: MapperGraph.add_edge triggers exponential redundant mapperify() rescans via recursive re-entry

MapperGraph.add_edge calls self.mapperify() unconditionally on every edge added:

def add_edge(self, u, v, reset_pos=True): super().add_edge(u, v, reset_pos) self.mapperify()

mapperify() scans every integer level between the graph's min and max function value and, for every edge crossing a level, calls self.subdivide_edge(...) to insert a subdivision vertex there:

def mapperify(self): ... for i in range(n_low, n_high + 1): e_list = [e for e in self.edges() if self.f[e[0]] < i and self.f[e[1]] > i] for e in e_list: w_name = self._get_next_mapperify_vert_name() self.subdivide_edge(*e, w_name, i)

subdivide_edge (defined on the parent ReebGraph class) internally calls self.add_edge(u, w) and self.add_edge(w, v) to reconnect the new subdivision vertex. Because self is a MapperGraph instance, these calls dispatch back through the overridden MapperGraph.add_edge — which calls self.mapperify() again, re-scanning the entire graph from scratch for every single subdivision vertex inserted. Each nested mapperify() call can itself trigger more subdivisions, each of which recurses again.

This means adding a handful of edges to a modest mapper graph triggers an explosive amount of redundant, repeated whole-graph scanning. A single edge spanning 5 integer levels triggers 11 total mapperify() calls (should need 1).

The problem is demostrated in the notebook and a fix is suggested.