Log fleet_k_q48/fleet_conserved; K holds exactly, 775/775 ticks (FABRIC-3.md §XVIII)

doe_log.c's per-heartbeat-tick CSV gains two columns: fleet_k_q48
(vm_physics_fleet_heat_sum() over ALL live VMs -- the genuine fleet-wide
conservation invariant K, not reconstructable from the 3 named-Tripod-
member heat columns already logged, which omit every identity VM's own
heat) and fleet_conserved (vm_physics_conserved() as 0/1). Requested
explicitly after the first heartbeat-telemetry analysis pass
(analysis-20260912/) omitted K entirely.

Kernel rebuilt on all three architectures, full 3x9x3 campaign rerun
(results-20260912-with-k/). K = 1.0000000000 (Q48.16 raw 65536) on every
one of 775 heartbeat-tick observations, sd(K) = 0, 100% fleet_conserved,
across amd64/aarch64/riscv64, nine identities, three replicates -- zero
deviation. Also a free regression check on both recent Stadium fixes
(§XVI/§XVII): neither disturbed the reservoir-transfer accounting K
depends on.

Found and fixed a tooling wrinkle along the way: fleet_conserved, being
the CSV row's very last field with nothing after it to bound a regex
match, can have a resumed trial digit merge into it with zero separator
on the wire -- combine.py now derives it from fleet_k_q48 directly (same
epsilon vm_physics_conserved() uses) instead of trusting the raw field.
fleet_k_q48 itself is unaffected either way.

Full analysis, discussion, and light/dark SVG->PDF figures written up as
a proper LaTeX report (report-20260912/report/std79_doe_report.pdf),
following experiments/bare_metal/analysis/report/bare_metal_doe_report.tex's
established style -- supersedes analysis-20260912/'s markdown-only first
pass as the primary deliverable for this dataset (kept, not discarded).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EXieurDfDSsDFdnSyusuWo

experiments/std79-doe/report-20260912/README.md

@@ -0,0 +1,64 @@
+# std79-doe LaTeX report: the fleet conservation invariant K
+
+**Read `report/std79_doe_report.pdf`** — this supersedes `../analysis-20260912/`
+(the first, markdown-only pass) as the primary deliverable for this dataset.
+That first pass was correctly flagged as inadequate: it never captured K
+(the fleet-wide execution-heat conservation invariant) at all, and presented
+results as a plain markdown page rather than in this project's established
+LaTeX report style (see `experiments/bare_metal/analysis/report/` for the
+precedent this follows — same document class, packages, colour palette,
+light/dark figure pairs, abstract-first structure with headline numbers).
+
+## What changed to make this possible
+
+`src/starkernel/doe_log.c`'s per-heartbeat-tick CSV gained two new columns
+(`fleet_k_q48`, `fleet_conserved` — see the file's own header comment for
+the full 20-column schema), requiring a kernel rebuild and a fresh DoE
+campaign rerun on all three architectures
+(`../results-20260912-with-k/*-doe-raw.log` + `*-heartbeat.csv`).
+
+## Pipeline (same shape as `../analysis-20260912/`, extended for K)
+
+1. `correlate_doe.py <raw_log> <out_csv>` — per architecture, against
+   `../results-20260912-with-k/{amd64,aarch64,riscv64}-doe-raw.log` →
+   `{amd64,aarch64,riscv64}-correlated.csv` (committed, not scratch).
+2. `combine.py <in1.csv> [in2.csv ...] <out.csv>` → `combined.csv` (775 rows).
+   Decodes `fleet_k_q48` to a plain float and **derives** `fleet_conserved`
+   from `fleet_k_q48` directly (using the same epsilon
+   `vm_physics_conserved()` uses, `capsule_vm_physics.c`) rather than
+   trusting the raw `fleet_conserved` field — that field is
+   `doe_log_tick_row()`'s very last printed column, with nothing after it
+   in the row to bound a regex match against, so a resumed trial digit can
+   merge into it with zero separator on the wire (confirmed live: raw
+   values like `19223372036854775807` where `0`/`1` was expected). See the
+   script's own comment for the full explanation.
+3. `analysis.R` (run from this directory) — computes the K result
+   (trivial: one distinct value, `sd = 0`, reported directly rather than
+   via a formal test — see the report's own \S5.1 for why), per-cell
+   architecture × identity summaries, and two-way ANOVA for 12 secondary
+   metrics; writes `charts/*.svg` (light+dark pairs, matching
+   `experiments/bare_metal/analysis/analyse_bare_metal.R`'s theme
+   convention) and `tables/*.csv`.
+4. SVGs converted to PDF for LaTeX via `rsvg-convert` into `report/figures/`.
+5. `report/std79_doe_report.tex`, hand-authored following
+   `experiments/bare_metal/analysis/report/bare_metal_doe_report.tex`'s
+   structure and conventions; compiled with `pdflatex` (three passes for
+   cross-references/TOC).
+
+## Headline result
+
+$K = 1.0000000000$ (raw Q48.16 `65536`) on **every one of 775
+heartbeat-tick observations**, `sd(K) = 0`, `fleet_conserved = 1` on 100% of
+rows, across all three architectures, nine identities, and three
+replicates. Zero deviation. See the report for the full discussion,
+including why this is also a free regression check on the two Stadium
+subsystem fixes (FABRIC-3.md §XVI–XVII) that landed the same week.
+
+## Known limitation
+
+Same interleaving-correlation caveat as `../analysis-20260912/`: 1 of 27
+`run_id` markers was unrecoverable on 2 of 3 architectures this run
+(aarch64/riscv64: `run_id` 1 folded into `run_id` 0). `fleet_k_q48` itself
+is entirely unaffected by this — it's read directly at each tick regardless
+of trial attribution, so the K result does not depend on trial labeling
+being correct at all. See report §2.3/§6.4 for the full explanation.