# FUM Historical Gallery

This gallery preserves figures, run logs, assessment outputs, and other execution artifacts from the **FUM generation**.

The three source repositories under `variants/` remain the historical code/document states. The gallery is a curated navigation layer above them: filenames have been normalized for readability, while the original gallery filename and SHA-256 for every carried-forward artifact are recorded in [`GALLERY_MANIFEST.csv`](GALLERY_MANIFEST.csv).

These artifacts should be read as **historical run evidence**, not as one normalized benchmark suite. They come from different FUM repository states, configurations, network sizes, experimental mechanisms, and points in development.

## 01 — Signature emergence

This is the most personally important visual artifact in the collection.

The first image is the structure I remember as the first time I saw FUM self-organize into something that was **obviously organized**. Until then, I had seen complex outputs and interesting dynamics. This one landed differently: the structure was visually unmistakable enough that it changed how I understood what I was building.

That form later became the basis for **Neuroca's logo**.

<p align="center">
  <img src="./01_signature_emergence/01_foundational_ukg_structure.png" alt="Foundational UKG Structure" width="760"/>
</p>

<p align="center">
  <img src="./01_signature_emergence/02_neuroca_logo_mark.png" alt="Neuroca logo mark derived from the emergent structure" width="180"/>
</p>

<p align="center">
  <img src="./01_signature_emergence/03_neuroca_banner.png" alt="Neuroca banner using the graph-derived logo" width="1000"/>
</p>

The significance of this artifact is historical and developmental. The visual organization itself is not being presented here as a standalone scientific proof; it is preserved because it marks a real turning point in the research lineage and directly influenced Neuroca's visual identity.

## 02 — Connectomes and structural snapshots

[`02_connectomes/`](02_connectomes/) contains preserved initial substrates, final connectomes, final structures, and related graph visualizations from multiple FUM runs.

These images are intentionally not collapsed into a single "best" result. Different runs produced materially different structures, including dense cores, long branches, multiple territories, fragmented-looking states, and highly centralized forms.

A few examples:

<p align="center">
  <img src="./02_connectomes/final_structure_01.png" alt="Historical FUM final structure" width="300"/>
  <img src="./02_connectomes/final_connectome_10.png" alt="Historical FUM final connectome" width="300"/>
  <img src="./02_connectomes/final_structure_04.png" alt="Historical FUM final structure" width="300"/>
</p>

The point of preserving the set is to make **variation across runs inspectable**, not to imply that every topology had the same interpretation.

## 03 — Spike rasters

[`03_spike_rasters/`](03_spike_rasters/) preserves FUM spike-raster outputs.

These are useful for comparing activity patterns against the corresponding structural and regulatory behavior in the logs.

## 04 — Dashboards and analysis figures

[`04_dashboards/`](04_dashboards/) contains historical dashboards and analysis figures, including:

- three artifacts carrying `VDM` dashboard naming from the FUM-to-Void-Dynamics transition;
- an early SIE stability simulation figure.

The naming is preserved conceptually rather than rewritten to imply that these are current VDM implementations.

## 05 — Run logs

[`05_run_logs/`](05_run_logs/) is one of the most useful parts of the gallery.

The preserved logs expose the system at mechanism level: initialization, dataset validation, SIE signals, VGSP/RE-VGSP updates, EHTP topology metrics, ADC domain cartography, GDSP/homeostatic intervention, memory accounting, spike statistics, and changes in experimental scale.

Examples in the collection include:

- early 1,600-synapse Phase-1 runs;
- a 2,400-synapse FUM Demo run;
- multiple 8,000-synapse runs at 100, 200, 1,000, and 10,000 presented stimuli;
- 1,000-neuron developmental runs with explicit self-regulation cycles and memory accounting.

The filenames now use the recorded `run_id` wherever one was available, which is substantially more useful than historical desktop-copy names such as `(11th copy)` or `(another copy)`.

## 06 — Assessments

[`06_assessments/`](06_assessments/) contains preserved assessment reports kept separate from raw execution logs.

## File custody

Normalization in this gallery changes **paths and filenames**, not the bytes of the carried-forward historical gallery artifacts.

For each file, [`GALLERY_MANIFEST.csv`](GALLERY_MANIFEST.csv) records:

- normalized path;
- original gallery/source name;
- SHA-256;
- byte size;
- artifact class;
- a short curation note.

The three signature-emergence images were supplied separately during the 2026 provenance curation and are identified as such in the manifest.

## Suggested use

If you are experimenting with the FUM variants, useful ways to use this gallery include:

1. compare a fresh run against historical structural outputs without assuming exact reproduction;
2. trace topology metrics in a log against structural snapshots from the same development era;
3. reproduce the multi-seed variation rather than selecting one visually compelling graph;
4. test whether later FUM topology instrumentation explains differences visible in earlier connectomes;
5. preserve new run artifacts under a separate experiment directory instead of overwriting this historical gallery.

The gallery exists to make the old system easier to interrogate, not to freeze one interpretation of what every image meant.