erdos-problems 0.1.4 → 0.1.6

package/docs/ERDOS_PROBLEMS_REPO_SPEC.md

@@ -4,120 +4,48 @@ Last updated: 2026-03-25
 
 ## Purpose
 
-This document defines a new public-facing sibling repository:
+This document defines the public-facing `erdos-problems` repository and npm package.
 
-- repo name target: `erdos-problems`
-- npm package target: `erdos-problems`
-- CLI executable target: `erdos`
+Targets:
+- repo name: `erdos-problems`
+- npm package: `erdos-problems`
+- CLI executable: `erdos`
 
-This repo is meant to be the polished atlas and harness for Paul Erdős problems.
+`sunflower-coda` remains the experimental lab.
+`erdos-problems` is the clean public atlas and harness surface.
 
-`sunflower-coda` remains the experimental proving ground.
-
-`erdos-problems` becomes the clean published surface.
-
-## Product Position
-
-`erdos-problems` should be:
+## Product position
 
+`erdos-problems` is:
 - a problem atlas for Paul Erdős problems
-- a research harness with staged workflows for selected problem families
-- CLI-first
-- dossier-first
-- evidence-linked
+- a dossier-first CLI research harness
+- provenance-aware
 - formalization-aware
+- agent-friendly from first install
 
-It should not be:
-
-- a clone of `sunflower-coda`
-- a raw mirror of every internal experiment
+It is not:
+- a raw mirror of `sunflower-coda`
 - a generic research framework
+- a place where local route state silently overwrites public upstream truth
 
-## Core Product Model
+## Core model
 
-The repo should have two layers.
+The repo has three layers.
 
 ### 1. Atlas layer
 
-All Erdős problems live here, including:
-
-- open problems
-- solved problems
-- prize information
-- source links
-- tags
-- references
-- dossier links
-- formalization status
-
-Every problem gets a documented home.
-
-### 2. Harness layer
-
-Selected problem families get deeper staged research machinery.
-
-The first deep harness should be:
-
-- `sunflower`
-
-Later families can be added only if they earn it.
-
-## Naming Recommendation
-
-Recommended repo and package name:
-
-- `erdos-problems`
-
-Reason:
-
-- broader than “conjectures”
-- matches the actual scope better
-- includes solved entries naturally
-- leaves room for atlas + dossiers + harnesses
-
-As of 2026-03-25, the following npm names appeared open when checked:
-
-- `erdos-problems`
-- `erdos-conjectures`
-- `erdos-conjecture`
-
-Recommended binary name:
-
-- `erdos`
-
-## Relationship To Existing Repos
-
-### `sunflower-coda`
-
-Role:
-
-- experimental lab
-- theorem-engine proving ground
-- route-design incubator
-- UI/app experimentation
-
-### `erdos-problems`
-
-Role:
+All seeded Erdős problems live here with canonical local dossiers.
 
-- public atlas
-- clean CLI surface
-- curated problem dossiers
-- stable research harness for selected families
+### 2. Upstream truth layer
 
-### `ode-to-erdos`
+The package ships a bundled snapshot of `teorth/erdosproblems` and can refresh a workspace-local snapshot.
 
-Role:
+### 3. Harness layer
 
-- narrower theorem-work CLI
-- can either remain a focused sibling or be absorbed as an internal package inside `erdos-problems`
+Selected problem families get deeper pack-specific context.
+The first pack is `sunflower`.
 
-Current recommendation:
-
-- keep `ode-to-erdos` as a focused implementation seed
-- let `erdos-problems` become the broader public shell above it
-
-## Repository Shape
+## Repository shape
 
 ```text
 erdos-problems/
@@ -129,94 +57,76 @@ erdos-problems/
     cli/
     commands/
     atlas/
-    dossiers/
-    packs/
-    reporting/
-    adapters/
+    runtime/
+    upstream/
   problems/
-    20/
-    536/
-    856/
-    857/
+    <id>/
   packs/
     sunflower/
-  schemas/
-  templates/
+      README.md
+      problems/<id>/
+      compute/<id>/
+  data/upstream/erdosproblems/
   docs/
 ```
 
-## Runtime Model
-
-Local runtime folder:
-
-- `.erdos/`
+## Runtime model
 
-Suggested shape:
+Local workspace runtime:
 
 ```text
 .erdos/
-  config.json
   state.json
   current-problem.json
-  problems.json
-  runs/
+  upstream/erdosproblems/
+  scaffolds/<id>/
+  pulls/<id>/
+    artifacts/
+    literature/
   registry/
-  checkpoints/
-  literature/
 ```
 
-## Problem Entry Model
+## Dossier model
 
-Every problem directory should contain:
-
-```text
-problems/<id>/
-  problem.yaml
-  STATEMENT.md
-  REFERENCES.md
-  EVIDENCE.md
-  FORMALIZATION.md
-  CHECKPOINTS/
-```
+Each canonical seeded dossier contains:
+- `problem.yaml`
+- `STATEMENT.md`
+- `REFERENCES.md`
+- `EVIDENCE.md`
+- `FORMALIZATION.md`
 
-Not every problem needs a deep route board immediately.
+## Pull model
 
-But every problem should have:
+`erdos pull problem <id>` creates a root pull bundle with two sub-lanes:
+- `artifacts/`
+- `literature/`
 
-- source URL
-- normalized statement
-- status
-- tags
-- references
-- evidence links
+`erdos pull artifacts <id>` and `erdos pull literature <id>` target those lanes directly.
 
-## Harness Model
+## Maintainer model
 
-The first harness pack should be:
+`erdos maintainer seed problem <id>` promotes a pulled bundle into a canonical local dossier.
 
-- `packs/sunflower/`
+This is the disciplined bridge from:
+- public upstream/site truth
+- to local canonical dossier truth
 
-That pack should preserve:
+## Sunflower pack model
 
-- open problem
-- active route
-- route breakthrough
-- problem solved
+The sunflower pack now explicitly covers the quartet:
+- `20`: strong sunflower core
+- `857`: weak sunflower core
+- `536`: natural-density LCM analogue
+- `856`: harmonic-density LCM analogue
 
-And also:
+Pack assets may include:
+- pack README context
+- per-problem context files
+- compute packets for deep problems
 
-- tickets
-- gates
-- atoms
-- ready queue
-- compute lane awareness
-- generated checkpoints
-- literature mapping
-
-## CLI Shape
-
-### Atlas commands
+## CLI shape
 
+Atlas:
 ```bash
 erdos problem list
 erdos problem show 857
@@ -225,70 +135,43 @@ erdos cluster list
 erdos cluster show sunflower
 ```
 
-### Dossier commands
-
+Workspace and artifacts:
 ```bash
-erdos dossier show 857
-erdos dossier build 857
-erdos checkpoints sync 857
-erdos report build 857
+erdos workspace show
+erdos problem artifacts 857 --json
+erdos scaffold problem 857
+erdos bootstrap problem 857
 ```
 
-### Sunflower harness commands
-
+Pull and maintainer flow:
 ```bash
-erdos sunflower setup 857
-erdos sunflower warnings 857
-erdos sunflower pass 857
-erdos sunflower frontier 857
-erdos sunflower status 857
+erdos pull problem 1
+erdos pull artifacts 857
+erdos pull literature 857 --include-site
+erdos maintainer seed problem 1 --from-pull .erdos/pulls/1 --cluster number-theory
 ```
 
-`erdos sunflower status` should be the public-shell place where route state and
-compute posture meet:
-
-- active route
-- breakthrough state
-- packaged compute lane
-- next compute action
-- whether paid approval is required
-
-## Content Policy
-
-The repo should not blindly mirror `erdosproblems.com` text.
-
-Safer default:
-
-- source URL
-- citation
-- normalized summary of the problem
-- our own dossier and evidence materials
-
-This keeps the repo publishable and reduces licensing ambiguity.
-
-## Launch Strategy
-
-### Phase 1
-
-- create repo
-- scaffold CLI
-- add core schema
-- seed sunflower cluster
-- ship atlas-only read commands
-
-### Phase 2
-
-- add dossier generation
-- add checkpoint shelf
-- add sunflower harness commands
-
-### Phase 3
+Sunflower pack:
+```bash
+erdos sunflower status 20
+erdos sunflower status 536
+erdos sunflower status 857 --json
+```
 
-- deepen route-level automation for sunflower family
-- expand to other Erdős clusters carefully
+## Canonical truth split
 
-## First Public Promise
+Public upstream truth:
+- `teorth/erdosproblems`
+- `data/problems.yaml`
+- `erdosproblems.com`
 
-The first public promise should be:
+Local atlas truth:
+- `problems/<id>/problem.yaml`
+- dossier markdown files beside it
 
-> `erdos-problems` is a CLI atlas and staged research harness for Paul Erdős problems, with the sunflower family as the first deeply integrated pack.
+Local harness truth:
+- research state
+- checkpoints
+- pack context
+- compute packets
+- workspace pull/scaffold artifacts

package/docs/RESEARCH_LOOP.md

@@ -0,0 +1,47 @@
+# Research Loop
+
+`erdos-problems` now carries the staged research loop that was proven out in the sunflower lab and refined by the `.gpd` flow in `longevity-research`.
+
+## Runtime layout
+
+Workspace runtime files live under `.erdos/`:
+- `config.json`
+- `state.json`
+- `STATE.md`
+- `QUESTION-LEDGER.md`
+- `checkpoints/CHECKPOINTS.md`
+- `checkpoints/CHECKPOINTS.json`
+- `registry/preflight/`
+- `registry/compute/`
+
+## Core loop
+
+1. Select or bootstrap a problem.
+2. Sync state.
+3. Run preflight.
+4. Set continuation mode.
+5. Sync checkpoints.
+6. Pull or scaffold artifacts.
+7. Work the active route.
+8. Sync checkpoints again at honest boundaries.
+
+## Commands
+
+```bash
+erdos problem use 857
+erdos state sync
+erdos preflight
+erdos continuation use route
+erdos checkpoints sync
+erdos workspace show
+```
+
+## Status ladder
+
+The public package uses the same ladder we converged on in the lab:
+- open problem
+- active route
+- route breakthrough
+- problem solved
+
+The key rule is that route breakthroughs are never silently inflated into solved-problem claims.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "erdos-problems",
-  "version": "0.1.4",
+  "version": "0.1.6",
   "description": "CLI atlas and staged research harness for Paul Erdos problems.",
   "type": "module",
   "bin": {

package/packs/sunflower/README.md

@@ -2,12 +2,14 @@
 
 The first deep harness pack in `erdos-problems`.
 
-Current seed cluster:
+Sunflower core quartet:
 - `20`
 - `536`
 - `856`
 - `857`
 
-Deep-harness problems now:
-- `20`
-- `857`
+Profiles:
+- `20`: deep harness, strong / uniform core
+- `857`: deep harness, weak / non-uniform core
+- `536`: dossier bridge, natural-density LCM analogue
+- `856`: dossier bridge, harmonic-density LCM analogue

package/packs/sunflower/compute/20/u3_uniform_transfer_window_v0.yaml

@@ -0,0 +1,16 @@
+lane_id: u3_uniform_transfer_window_v0
+problem_id: "20"
+cluster: sunflower
+question: Can the uniform k=3 lane be frozen into a transfer-ready reduction packet that talks cleanly to the weak 857 export route?
+claim_level_goal: Verified
+status: ready_for_local_scout
+price_checked_local_date: 2026-03-25
+recommendation: local_scout_first
+approval_required: false
+summary: Problem 20 now has a frozen local-scout packet for the uniform k=3 transfer window. The job is to verify a small explicit reduction bundle before any paid compute rung is considered.
+source_repo: /Volumes/Code_2TB/code/sunflower-coda/repo
+public_feature: uniform_k3_frontier
+rungs:
+  - label: local_scout
+    mode: local
+    goal: verify the reduction packet, pull artifacts, and checkpoint the first honest route result

package/packs/sunflower/problems/20/CONTEXT.md

@@ -0,0 +1,14 @@
+# Problem 20 Sunflower Context
+
+Role in family:
+- strong / uniform sunflower core
+- deep-harness sibling to the weak sunflower program on 857
+
+Bootstrap focus:
+- start from the uniform k=3 lane
+- keep explicit bridges back to 857 whenever a route or method transfers cleanly
+
+Frontier framing:
+- the live goal is not “solve strong sunflower in one jump”
+- the live goal is to package the uniform k=3 lane into a recurrence-ready and transfer-ready route
+- checkpoint every route breakthrough before widening the public claim surface

package/packs/sunflower/problems/20/context.yaml

@@ -0,0 +1,31 @@
+problem_id: "20"
+family_role: strong_sunflower_core
+harness_profile: deep
+default_active_route: uniform_k3_frontier
+bootstrap_focus: Frame the strong or uniform k=3 sunflower lane as the direct sibling frontier to the weak 857 route.
+route_story: Keep the uniform k=3 lane explicit, preserve every clean bridge to problem 857, and do not let local route work masquerade as full strong-sunflower closure.
+frontier_label: uniform_k3_frontier
+frontier_detail: Tighten the uniform k=3 route into a recurrence-ready and transfer-ready reduction packet that can talk honestly to the weak 857 export program.
+checkpoint_focus: Record honest uniform-family route claims, bridge notes to 857, and any reusable reductions without overstating global closure.
+next_honest_move: Pull the uniform k=3 dossier, sync checkpoints, and pressure the smallest recurrence-facing or reduction-facing obligation.
+related_core_problems:
+  - "857"
+  - "536"
+  - "856"
+literature_focus:
+  - strong sunflower problem
+  - k-uniform families
+  - k=3
+  - Erdos-Rado sunflower
+artifact_focus:
+  - uniform-family reductions
+  - k=3 active route notes
+  - bridge notes to 857
+question_ledger:
+  open_questions:
+    - Which uniform k=3 reduction is the next smallest honest route step?
+    - What bridge result would cleanly transfer from problem 20 back into the weak 857 program?
+  active_route_notes:
+    - Keep the active route stated in uniform-family language, not generic sunflower shorthand.
+  route_breakthroughs:
+    - If a route breakthrough lands, checkpoint it before widening to strong-sunflower victory language.

package/packs/sunflower/problems/536/CONTEXT.md

@@ -0,0 +1,9 @@
+# Problem 536 Sunflower Context
+
+Role in family:
+- natural-density LCM analogue inside the sunflower quartet
+- dossier-first bridge problem rather than a deep theorem harness today
+
+Bootstrap focus:
+- keep the combinatorial relation to 857 explicit
+- collect number-theoretic translations, references, and bridge artifacts

package/packs/sunflower/problems/536/context.yaml

@@ -0,0 +1,17 @@
+problem_id: "536"
+family_role: natural_density_lcm_analogue
+harness_profile: dossier_bridge
+default_active_route: natural_density_lcm_bridge
+bootstrap_focus: Treat 536 as the number-theoretic natural-density analogue that should stay tightly linked to weak sunflower progress.
+related_core_problems:
+  - "857"
+  - "856"
+  - "20"
+literature_focus:
+  - lcm analogue
+  - natural density
+  - weak sunflower transfer
+artifact_focus:
+  - analogue statements
+  - bridge notes to 857
+  - number-theoretic translation artifacts

package/packs/sunflower/problems/856/CONTEXT.md

@@ -0,0 +1,9 @@
+# Problem 856 Sunflower Context
+
+Role in family:
+- harmonic-density LCM analogue inside the sunflower quartet
+- dossier-first bridge problem with direct implications for 857-style exponents
+
+Bootstrap focus:
+- keep the exponent-transfer story explicit
+- collect references and structured bridge artifacts for future deepening

package/packs/sunflower/problems/856/context.yaml

@@ -0,0 +1,17 @@
+problem_id: "856"
+family_role: harmonic_density_lcm_analogue
+harness_profile: dossier_bridge
+default_active_route: harmonic_density_lcm_bridge
+bootstrap_focus: Treat 856 as the harmonic-density LCM analogue whose exponent behavior should stay synchronized with weak sunflower progress.
+related_core_problems:
+  - "857"
+  - "536"
+  - "20"
+literature_focus:
+  - harmonic density
+  - lcm analogue
+  - exponent transfer to 857
+artifact_focus:
+  - analogue statements
+  - density bridge notes
+  - exponent-comparison artifacts

package/packs/sunflower/problems/857/CONTEXT.md

@@ -0,0 +1,14 @@
+# Problem 857 Sunflower Context
+
+Role in family:
+- weak / non-uniform sunflower core
+- current deep-harness flagship problem in the public atlas
+
+Bootstrap focus:
+- preserve the live route framing in the scaffold
+- keep literature, artifact, and compute packets aligned with the active frontier rather than generic sunflower chatter
+
+Frontier framing:
+- preserve the active export/compression route as the live frontier
+- keep the recurrence-facing remainder terms explicit in every checkpoint
+- do not blur route breakthroughs into claims that the open problem is solved

package/packs/sunflower/problems/857/context.yaml

@@ -0,0 +1,32 @@
+problem_id: "857"
+family_role: weak_sunflower_core
+harness_profile: deep
+default_active_route: anchored_selector_linearization
+bootstrap_focus: Start from the weak sunflower asymptotic route and preserve the active export/compression frontier in the scaffold.
+route_story: Keep the weak sunflower route pointed at recurrence-facing export and compression work, and preserve the distinction between route breakthroughs and full problem closure.
+frontier_label: anchored_selector_linearization
+frontier_detail: Preserve the live export/compression frontier and keep the recurrence-facing remainder terms explicit rather than flattening them into generic sunflower notes.
+checkpoint_focus: Record route breakthroughs, keep the open-problem versus active-route distinction sharp, and preserve the live remainder/compression obligations in every checkpoint.
+next_honest_move: Sync checkpoints, then keep compressing the explicit O1a export into a true recurrence-facing step.
+related_core_problems:
+  - "20"
+  - "536"
+  - "856"
+literature_focus:
+  - weak sunflower problem
+  - m(n,3)
+  - asymptotic routes
+  - container and export methods
+artifact_focus:
+  - active route notes
+  - formal frontier checkpoints
+  - compute lane packets when frozen
+question_ledger:
+  open_questions:
+    - Which explicit compression step is the next honest reduction on the active weak-sunflower route?
+    - Which exported remainder term is still dominating the recurrence-facing frontier?
+  active_route_notes:
+    - Keep active-route notes tied to the explicit export/compression frontier, not generic sunflower chatter.
+    - Preserve the distinction between route breakthrough and full-problem solution in every artifact.
+  route_breakthroughs:
+    - When a route breakthrough lands, checkpoint it before selecting the next active route.

package/problems/20/problem.yaml

@@ -20,14 +20,15 @@ prize:
   display: $1000
 related_problems:
   - "857"
+  - "536"
+  - "856"
 family_tags:
   - sunflower
   - uniform-families
 harness:
   depth: deep
 statement:
-  short: Determine the strong sunflower threshold for k-uniform set systems, with
-    the k=3 lane as the immediate active frontier.
+  short: Determine the strong sunflower threshold for k-uniform set systems, with the uniform k=3 lane as the immediate active frontier.
   normalized_md_path: STATEMENT.md
 references_path: REFERENCES.md
 evidence_path: EVIDENCE.md
@@ -36,3 +37,8 @@ formalization:
   status: active
   upstream_state: yes
   upstream_last_update: 2025-08-31
+research_state:
+  open_problem: true
+  active_route: uniform_k3_frontier
+  route_breakthrough: false
+  problem_solved: false