# Cached entropy estimates

Machine-generated by [`.github/workflows/estimates.yml`][wf] on `main` — do not
edit by hand. Every dispatch runs the whole grid again with fresh, independent
pasts, appends one record per cell to `runs.jsonl`, and rebuilds
`estimates.json` from the full log, so the means tighten run after run.

- **`runs.jsonl`** — append-only, one JSON object per (cell, dispatch): the
  cell's parameters, the per-past values with their sum and sum of squares,
  the per-past resolved Gibbs thinning and probe autocorrelation times
  (`thin='auto'` matches thinning to each chain's measured mixing), the
  seed, and the code commit that produced them.
- **`estimates.json`** — pooled `mean` and `se` per cell over every past ever
  drawn for it, alongside the analytic `predicted` rate for comparison.

The estimand is H(z_(M+1) | z_1..z_M) in bits/sample — an upper bound on the
entropy rate of z that tightens as the conditioning window M grows.

Fetch it from a browser or a script (raw.githubusercontent.com sends
`access-control-allow-origin: *`):

    https://raw.githubusercontent.com/concept-collection/timeseries-entropy/estimates/estimates.json

## Current state (155 runs, generated 2026-08-17T08:39:15Z)

| cell | sigma | M | mean | se | pasts | predicted |
| --- | --- | --- | --- | --- | --- | --- |
| bp300-6000_s1 | 1 | 512 | 0.9338 | 0.0035 | 1240 | 0.9513 |
| bp300-6000_s16 | 16 | 512 | 2.5529 | 0.0075 | 1240 | 2.5680 |
| bp300-6000_s2 | 2 | 512 | 1.3133 | 0.0033 | 1240 | 1.3319 |
| bp300-6000_s32 | 32 | 512 | 2.9845 | 0.0182 | 1240 | 2.9938 |
| bp300-6000_s4 | 4 | 512 | 1.7263 | 0.0042 | 1240 | 1.7342 |
| bp300-6000_s8 | 8 | 512 | 2.1298 | 0.0054 | 1240 | 2.1476 |
| lp3000_s1 | 1 | 512 | 0.4809 | 0.0067 | 1240 | 0.6166 |
| lp3000_s16 | 16 | 512 | 1.4442 | 0.0060 | 1240 | 1.4888 |
| lp3000_s2 | 2 | 512 | 0.7588 | 0.0076 | 1240 | 0.8251 |
| lp3000_s32 | 32 | 512 | 1.6672 | 0.0060 | 1240 | 1.7179 |
| lp3000_s4 | 4 | 512 | 0.9927 | 0.0054 | 1240 | 1.0415 |
| lp3000_s8 | 8 | 512 | 1.2087 | 0.0060 | 1240 | 1.2632 |
| ma8_s1 | 1 | 512 | 0.4816 | 0.0060 | 1240 | 0.6100 |
| ma8_s16 | 16 | 512 | 3.1742 | 0.0068 | 1240 | 3.1777 |
| ma8_s2 | 2 | 512 | 0.9747 | 0.0027 | 1240 | 0.9825 |
| ma8_s32 | 32 | 512 | 4.0890 | 0.0069 | 1240 | 4.1125 |
| ma8_s4 | 4 | 512 | 1.5559 | 0.0030 | 1240 | 1.5529 |
| ma8_s8 | 8 | 512 | 2.3082 | 0.0047 | 1240 | 2.3064 |
| none_s1 | 1 | 512 | 2.1054 | 0.0019 | 1240 | 2.1048 |
| none_s16 | 16 | 512 | 6.0215 | 0.0164 | 1240 | 6.0473 |
| none_s2 | 2 | 512 | 3.0652 | 0.0032 | 1240 | 3.0620 |
| none_s32 | 32 | 512 | 7.0417 | 0.0292 | 1240 | 7.0472 |
| none_s4 | 4 | 512 | 4.0411 | 0.0041 | 1240 | 4.0508 |
| none_s8 | 8 | 512 | 5.0464 | 0.0079 | 1240 | 5.0480 |

[wf]: https://github.com/concept-collection/timeseries-entropy/blob/main/.github/workflows/estimates.yml