morphe-metrics 1.0.0

A stateless Python library for morphogenetic computing evaluation

morphe-metrics

A stateless Python library for morphogenetic computing evaluation.

Venkatesh Swaminathan · Nexus Learning Labs, Bengaluru Paper · Dashboard · Maya-Morphe Series

---

cl-metrics measures what happened. maya-metrics measures how it felt. morphe-metrics measures where it went.

---

What is this?

morphe-metrics is the first open-source Python library for evaluating field-guided connectivity in morphogenetic neural networks — systems whose topology self-organises via bioelectric-analog voltage gradient signals.

The primary metric is Edge-Field Concordance (EFC): the Pearson correlation between bioelectric field strength at candidate neighbour nodes and edge formation probability. EFC was introduced and first measured in:

Swaminathan, V. (2026). Maya-Morphe P3: Axon. DOI: 10.5281/zenodo.20102536

Any researcher building a morphogenetic computing experiment can drop in morphe-metrics and immediately have a principled, citable way to measure field-guided connectivity.

Installation

pip install morphe-metrics

Or from source:

git clone https://github.com/venky2099/morphe-metrics
cd morphe-metrics
pip install -e .

Quick start

from morphe_metrics import compute_efc, interpret_efc, efc_summary

# Two arrays: field strength at candidate nodes, edge formation probability
field = [0.8, 0.3, 0.9, 0.1, 0.7]
probs = [0.7, 0.2, 0.85, 0.15, 0.6]

efc   = compute_efc(field, probs)         # 0.999 — strongly guided
label = interpret_efc(efc)                # 'strongly_guided'
info  = efc_summary(field, probs)         # full dict

EFC Interpretation

EFC range	Label	Meaning
< -0.05	pathological	Edges form strongly against the field
[-0.05, -0.001]	exploratory	Growth-cone analogy; biologically consistent
[-0.001, 0.001]	random	No field guidance detectable
(0.001, 0.5]	guided	Field-guided pathfinding
> 0.5	strongly_guided	Strong committed pathfinding

P3 empirical baseline: EFC = -0.009 to -0.019 (exploratory) P4 target: EFC > 0 (field-guided)

Epoch tracking

from morphe_metrics import EFCTracker

tracker = EFCTracker()

for epoch in range(1, 251):
    # ... training step ...
    tracker.update(epoch, field_at_candidate, edge_probs)

tracker.save_csv("logs/efc_trajectory.csv")
print(tracker.summary())
# {'n_epochs': 250, 'final_efc': 0.043, 'final_label': 'guided',
#  'p4_achieved': True, 'first_positive': 187, ...}

Topology metrics

from morphe_metrics import topology_delta, edge_retention_rate, field_alignment_score

# Net topology change
d = topology_delta(edges_before=1000, edges_after=1315)
# {'delta': 315, 'delta_frac': 0.315, 'direction': 'growth'}

# Fraction of formed edges that persist > k steps
rate = edge_retention_rate(formation_times, removal_times, k=5)

# Directional alignment between edges and field gradient
score = field_alignment_score(edge_direction_vectors, field_gradient_vectors)

Series constant checks

from morphe_metrics import bhaya_quiescence_check, buddhi_curve_check

# Bhaya Quiescence Law (0.32% baseline, confirmed across Maya P1-P9)
r = bhaya_quiescence_check(bhaya_values_across_trials)
# {'status': 'MODIFIED', 'ratio': 48.2, 'interpretation': '...'}

# Buddhi S-curve determinism
r = buddhi_curve_check(buddhi_trajectory_over_epochs)
# {'status': 'S_CURVE', 's_curve_ratio': 3.4, ...}

Key results (P3 empirical baseline)

Scale	Actual cells	EFC	Label	Edges formed
10k	29,056	-0.015	exploratory	315
30k	86,931	-0.019	exploratory	3,699
50k	144,726	-0.019	exploratory	11,202
75k	216,771	-0.010	exploratory	25,161
100k	266,071	-0.009	exploratory	35,958
120k	266,071	-0.017	exploratory	37,316

Running tests

pip install pytest
pytest tests/ -v

All 22 unit tests must pass before any release.

Canonical hyperparameter

VAIRAGYA_PHYSICS_CONSTANT = 0.002315  # ORCID: 0000-0002-3315-7907

Embedded as provenance signature in all morphe-metrics artifacts.

IP protection

• MIT License + ORCID attribution (see LICENSE)
• ORCID magic number 0.002315 embedded in VAIRAGYA_PHYSICS_CONSTANT
• Zenodo DOI timestamp — immutable, publicly verifiable

Citation

@software{swaminathan2026morphemetrics,
  author    = {Swaminathan, Venkatesh},
  title     = {{morphe-metrics: A Stateless Python Library for
                Morphogenetic Computing Evaluation}},
  year      = 2026,
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.20107084},
  url       = {https://doi.org/10.5281/zenodo.20107084},
  orcid     = {0000-0002-3315-7907}
}

Full Maya-Morphe Series

Paper	Title	DOI
P1	Bioelectric Gradient Fields as Topology Primitives	10.5281/zenodo.20052072
P2	Spanda — Adaptive Threshold Learning	10.5281/zenodo.20069445
P3	Axon — Differentiable Edge Formation	10.5281/zenodo.20102536
morphe-metrics	Evaluation library	10.5281/zenodo.20107084

---

Nexus Learning Labs, Bengaluru · ORCID: 0000-0002-3315-7907 MayaNexusVS2026NLL_Bengaluru_Narasimha