---

Todo List

Member DAG::ComputeEigenSum(const Matrix &adj, int nVals)

Find out the impact of summing all the eigenvalues instead of the $k$ largest ones.5s3l1

Member DAG::DeleteSubDAG(leda_node v)

This function does not support some kinds of DAGs. More specifically, DAGs with loops are not allowed. This must be fixed in the future.

Member DAG::SplitSubGraph(leda_node v, DAGPtr &ptrDag)

The algorithm doesn't work well with DAG because it looses the multiple parents info of nodes in with more than one of their parents belong the the splitted subgraph. Since the TSVs of the subgraph are never recomputed, the missing edges won't affect the matching algorithm as it's currently implemented.

Class SGNode

At this moment, the type of a branch is not 1..4 but instead the set of numbers that the shock graph detection algorithm returns. SHOCK_base=100, SHOCK_2or3=100, SHOCK_1=101, SHOCK_2=102, SHOCK_3=103, SHOCK_4=104, ROOT=105 (values defined in Shock.h).

Member operator>>(istream &is, GGRelation &rel)

these pair of functions should cal read and write, but we still need a version of read that reads in text format for reading the GG from a text file.

Member OccludeDAG(DAGPtr &ptrDag, double dMaxOcclusionRate)

The DeleteSubDAG() function does not work for DAGs with loops.

---