The GCG Detectors
During the detection, the included detectors will try to find a suitable decomposition for your problem. For more information on how they are called, please visit the page on Detection Process.
List of Detectors
Here you find a list of all detectors available in GCG. Each of those fulfills at least one function out of
- propagating (assigning variables/constraints to a block or the master)
- finishing (assigning all remaining variables/constraints to a block or the master)
- postprocessing (reassigning all remaining variables/constraints to a block or the master to get a better finished decomposition).
For an overview of which detector does what of those three core functionalities, please check the Detector Functionality Overview.
Detectors based on constraint or variable classes:
Detectors finding staircase structures:
Detectors finding inter- and intra-constraint connections:
- Connected Detector (base)
- Connected Detector (nonewlinkingvars)
- Constraint Neighborhood Detector
- Constraint Density Detector
Detectors finding set structures (covering, partitioning and packing):
- Set Covering Detector
- General Set Covering Detector
- Set Packing Detector
- General Set Packing Detector
- Set Partitioning Detector
- General Set Partitioning Detector
Detectors finding arrowhead and single-bordered structures (require hMETIS):
- Hypergraph Partitioning Detector (Columns)
- Hypergraph Partitioning Detector (Rows)
- Hypergraph Partitioning Detector (Columns & Rows)
Detectors performing clustering (require GSL):
Symmetry detectors (require bliss):
Detectors used for detection algorithmics:
Adding own Detectors
If you want to write your own detector, i.e. define how you want to use the information given by the classifiers to decompose your problem, please consult the "How to" for that.