a global whole-schedule quality measure	Contact coverage and repeat avoidance

A large preference weight does not make that preference a requirement. It only increases its cost relative to other weighted terms.

Requirements

Requirement	Measurement
Fixed Placements	Every selected person who participates in a scoped session is assigned to the Target Group.
Keep Together	All selected people participating in a scoped session share one group. With three or more people, this is one all-together requirement.
Keep Apart	Every selected pair is separated in a scoped session when both people participate. Three selected people create three pairwise separations.

Each form requires people; Fixed Placements also requires a group. All Sessions includes future sessions. Selecting session numbers creates a fixed current-session list.

If requirements conflict or exceed capacity, GroupMixer cannot return a valid schedule. Change the people, group, session scope, or capacity named by the error.

Preferences

Preference	Measurement
Prefer Together	Scores every selected pair for being together in the selected sessions.
Prefer Apart	Scores every selected pair for being apart in the selected sessions.
Balance Attributes	Compares the count of each selected attribute value with a desired distribution for each scoped group and session. Exact penalizes shortages and excesses; At least penalizes shortages only.
Attribute Totals	Sums a numeric person attribute in each scoped group and session, then compares it with an exact, minimum, maximum, or range target. Missing or nonnumeric values contribute 0.
Encounters	Counts sessions in which all selected people share one group, then compares that count with the configured minimum and maximum.
Group Visits	Counts how often scoped people are assigned to scoped groups. See Group Visits.

Prefer Together, Prefer Apart, and Keep Apart use every pair among the selected people. Encounters and Keep Together instead evaluate the selected set as a whole.

Understand scope

On scope controls for People, Groups, and Sessions:

All means all current and future items.
Selecting items creates an explicit fixed list.
Selecting every item individually is not the same as All; later items will not be added.
An empty explicit selection cannot be saved where the preference requires a scope.

The summary on each saved row shows the resulting scope.

Understand targets and cost

Minimum and maximum fields define the target range. A value inside the range has zero raw deviation. A value below the minimum has a shortfall; a value above the maximum has an excess.

Where the form offers a penalty function, raw deviation d and weight w are scored as:

Penalty function	Cost for one measurement
Linear	`w × d`
Squared	`w × d²`

With weight 10 and a deviation of 2, Linear costs 20 and Squared costs 40. The solver sums the measurements produced by that preference.

Weights are relative. Change one weight, target, or scope at a time, generate again, and inspect the corresponding compliance evidence. Where a form allows weight 0, the preference adds no optimization cost, but Results may still show its raw deviations.

Configure global objectives

Maximize Contacts rewards distinct participant pairs meeting at least once. Its weight competes with weighted preferences but never overrides requirements. Set the weight to 0 to deactivate it.

Avoid Repeats penalizes pair meetings above the configured global threshold. Linear cost adds the same amount for each excess meeting; Squared cost rises faster when repeats concentrate on the same pair. Its weight is relative to Maximize Contacts and the scoped preferences.

Avoid Repeats and Maximize Contacts are different: Avoid Repeats penalizes meetings above a pair threshold, while Maximize Contacts rewards broader unique contact coverage.

After generation, use Score Breakdown and the pair-meeting views to inspect objective performance. Expand Constraint Compliance for required-rule status and scoped preference deviations. See Understand results.

Open in GroupMixer

These links use registered product destinations. Result-specific links remain unavailable until a saved result is selected.

Fixed PlacementsFix selected people to a specific group in the chosen sessions.Keep TogetherRequire selected people to remain in the same group.Keep ApartRequire selected people to remain in different groups.Prefer TogetherPrefer selected people to stay together without making it mandatory.Avoid RepeatsPenalize pair meetings above a global threshold across sessions.

Scenario EditorEnter the model, rules, preferences, and objective used to generate groups.Group VisitsSet and verify how often people are assigned to selected groups.Run and improve a solveGenerate a result, configure advanced solving, save progress, or continue from history.Understand resultsVerify scores, constraint evidence, pair meetings, assignments, and exports.Guide: Random groups vs balanced groups vs constrained groupsUse the least complicated method that can produce an acceptable result. Random grouping needs the least data; balance adds a composition preference; constraints add rules whose violation would make the assignment unusable.