"FeatureVector"
computes some features of
the WolframModel
evolution. For now, it only computes properties associated with the causal graph g
:
In[] := WolframModel[{{x, y}, {x, z}} -> {{x, z}, {x, w}, {y, w}, {z, w}}, {{0, 0}, {0, 0}}, 5]["FeatureVector"]
Out[] = {22, 42, 0, 2, 2, 2, 6, 6}
The list of properties is:
VertexCount
: The number of vertices in the causal graph. Related to the total number of eventsEdgeCount
: The number of edges in the causal graph. Related to the total number of expressionsVertexConnectivity
: The smallest number of vertices whose deletion fromg
disconnectsg
. This is computed on the undirected version of the causal graph.VertexDegree
Quantiles: The quantiles 0, 0.25, 0.50, 0.75, 1 of the vertex degrees distribution.
This property is useful for applying machine learning to Wolfram Models explorations.
inits = Partition[#, 2] & /@ Tuples[ConstantArray[Range[0, 3], 4]];
In[] := FeatureSpacePlot[#["FeatureVector"] -> #[
"CausalGraph"] & /@ (WolframModel[{{x, y}, {x, z}} -> {{x,
z}, {x, w}, {y, w}, {z, w}}, #, 6] &) /@ inits]