population op• Data kinds: matrix × matrix → table
• Call: import fullseye as fs; fs.ledger.graph_strength_growth(a, b, hub_fraction=0.1) (to call the implementation directly, import graphinv; graphinv.graph_strength_growth(a, b, hub_fraction=0.1); from the registry, opsgraph.get("graph_strength_growth"))
Where the new synapses went — one wiring matrix `a (earlier) against b` (later).
Both are weight matrices (synapse counts) on the same node order; self-loops are
dropped. With `S = sum of each matrix and dS = S_b - S_a`:
• per node: `in_a, out_a, in_b, out_b (strengths), degree_a` (number of
partners, in + out, presence `> 0), gain_in = in_b - in_a, gain_out`.
Identity: `gain_in.sum() == gain_out.sum() == dS` (every synapse has one pre and one post).
• `strengthened / weakened (weight change on edges present in both), added`
(weight of edges only in `b), lost (weight of edges only in a`);
`strengthened - weakened + added - lost == dS`.
• `rho_in / rho_out — Spearman rank correlation of degree_a with gain_in` /
`gain_out over nodes with degree_a > 0` (0.0 with fewer than 3 such nodes or no
variance); `n_ranked`.
• hubs = the top `hub_fraction of ranked nodes by degree_a` (at least 1):
`hub_share_in_a (their share of S_a as post-synaptic partners), hub_share_gain_in`
(their share of `dS` arriving as inputs), and the same for outputs. If new synapses were
spread in proportion to existing strength, the two shares would be equal — the gap is
what "hubs grow disproportionately" means in numbers.
Raises `ValueError`: either matrix not square / non-finite / negative / a string or
masked array; shapes differ; `hub_fraction outside (0, 1]; dS == 0` is allowed
(shares of the gain are then 0).
• Sample-data catalog (download URLs / licences) — 2-D uses skimage.data (BSD/public domain) plus synthetic images; 3-D lists download URLs for real data sources (Stanford, PDS, …).
• Operator provenance and references — the sources of the research/methods this op family came from.
• The canonical algorithm (author, year) and its uses are named in the family usage guide above.
• poc_worm_synapses_vs_neurites — py -3.11 examples/poc_worm_synapses_vs_neurites.py
table as input)graph_edge_consensus · graph_core_persistence · tree_morphometry · tree_sholl · tree_run_length
population)graph_edge_consensus · graph_core_persistence
*Provenance: graphinv.py — GRAPH operator registry. This per-op note is generated by tools/opdocs.py md (do not hand-edit).*
© 2026 Kazufumi Furuse — Fullseye operator documentation. Licensed under Apache-2.0.