Kmesh distance floor

AGENTS.md

@@ -78,15 +78,16 @@ record keeps the convention it was published with, so a consumer must read the
 base from the record rather than assume it.
 
 The ladder is built from the k-distances at which `ceil(|b_i| / k_distance)`
-changes on any axis — that is, from `|b_i| / n`. The enumeration cap on `n` is
-applied per axis, so axes with different `|b_i|` exhaust their breakpoints at
-different k-distances: the ladder is a complete set of meshes only for
-`k_distance >= max(|b_i|) / n_max`, and below that it silently skips reachable
-meshes. Any recomputation or published `k_index` column must state the cap it
-used.
-
-This convention is shared with `goldilocks-core`. Changing it invalidates every
-`k_index` value already recorded or trained on.
+changes on any axis — that is, from `|b_i| / n`. Enumeration stops at a
+resolution floor `min_k_distance` (default `0.03` Å⁻¹, on the solid-state 2π
+lengths — the AiiDA-QuantumESPRESSO convention). The floor is the same for every
+axis, so all axes stop together and the ladder is a complete, gap-free set of
+meshes over the whole `[min_k_distance, ∞)` range. Any recomputation or
+published `k_index` column must state the floor it used; a `k_index` from a
+different floor is comparable only where the two ranges overlap.
+
+This convention must stay in step with `goldilocks-core` and `goldilocks-ml`.
+Changing it invalidates every `k_index` value already recorded or trained on.
 
 ## Commands
 

campaigns/qe/kpoints/README.md

@@ -9,7 +9,7 @@ PseudoDojo PBEsol campaign and the SSSP comparison analysis.
 ```text
 kpoints/
   campaign.yaml       human-readable campaign configuration
-  scripts/            initial submission, extension, and monitoring
+  scripts/            extension and monitoring
   notebooks/          curated analysis and visualisation
   results/            snapshot, summary, and provenance manifest
 ```
@@ -49,8 +49,6 @@ uv run --extra aiida --extra kmesh python campaigns/qe/kpoints/scripts/monitor.p
 ```
 
 The monitor calls `extend.py` in a fresh process. Stop it with `Ctrl-C`.
-`submit_initial.py` is for the original seed campaign and additionally needs
-the historical convergence summary files; run `--help` for its full interface.
 
 ## Convergence definition
 

campaigns/qe/kpoints/scripts/extend.py

@@ -16,7 +16,8 @@
 from goldilocks_data.aiida.submit import existing_kindices_by_source, submit_jobs
 from goldilocks_data.codes import DftCode
 from goldilocks_data.intents import CalculationIntent
-from goldilocks_data.sweeps import AiidaJobSpec, KindexExtension, kindex_points, plan_well_not_ultra_extensions
+from goldilocks_data.kmesh import kindex_points
+from goldilocks_data.sweeps import AiidaJobSpec, KindexExtension, plan_well_not_ultra_extensions
 
 TASK_ROOT = Path(__file__).resolve().parents[1]
 DEFAULT_SNAPSHOT_DIR = TASK_ROOT / "results"

docs/published-records.md

@@ -7,16 +7,18 @@ identifier and can be cited.
 These are snapshots. The AiiDA database remains the authoritative calculation
 record; a published dataset is a documented view of it at one point in time.
 
-## Quantum ESPRESSO no-spin SCF calculations (SSSP, K-index)
+## Quantum ESPRESSO no-spin SCF calculations (SSSP, k-index)
 
-[`d5ds2-64f16`](https://data-collections.psdi.ac.uk/records/d5ds2-64f16) · v1 ·
+[`52713-55d86`](https://data-collections.psdi.ac.uk/records/52713-55d86) · v1.0 ·
 CC BY 4.0
 
-The converged k-point mesh for 17,757 MC3D structures. No spin polarisation,
-SSSP PBEsol pseudopotentials, every mesh unshifted and therefore
-gamma-inclusive.
+The current SSSP k-index dataset: the converged k-point mesh for 17,757 MC3D
+structures, numbered on the **1-based** ladder (rung 1 the Γ-only `(1, 1, 1)`
+mesh) and built with the resolution floor `min_k_distance = 0.03` Å⁻¹ rather
+than a per-axis k-point cap. No spin polarisation, SSSP PBEsol pseudopotentials,
+every mesh unshifted and therefore gamma-inclusive.
 
-Convergence is the first of three consecutive k-distances whose total energies
+Convergence is the first of three consecutive ladder meshes whose total energies
 agree within **1 meV per atom**. Energy only — no force criterion.
 
 | File | Contents |
@@ -28,28 +30,9 @@ The archive carries more structures than the table has rows: 463 structures were
 calculated but never met the criterion within the range of meshes swept, so they
 have a structure file and no converged answer.
 
-### The k_index in this record
-
-`k_index` in this record is **0-based, with rung 0 the gamma-only `(1, 1, 1)`
-mesh**, and was computed with a per-axis enumeration bound of **50**, the ladder
-truncated at the first rung where an axis count would rise by more than one.
-
-!!! warning "This record is 0-based; the convention since is 1-based"
-
-    Everything produced after this record numbers the same ladder from 1, so
-    rung *n* here is rung *n + 1* under the current convention. The published
-    record is not rewritten: it keeps the convention it was published with, and
-    a consumer reads the base from the record rather than assuming it.
-
-That bound is part of the definition, not an implementation detail. The change
-points of the ladder are `|b_i| / n`, the bound applies per axis, and axes with
-different `|b_i|` exhaust their change points at different k-distances — so the
-ladder is a complete set of meshes only down to `max(|b_i|) / 50`. Below that a
-reachable mesh can be skipped. Any recomputation must state the bound it used or
-its `k_index` values are not comparable with these.
-
-Raising the bound only appends rungs and never renumbers an existing one, so
-these values stay valid under a larger enumeration.
+`k_index` needs both its base and its floor to mean anything; the record's
+`README.md` and `manifest.json` carry both, and a consumer reads them from there
+rather than assuming.
 
 See [convergence criteria](reference/convergence.md) for how labels are assigned,
 and the record's own `README.md` for the full definition and reproduction code.

docs/publishing.md

@@ -49,7 +49,7 @@ convention a consumer must share are required fields:
   "rows": 17757,
   "columns": [{"name": "k_index", "dtype": "int", "description": "..."}],
   "conventions": {
-    "kmesh_ladder": {"base": 0, "rung_0": "gamma_only", "max_kpoints_per_axis": 50}
+    "kmesh_ladder": {"base": 1, "rung_1": "gamma_only", "min_k_distance": 0.03}
   },
   "provenance": {"code": "quantum_espresso", "calculation": "scf", "spin": "none"}
 }