The figure shows the dialog for the Replay using Decomposition plug-in (DecomposedReplayer 6.5.61).

### Select configuration

Here you can select the decomposition configuration, which handles the decomposition for you. There are seven possible configurations. To explain these configurations, assume that the maximal decomposition would result in *N* decomposed parts, and that the decomposition slider below is set at *D*%.

#### Decompose

Decomposes the net (log) into (*N* × *D*) /100 (rounded down, minimal value is 1) subnets (sublogs).

#### Decompose 100%

Decomposes the net (log) into *N* subnets (sublogs).

#### Decompose 50%

Decomposes the net (log) into *N*/2 (rounded down, minimal value is 1) subnets (sublogs).

#### Decompose 75%

Decomposes the net (log) into (3 × *N*)/4 (rounded down, minimal value is 1) subnets (sublogs).

#### Do not decompose

Uses the net (log) itself as the only subnet (sublog).

#### Hide

Splits the net (log) into (*N* × *D*) /100 (rounded down, minimal value is 1) subnets (sublogs). Every subnet is obtained from the overall net and the decomposed subnet by hiding all transitions in the first that do not occur in the second.

#### Hide and reduce

Splits the net (log) into (*N* × *D*) /100 (rounded down, minimal value is 1) subnets (sublogs). Every subnet is obtained from the overall net and the decomposed subnet by hiding all transitions in the first that do not occur in the second, and by reducing as many hidden transitions as possible.

#### Decomposition percentage

Here you can influence the number of subnets (sublogs) obtained by the **Decompose**, **Hide**, and **Hide and reduce** configurations. This slider is disabled for all other configurations.

### Select classifier

Here you can select the classifier to use. The classifier determines the activity that corresponds to an event. The plug-in assumes that a transition matches an event if and only if the transition label matches the activity perfectly.

#### Move on Log Costs and Move on Model Costs

Here you can set the costs for log moves and visible model moves. The plug-in assumes that the costs of synchronous moves and invisible model moves are both 0.

After having decomposed the net (log) into subnets (sublogs), every sublog will be replayed on the corresponding subnet using the cost-based replayer and the cost structure as defined by you. Afterwards, the resulting subalignments will be merged into a single *pseudo* alignment, where a pseudo alignment is an alignment except for the fact that the transition sequence in a pseudo alignment need not to be executable in the net. In general, it is not possible to merge subalignments into an alignment.