jpmschweitzer/settled-reach
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(simulation): static scattered freeze/snow model (T-1030) #163

Closed
jpmschweitzer wants to merge 0 commits from freeze-snow-model into main
pull from: freeze-snow-model
merge into: :main
Conversation 1 Commits - Files Changed -
jpmschweitzer commented 2026-06-08 17:16:48 +02:00
Owner
Copy Link
Copy Source

Summary

T-1030 — the static scattered freeze/snow model (D-239 §3). Adds a SeasonalCover { None, Snow, Ice } overlay axis to VoxelColumn (D-228 — cover, not TerrainMaterial), derived after the family generator in derive_voxel_column (one site; the 8 family generators are untouched).

The §3 model

  • Spatially-coherent cluster scatter (the load-bearing requirement — clustered patches, never per-tile dice): coarse 6 m cluster cells via div_euclid → Cantor pair → splitmix64 ^ sub_chunk_seed, so all voxels in a cell share the freeze decision → ragged patches. Integer/hash only (D-010).
  • Bands by surface (classify_surface from morphology_zone + water):
    • Freshwater (Lake / wet RiverBank·Meander·Delta·Estuarine·TidalFlat): +5 °C → −10 °C (the +5 = night-frost reaching the coldest scattered tiles).
    • Salt water (OpenOcean): own lower/wider band, −2 °C onset → −14 °C, with a 3×-coarser sheets + leads pack-ice pattern.
    • Land snow: +2 °C → −10 °C, moisture-gated (moisture_q < 30 → None, bare frozen ground).
    • Frozen fraction grows linearly across the band; permanent (no scatter) below it (poles/glaciers).
  • Airless (temperature_c == None) → cover None.

Deferred (Q-105 forward contract)

The transient/clock-bound layer (dawn frost that melts by noon → time-of-day passability shifts) requires Q-105's regional clock state and is not built — only the static mean-state cover field. Documented as a forward contract on the enum + derive_cover.

Tests

16 new cover tests: real spatial-coherence (≥90% intra-cluster agreement + inter-cluster variety), band edges per surface (freshwater/salt/snow above/within/below), moisture gating, salt-vs-fresh distinction, airless→None, determinism.

Verification

cargo test passes, clippy --all-targets -D warnings clean, fmt clean. D-010 integer throughout (temperature as i32, hash-based scatter — no f64/trig).

Ticket: T-1030 (epic T-974). With T-1031 (the verification harness) this completes the epic.

🤖 Generated with Claude Code

## Summary **T-1030** — the static scattered freeze/snow model (D-239 §3). Adds a `SeasonalCover { None, Snow, Ice }` overlay axis to `VoxelColumn` (D-228 — cover, **not** TerrainMaterial), derived after the family generator in `derive_voxel_column` (one site; the 8 family generators are untouched). ### The §3 model - **Spatially-coherent cluster scatter** (the load-bearing requirement — *clustered patches, never per-tile dice*): coarse 6 m cluster cells via `div_euclid` → Cantor pair → `splitmix64 ^ sub_chunk_seed`, so all voxels in a cell share the freeze decision → ragged patches. Integer/hash only (D-010). - **Bands by surface** (`classify_surface` from `morphology_zone` + water): - Freshwater (Lake / wet RiverBank·Meander·Delta·Estuarine·TidalFlat): +5 °C → −10 °C (the +5 = night-frost reaching the coldest scattered tiles). - Salt water (OpenOcean): own lower/wider band, −2 °C onset → −14 °C, with a 3×-coarser **sheets + leads** pack-ice pattern. - Land snow: +2 °C → −10 °C, **moisture-gated** (`moisture_q < 30` → None, bare frozen ground). - Frozen fraction grows linearly across the band; permanent (no scatter) below it (poles/glaciers). - **Airless** (`temperature_c == None`) → cover `None`. ### Deferred (Q-105 forward contract) The transient/clock-bound layer (dawn frost that melts by noon → time-of-day passability shifts) requires Q-105's regional clock state and is **not built** — only the static mean-state `cover` field. Documented as a forward contract on the enum + `derive_cover`. ### Tests 16 new cover tests: **real spatial-coherence** (≥90% intra-cluster agreement + inter-cluster variety), band edges per surface (freshwater/salt/snow above/within/below), moisture gating, salt-vs-fresh distinction, airless→None, determinism. ### Verification `cargo test` passes, `clippy --all-targets -D warnings` clean, `fmt` clean. D-010 integer throughout (temperature as i32, hash-based scatter — no f64/trig). Ticket: T-1030 (epic T-974). With T-1031 (the verification harness) this completes the epic. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
jpmschweitzer added 1 commit 2026-06-08 17:16:48 +02:00
feat(simulation): static scattered freeze/snow model (T-1030) ed7380245d 
D-239 §3 — the static mean-state freeze/snow cover. New SeasonalCover
{None,Snow,Ice} overlay axis on VoxelColumn (D-228 cover, NOT TerrainMaterial),
derived after the family generator in derive_voxel_column.

- Spatially-COHERENT cluster scatter (never per-tile dice): coarse 6m cluster
  cells (div_euclid) -> Cantor pair -> splitmix64 ^ sub_chunk_seed, so voxels in
  a cell share the freeze decision -> ragged patches. All integer (D-010).
- Bands by surface (classify_surface from morphology_zone + water): freshwater
  +5..-10C; salt water (OpenOcean) -2C onset / -14C, 3x-coarser sheets+leads;
  land snow +2..-10C, moisture-gated (moisture_q<30 -> None). Linear frozen
  fraction across the band; permanent below the band (poles/glaciers).
- Airless (temperature_c None) -> cover None.
- Transient/clock-bound layer (dawn frost melting by noon) doc'd as a Q-105
  forward contract; only the static model is built.

16 new cover tests incl real spatial-coherence (>=90% intra-cluster agreement),
band edges per surface, moisture gating, determinism. cargo test passes, clippy
-D warnings clean, fmt clean.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
jpmschweitzer added 1 commit 2026-06-08 17:31:23 +02:00
fix(simulation): address PR #163 review (freeze/snow model) 88b4323585 
Hoshe found a CRITICAL coherence defect (Tyre APPROVE):

- CRITICAL: cluster_scatter received the per-voxel sub_chunk_seed as its cluster
  seed, so voxels in the same 6m cell got different scatter -> salt-and-pepper,
  not the §3-mandated clustered patches. Re-key on (world, body, cluster-cell):
  add SeedDomain::Cover=10 (pinned); cluster_scatter(world_seed, body_id, x, y)
  derives for_body(world_seed,body_id).derive(Cover, cluster_pair_id). derive_cover
  now takes (world_seed, body_id, ...). (Hoshe #1)
- The coherence test was tautological (constant seed bypassed the broken path).
  Rewritten to the production derive_cover signature in a cold-lake scatter band;
  it FAILED at 20% on the old code, passes 100% now. (Hoshe #2)
- Zigzag-encode cluster (cx,cy) before Cantor pairing (negatives were cast as u64
  directly); fix the misleading 'no collision' comment. (Hoshe #3, Tyre)
- Sea-ice band was inverted (~85% ice at the -2 onset). Now monotonic: ice grows
  ~8% (onset) -> ~92% (cold edge); leads (open gaps) layered only in the cold half
  via a 3x super-cell, passing the super-cell index directly. Dead depth==0 guard
  removed. (Hoshe #4, Tyre #1/#3)

cargo test passes, clippy -D warnings clean, fmt clean.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
jpmschweitzer commented 2026-06-08 17:32:30 +02:00
Author
Owner
Copy Link
Copy Source

Review: freeze-snow-model → main (type: code)

Reviewers: Hoshe (code quality) + Tyre (architecture). Read from the branch worktree. No merge-path surface.

Tyre (Architecture): APPROVE

SeasonalCover correctly modeled as a D-228 cover overlay (orthogonal to TerrainMaterial), appended after the family generators; §3 fidelity solid across all three surfaces; Q-105 forward contract cleanly scoped (static mean-state only). Three clarity notes (dead onset branch, param naming, pack-ice quantization) — all folded into the fixes below.

Hoshe (Code Quality): REQUEST_CHANGES → resolved

Caught a critical defect Tyre and the lead both missed.

Findings & resolution (88b43235)

# Sev Finding Resolution
1 critical cluster_scatter was keyed on the per-voxel sub_chunk_seed, so voxels in the same 6 m cell got different scatter → salt-and-pepper, violating §3's "clustered patches, never per-tile dice" Re-keyed on (world, body, cluster-cell): new SeedDomain::Cover = 10; cluster_scatter(world_seed, body_id, …) derives for_body(...).derive(Cover, cluster_pair_id).
2 critical The coherence test passed a constant seed to derive_cover, bypassing the per-voxel variation — so it couldn't catch #1 (tautological) Rewritten to the production derive_cover signature in a cold-lake scatter band. It failed at 20% on the old code and passes at 100% now — it genuinely gates the contract.
3 med Cantor pairing cast negative cluster coords as u64 directly ("no collision" claim false) Zigzag-encode (cx,cy) before pairing (the voxel_pos_to_id pattern); comment fixed.
4 med Sea-ice band inverted (~85% ice at the −2 onset) + a dead depth==0 guard Now monotonic: ice grows ~8% (onset) → ~92% (cold edge); leads layered only in the cold half via a 3× super-cell (super-cell index passed directly); dead guard removed.

Verdict: APPROVED (after resolution)

The critical fix restores the §3 spatial-coherence contract, and the rewritten test now actually fails against the broken implementation (verified: lead also confirmed the seed path is body/cluster-keyed, not per-voxel). cargo test passes, clippy -D warnings clean, fmt clean. Merging.

🤖 Generated with Claude Code

## Review: freeze-snow-model → main (type: code) Reviewers: **Hoshe** (code quality) + **Tyre** (architecture). Read from the branch worktree. No merge-path surface. ### Tyre (Architecture): APPROVE `SeasonalCover` correctly modeled as a D-228 cover overlay (orthogonal to TerrainMaterial), appended after the family generators; §3 fidelity solid across all three surfaces; Q-105 forward contract cleanly scoped (static mean-state only). Three clarity notes (dead onset branch, param naming, pack-ice quantization) — all folded into the fixes below. ### Hoshe (Code Quality): REQUEST_CHANGES → resolved Caught a **critical** defect Tyre and the lead both missed. ### Findings & resolution (`88b43235`) | # | Sev | Finding | Resolution | |---|-----|---------|------------| | 1 | **critical** | `cluster_scatter` was keyed on the **per-voxel** `sub_chunk_seed`, so voxels in the same 6 m cell got different scatter → salt-and-pepper, violating §3's "clustered patches, never per-tile dice" | Re-keyed on `(world, body, cluster-cell)`: new `SeedDomain::Cover = 10`; `cluster_scatter(world_seed, body_id, …)` derives `for_body(...).derive(Cover, cluster_pair_id)`. | | 2 | **critical** | The coherence test passed a **constant seed** to `derive_cover`, bypassing the per-voxel variation — so it couldn't catch #1 (tautological) | Rewritten to the production `derive_cover` signature in a cold-lake scatter band. **It failed at 20% on the old code and passes at 100% now** — it genuinely gates the contract. | | 3 | med | Cantor pairing cast negative cluster coords `as u64` directly ("no collision" claim false) | Zigzag-encode `(cx,cy)` before pairing (the `voxel_pos_to_id` pattern); comment fixed. | | 4 | med | Sea-ice band inverted (~85% ice at the −2 onset) + a dead `depth==0` guard | Now monotonic: ice grows ~8% (onset) → ~92% (cold edge); leads layered only in the cold half via a 3× super-cell (super-cell index passed directly); dead guard removed. | ### Verdict: APPROVED (after resolution) The critical fix restores the §3 spatial-coherence contract, and the rewritten test now actually fails against the broken implementation (verified: lead also confirmed the seed path is body/cluster-keyed, not per-voxel). `cargo test` passes, clippy `-D warnings` clean, fmt clean. Merging. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
jpmschweitzer closed this pull request 2026-06-08 17:34:04 +02:00

Pull request closed

This pull request cannot be reopened because the branch was deleted.
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
No items
No labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
jpmschweitzer (Jeroen Schweitzer)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: jpmschweitzer/settled-reach#163