ShortFactory Research

Distributed substrate engineering · adaptive lattices · embodied computation
Somerset · UK · Daniel Chipchase · ORCID 0009-0003-9001-2493

This portal hosts technical artefacts for external researchers, engineers, and reviewers. Each link below is a self-contained engineering deliverable. Falsifiability and reproducibility are prioritised over scope.

If you've come from elsewhere on the empire, this page is intentionally sparse. Conceptual and synthesis work is hosted separately. Dent Matrix is the active research thread.

Active programme · Dent Matrix v0.1

Live demonstrator

RUNNING NOW 100 nodes · 60 Hz · Python · stdlib only

An HCP lattice running real-time physics — adaptive coupling, energy flow, Bellman-Ford routing, self-repair on heartbeat timeout. Click any node to damage it; watch surviving nodes reroute around it within 3–5 ticks. Click RECHARGE to restore energy. Apply load to see phase transitions cascade.

Open demonstrator → Telemetry JSON

Preprint

v1 · 9 May 2026 DOI 10.5281/zenodo.20095798

Distributed Local-Only Adaptive Coupling in a Self-Healing Electromagnetic Lattice

D. Chipchase · ShortFactory Research

Engineering description of the substrate primitive: hex lattice, six-phase magnetic coupling, Ohm's-law energy gradient, heartbeat-triggered Bellman-Ford rerouting. Full BOM with off-the-shelf components, complexity analysis, and §7 limitations including synchronisation instability, magnetic interference, and route convergence latency.

Read preprint → References (11)

Reference implementation

Apache 2.0 Python ≥ 3.10 no pip deps

Public GitHub repository · clone, run, instrument, fork, falsify.

git clone https://github.com/eliskcage/dent-matrix cd dent-matrix python3 dent_server.py

Includes dent_mesh.py (physics engine), dent_server.py (HTTP/telemetry layer), full docs/ tree, eight standardised benchmarks, and a runs/ directory for committed datasets.

github.com/eliskcage/dent-matrix →

Hardware roadmap · M1 build doc

£129 lab build ≈6 weeks to first dataset

7-node observability-first demonstrator · ESP32-C3 · 5 F supercap · controlled-asymmetry layout · pluggable coils · UART live logging.

Each node ≈ £3.83 in single quantities. The build doc specifies four numbered measurement procedures (P1–P4) with explicit pass/fail thresholds, plus a failure-analysis table predicting common bring-up issues. Optimised for observability: oversized traces, exposed test pads, removable coils, external power injection.

M1 build doc →

Standardised benchmark scenarios

Eight reproducible tests that any implementation can run. Each specifies setup, stimulus, measurements, and pass thresholds — so external builds can produce comparable numbers.

ScenarioTestsPass threshold
random_node_deathreroute latencyconvergence ≤ 5 ticks (N=1)
cascading_overloaddamage propagationdamage stays within 1-hop of stimulus
partition_recoveryre-merge after severcross-island routes within 8 ticks of bridge heal
load_transferphase wave propagationadjacent nodes ≥ phase 3 within 10 ticks
asymmetry_stressweak-edge avoidanceload detours around weak edge
reroute_latencyscaling with cluster sizesub-linear in N
energy_starvationbrown-out propagationlocalised to loaded cluster
healing_under_loadself-repair under stress≥ 70% routes restored within 20 ticks

Specifications: docs/benchmarks.md

Milestones

MilestoneGoalStatus
M0 simulatordistributed adaptive coupling proven in softwarecomplete · live
M1 7-node panelfirst physical proof under load + damagebuild doc published
M2 19-node panelcoupling-force vs current curves under physical loadspec'd · awaiting build
M3 100-node panel1:1 hardware parity with simulatorspec'd
M4 mobile shard transportmetabolic loop closureresearch
M5 3D folded latticelocomotion via coordinated phase changeshorizon

Datasets

Empirical runs are committed to the public repo under runs/. Time-series evidence of substrate evolution accumulates in git history. Cite specific commits in publications.

runs/ on GitHub

Citation

@misc{chipchase2026dent, author = {Chipchase, Daniel}, title = {Distributed Local-Only Adaptive Coupling in a Self-Healing Electromagnetic Lattice}, year = {2026}, month = {May}, publisher = {Zenodo}, version = {v0.1.0}, doi = {10.5281/zenodo.20095798}, url = {https://doi.org/10.5281/zenodo.20095798} }

Contact

Emaildan@shortfactory.shop
ORCID0009-0003-9001-2493
GitHubeliskcage
LocationSomerset, UK · willing to travel for collaborator visits

Open to: collaboration with university robotics groups, prosthetics labs, adaptive manufacturing programmes, or independent researchers running the benchmark suite. Particularly interested in correspondence about ESP32-NOW synchronisation under high broadcast density and ferrite core selection for sub-15 mm node spacing.