You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The IdFragmentAndAP class made sense when denoptim used an index-based approach for identifying building blocks and attachment points.
It should be possible to remove it and replace the related index-based code with reference-based one.
The text was updated successfully, but these errors were encountered:
The presence of IdFragmentAndAP in the FragSpaceExplorer, namely here, prevents the change of the library of fragments in a hypothetical process where graphs are build by a sequence of constrained FSE acting on the compatibility matrix and on the presence of fragments in the fragment's library. In practice, if we build root graphs with lib_frags-version-A that contains a vertex with ID=50 that have 4 APs, then that vertex must be in the same position in any further use of FSE on that root graph, for instane a further enrichment using lib_frags-version-B. In practice any change to the library of vertexes must not affect the previously used vertexes. Failure in adhering to this contract, i.e., failure in retaining at position 50 of any lib_frags-version-B/C/d/... the 4-AP vertex will eventually result lead to java.lang.IndexOutOfBoundsException (from here) at ID 50 now we have a vertex with N<4 attachment points.
The IdFragmentAndAP class made sense when denoptim used an index-based approach for identifying building blocks and attachment points.
It should be possible to remove it and replace the related index-based code with reference-based one.
The text was updated successfully, but these errors were encountered: