Restore the ITensors.jl cross-check and key the results by measured revision

.github/workflows/ci.yml

@@ -14,6 +14,8 @@ jobs:
       - uses: dtolnay/rust-toolchain@stable
       - uses: Swatinem/rust-cache@v2
       - run: cargo fmt --all -- --check
+      - name: Check the generated result index
+        run: python3 scripts/gen_index.py --check
       - run: cargo clippy --locked --all-targets -- -D warnings
       - run: cargo test --release --locked
       - name: smoke Fourier elementwise
@@ -30,4 +32,14 @@ jobs:
         run: >
           BENCH_NS=2 BENCH_RUNS=1 BENCH_WARMUPS=0
           BENCH_INPUT_CACHE_DIR=/tmp/smoke-cache OUT_DIR=/tmp/smoke/raw
+          EXPORT_HDF5=/tmp/smoke-export
           cargo run --release --locked --bin mpo_mpo_aniso_patched
+      - uses: julia-actions/setup-julia@v2
+        with:
+          version: "1"
+      - name: Julia cross-check of the MPO contraction inputs
+        # Reads the HDF5 the smoke step exported and evaluates it in ITensors.jl
+        # against the closed form the instance approximates.
+        run: |
+          julia --project=julia -e 'using Pkg; Pkg.instantiate()'
+          julia --project=julia julia/check_mpo_mpo.jl /tmp/smoke-export 2

AGENTS.md

@@ -24,7 +24,11 @@ single-core profile stays the reference, and the two profiles are never merged.
 
 ## Results
 
-Files under `result/<profile>/` are generated by `scripts/run_all.sh <profile>` and committed deliberately. Do not edit them manually. Probe runs belong under `/tmp`. One profile belongs to one physical machine and contains no hostname.
+Files under `result/<profile>/<rev12>/` are generated by
+`scripts/run_all.sh <profile>` and committed deliberately. `result/INDEX.md` is
+generated by `scripts/gen_index.py`. Do not edit either by hand. Probe runs
+belong under `/tmp`. One profile directory belongs to one physical machine and
+contains no hostname.
 
 ## Input cache
 

README.md

@@ -4,22 +4,17 @@ Reproducible benchmarks for tensor network algorithms in [tensor4all-rs](https:/
 
 ## Results
 
-Every committed sweep is a directory under [`result/`](result/), one per hardware
-profile. Each holds `run.yaml` (machine, revisions, thread count and CPU set),
-`raw/` (one JSON record per case, arm and instance) and the generated
-`report.md`. Read the reports:
-
-| Profile | Machine | Threads | tensor4all-rs | Report |
-| --- | --- | ---: | --- | --- |
-| `linux-ryzen-ai-9-hx-470` | AMD Ryzen AI 9 HX 470, 12 physical cores | 1 | `a515738` | [report.md](result/linux-ryzen-ai-9-hx-470/report.md) |
-| `linux-ryzen-ai-9-hx-470-4t` | the same machine, CPUs 0-3 | 4 | `a515738` | [report.md](result/linux-ryzen-ai-9-hx-470-4t/report.md) |
-| `linux-epyc-7713p` | AMD EPYC 7713P | 1 | `9e9aeda` | [report.md](result/linux-epyc-7713p/report.md) |
-
-The one-thread profile is the reference configuration, and each report repeats
-the machine, revisions, thread count and CPU set in its header. The `-4t`
-profile is scaling evidence only: read it against the one-thread profile of the
-same machine, never instead of it. [`result/README.md`](result/README.md)
-describes the layout.
+[`result/INDEX.md`](result/INDEX.md) lists every recorded sweep: one row per
+(hardware profile, measured tensor4all-rs revision), each linking to that
+sweep's generated `report.md`. It is generated by `scripts/gen_index.py`, so it
+cannot drift from the tree, and CI fails when it is out of date.
+
+A sweep is a directory `result/<profile>/<rev12>/` holding `run.yaml` (machine,
+revisions, thread count and CPU set), `raw/` (one JSON record per case, arm and
+instance) and the `report.md` those records generate. The one-thread profile of
+each machine is the reference configuration; its optional `-<n>t` scaling sweep
+is read against it, never instead of it. [`result/README.md`](result/README.md)
+has the layout.
 
 ## Maintained cases
 
@@ -66,6 +61,32 @@ Case 3 compares global TreeTN fit with patched chain-TreeTN fit. Its reference i
 
 Input construction, cache I/O, global input compression, format conversion, patch preparation, output conversion, and accuracy evaluation are outside timed regions.
 
+## Julia cross-check
+
+Case 3 exports its two input QTTs on request, and `julia/check_mpo_mpo.jl`
+evaluates them in ITensors.jl against the closed form the interpolative
+construction approximates. This is the one place where an artifact written by
+tensor4all-rs is read back and contracted by an independent implementation.
+
+```bash
+BENCH_NS=2 OUT_DIR=/tmp/smoke/raw EXPORT_HDF5=/tmp/smoke-export \
+  cargo run --release --locked --bin mpo_mpo_aniso_patched
+julia --project=julia -e 'using Pkg; Pkg.instantiate()'
+julia --project=julia julia/check_mpo_mpo.jl /tmp/smoke-export 2
+```
+
+`EXPORT_HDF5` writes `instance-n<N>.h5` (the fused site-dimension-4 QTTs as
+ITensorMPS `MPS` groups `f` and `g`) and `instance-n<N>.json` (the analytic
+mixtures, the box, `r` and the tolerances) into that directory. The script
+asserts the case name before reading, because the Gaussian cases once shared
+these file names.
+
+It prints the largest sampled relative deviation and gates at `1e-4`,
+overridable with `CHECK_THRESHOLD`. At the default instances that deviation is
+between `5e-6` and `9e-6`, which is the scale of the `1e-6` input truncation: the
+check catches a broken export or a wrong index convention, not an accuracy
+bound. CI runs it on the `N = 2` smoke instance.
+
 ## Running
 
 Requirements are Rust, HDF5, and a BLAS/LAPACK implementation. A complete single-core run is:
@@ -114,7 +135,8 @@ cargo fmt --all -- --check
 cargo test --release
 cargo clippy --locked --all-targets -- -D warnings
 cargo check --locked --all-targets
-python3 scripts/report.py result/<profile>
+python3 scripts/report.py result/<profile>/<rev12>
+python3 scripts/gen_index.py --check
 git diff --check
 ```