Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 6 additions & 7 deletions AGENTS.md
Original file line number Diff line number Diff line change
Expand Up @@ -17,9 +17,9 @@ Authority and invariants:
variable.
- Fiqs gate internal core propagation according to Fickian field motion. That
is timing, not gradient descent.
- UCNS integration is suspended until a PTCNA-specific higher-gonol producer
profile exists. Do not activate archived surfaces such as `a0_safe`,
`UCNSObject`, or `factor_search`.
- UCNS integration is active only for the exact pinned `157x7x7x53` candidate
receipt. Other shapes remain suspended. Do not activate archived surfaces
such as `a0_safe`, `UCNSObject`, or `factor_search`.
- EDCM measurement authority belongs to `The-Interdependency/edcm`. PTCNA may
consume an explicitly injected measurement provider; it must not maintain a
shadow EDCM implementation.
Expand All @@ -36,9 +36,8 @@ Authority and invariants:
representative workload, comparator, exact metrics and aggregation,
thresholds, resource bounds, stopping rules, and failure propagation before
inspecting outcomes. This freeze governs the verdict, not construction.
- Build PTCNA-local prime and ring structures under local provenance. Claiming
those structures are UCNS-produced remains suspended until an exact pinned
UCNS receipt exists.
- Build non-receipted PTCNA prime and ring structures under local provenance.
Only the exact receipt-covered state may carry UCNS candidate provenance.
- Record terminal status before repair: `FALSIFIED`, `SURVIVED — not proved`,
or `UNRESOLVED`. A simpler recovery method may supply the fallback; it does
not invalidate the architecture that selected the discovery path.
Expand Down Expand Up @@ -78,6 +77,6 @@ collection check, ratios gate, wheel smoke test, and downstream

## hmmm

- A PTCNA-specific UCNS higher-gonol producer identity does not yet exist.
- Continuous seven-fold geometry is not established by the candidate receipt.
- Training and timing behavior across the complete four-layer seam remains
unfalsified under sustained load.
4 changes: 4 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,10 @@

## Unreleased

- Consumed the exact UCNS `157×7×7×53` candidate-state receipt from merged
producer commit `b7b6f35cce69c273860923489a1c8b5372d14eb0`.
- Added independent state materialization verification and fail-closed receipt
tamper rejection; shapes outside the receipt remain locally attributed.
- Added a public `PTCNAEngine` receipt spanning neural, circle, seed, and core.
- Added a deterministic hashed-linear fallback behind the same task interface.
- Added explicit, attributed fallback routing; target failures still raise by
Expand Down
23 changes: 21 additions & 2 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,23 @@ continued = runtime.infer("question", fallback_on_error=True)
runtime.reward(continued, outcome=1.0)
```

The default target consumes the bundled UCNS candidate receipt pinned to
`The-Interdependency/ucns@b7b6f35cce69c273860923489a1c8b5372d14eb0`.
It independently materializes and verifies the exact `157×7×7×53` positive-zero
state before inference. Inspect or validate a persisted receipt explicitly:

```python
import json
from pathlib import Path
from ptcna import consume_ucns_receipt

status = consume_ucns_receipt(json.loads(Path("ucns-receipt.json").read_text()))
assert status.adapter_active
```

Receipt success establishes compatible construction only. It does not select
continuous seven-fold geometry or establish efficacy or production privacy.

Freeze a representative labeled workload before inspecting outcomes:

```python
Expand Down Expand Up @@ -86,14 +103,16 @@ frozen and executed, whether PTCNA works remains `hmmm`.
## Status

Alpha (`0.1.1`). All four layers import and the repository test suite passes.
The exact default state now carries producer-validated UCNS candidate
provenance; nonmatching shapes remain explicitly suspended and PTCNA-local.
The layer boundary is now executable rather than only descriptive:

- `ptcna.circle.CircleTensor` is the one circle type used by circle, seed, and
core composition.
- `ptcna.neural.NeuralScalar` is the only PTCNA type that owns reverse-mode
gradients; structural hosts carry it opaquely.
- UCNS integration reports a typed `suspended` state and fails closed until a
reviewed PTCNA-specific higher-gonol producer profile exists.
- UCNS integration validates the exact candidate receipt for the default shape;
mismatched shapes report typed `suspended` state and remain locally attributed.
- EDCM remains an external authority. `ZetaEngine` accepts an explicitly
injected measurement provider; PTCNA contains no shadow EDCM module.
- `PTCNAEngine` joins the live neural engine to the complete local core and
Expand Down
12 changes: 9 additions & 3 deletions docs/architecture.md
Original file line number Diff line number Diff line change
Expand Up @@ -93,9 +93,8 @@ outcomes. That preregistration governs the verdict, not whether construction may
begin.

Record `FALSIFIED`, `SURVIVED — not proved`, or `UNRESOLVED` before repair
or criterion change. PTCNA-local prime and ring structures may be constructed
now; only a claim that they are UCNS-produced remains suspended until an exact
pinned UCNS receipt exists.
or criterion change. The exact default initialization may carry its pinned UCNS
candidate provenance; other prime and ring structures remain PTCNA-local.

The executable boundary lives in `ptcna.runtime`: `PTCNAEngine` reports all
four live layers, `HashedLinearFallback` remains separately identified, and
Expand All @@ -105,6 +104,13 @@ enabled. `ptcna.evaluation` accepts only an immutable, digest-bearing
representative workload is bundled; selecting one remains evidence work, not a
construction prerequisite.

The default core shape consumes a bundled receipt produced by exact UCNS merge
`b7b6f35cce69c273860923489a1c8b5372d14eb0`. UCNS owns the candidate state
contract and provenance; PTCNA validates that contract and independently checks
the materialized byte stream. This replaces the former blanket suspension only
for `157×7×7×53`. Other shapes retain typed suspension, and integration success
does not transfer proof, efficacy, geometry-selection, or privacy standing.

Semantic authority for discovery-before-recovery remains with
`The-Interdependency/metapat`; this correction consulted
`metapat@53315e30c54aba881a5b48cbf395890e83ab05c5`,
Expand Down
6 changes: 3 additions & 3 deletions docs/chapter-5.md
Original file line number Diff line number Diff line change
Expand Up @@ -140,6 +140,6 @@ class names in the core layer remain a standing compatibility choice — right
layer, published API, revisitable, and recorded so that revisiting it starts
from evidence rather than surprise.

**hmmm — a reviewed PTCNA-specific UCNS higher-gonol producer profile does not
yet exist, so UCNS integration remains typed and suspended; sustained-load
behavior across the complete four-layer seam is still unfalsified.**
**hmmm — the exact default initialization now has a reviewed UCNS candidate
receipt, but continuous seven-fold geometry and sustained-load behavior across
the complete four-layer seam remain unfalsified.**
18 changes: 9 additions & 9 deletions docs/work-graphs/repository-plan-report.json
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@
},
"source": {
"branch": "main",
"commit": "4509d33419aa25e6a9cfef415055e378b8f37edc",
"commit": "d69c94a4a43cf4b5b286af047217c5624a8d4f3f",
"generated_at": "2026-08-17",
"note": "This report describes the executable PTCNA 0.1.1 runtime at the cited main commit plus the current repo-owned plan. The 2026-08-17 policy correction removes an unauthorized evaluation-as-permission gate: intended architecture construction, dependable fallback construction, and evaluation are distinct work streams. Operational-shell and report-only maintenance do not upgrade runtime, mathematical, measurement, or architectural evidence standing."
},
Expand All @@ -34,7 +34,7 @@
"UCNS proof or representation status does not establish PTCNA learning efficacy",
"EDCM measurement results do not become PTCNA-owned evidence merely because a caller injects them",
"a functioning four-layer implementation does not establish that prime structure, current ring constants, or current weight topology are necessary or superior",
"suspended integration remains suspended until an exact producer contract is independently available and reviewed",
"a validated UCNS candidate-state receipt transfers neither proof status nor PTCNA efficacy standing",
"upstream evidence and simpler-baseline performance inform attribution, evaluation, and fallback routing; they neither grant nor deny permission to construct the selected PTCNA architecture"
]
},
Expand All @@ -50,7 +50,7 @@
},
"status": {
"state": "active alpha runtime; explicit target, fallback, and frozen-verdict interfaces are implemented under separate evidence identities",
"current_claim": "PTCNA 0.1.1 implements one four-layer runtime in which reverse-mode gradients are owned by the neural layer while circle, seed, and core remain non-differentiating structural hosts. The unreleased package surface adds an attributed four-layer target receipt, an independently test-backed hashed-linear fallback, explicit failover, and immutable evaluation plans. UCNS higher-gonol attribution remains suspended, EDCM remains external, and the critical 'does it work?' verdict remains unearned until a representative plan is frozen and executed.",
"current_claim": "PTCNA implements one four-layer runtime in which reverse-mode gradients are owned by the neural layer while circle, seed, and core remain non-differentiating structural hosts. The exact default 157x7x7x53 initialization consumes and independently verifies a receipt from UCNS merge b7b6f35cce69c273860923489a1c8b5372d14eb0. The hashed-linear fallback remains separately attributed, EDCM remains external, and the critical 'does it work?' verdict remains unearned until a representative plan is frozen and executed.",
"latest_main_milestone": {
"commit": "4509d33419aa25e6a9cfef415055e378b8f37edc",
"description": "merged the reconciled four-layer v0.1.1 runtime with one CircleTensor, neural-only reverse-mode scalar ownership, fail-closed UCNS integration, and external EDCM measurement injection"
Expand All @@ -69,8 +69,8 @@
},
{
"surface": "UCNS integration boundary",
"status": "typed suspended fail-closed adapter",
"boundary": "no PTCNA-specific higher-gonol producer profile is currently authorized or consumed"
"status": "exact receipt consumer active for 157x7x7x53; all other shapes suspended",
"boundary": "candidate construction compatibility does not select geometry or transfer proof, efficacy, or privacy standing"
},
{
"surface": "external measurement boundary",
Expand Down Expand Up @@ -98,7 +98,7 @@
"freeze a representative 'does it work?' EvaluationPlan before outcome inspection, including workload, training epochs and reward, exact thresholds, repetitions, resource bounds, and failure propagation; the comparator identity, post-training metric, aggregation, stopping rule, and receipt schema are now executable",
"execute the preserved plan without repair or criterion change and record FALSIFIED, SURVIVED — not proved, or UNRESOLVED",
"qualify whether the hashed-linear fallback preserves useful operation on the representative workload rather than inferring utility from deterministic contract tests",
"consume exact UCNS producer receipts when available for UCNS attribution while allowing locally attributed PTCNA prime and ring construction to continue",
"preserve exact UCNS receipt validation while evaluating the candidate architecture independently",
"resolve whether the embedded ZetaEngine remains a bounded PTCNA-local measurement-to-learning adapter or should be removed in favor of separate ZFAE authority before further Zeta development",
"test sustained-load and long-horizon state behavior through the explicit target/fallback runtime boundary"
],
Expand All @@ -120,14 +120,14 @@
}
],
"blocked": [
"claiming that PTCNA-local prime or ring construction is UCNS-produced is blocked until an exact reviewed PTCNA-specific UCNS producer receipt exists",
"claiming continuous seven-fold UCNS geometry from the candidate initialization receipt remains blocked",
"claiming that PTCNA works, learns usefully, or is necessary or superior is blocked until the frozen evaluation earns that standing",
"ZetaEngine ownership migration or expansion is blocked until the PTCNA-local versus separate-ZFAE authority boundary is frozen"
],
"cross_repository_relations": [
{
"repository": "The-Interdependency/ucns",
"relation": "mathematical and relational producer authority; exact UCNS attribution requires a reviewed producer receipt, while PTCNA-local construction proceeds under local provenance",
"relation": "mathematical and relational producer authority for the exact pinned 157x7x7x53 candidate-state receipt; PTCNA owns its consumption and inference",
"authority_transfer": false,
"proof_status_transfer": false
},
Expand Down Expand Up @@ -171,7 +171,7 @@
"quality_workflow": ".github/workflows/quality.yml"
},
"hmmm": [
"the exact PTCNA-specific UCNS producer profile and receipt for any structure later claimed as UCNS-produced",
"continuous seven-fold geometry beyond the exact candidate-state receipt",
"whether the constructed PTCNA architecture works under the frozen representative evaluation",
"the exact representative workload and externally justified training epochs, reward, target threshold, comparator-deficit allowance, repetitions, resource bounds, and backend-error status for the implemented evaluation API",
"whether the implemented hashed-linear fallback preserves useful operation on that representative workload",
Expand Down
14 changes: 12 additions & 2 deletions ptcna/__init__.py
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
# ratios: loc_comments=31:51 imports_exports=4:1 calls_definitions=1:0
# ratios: loc_comments=41:51 imports_exports=4:1 calls_definitions=1:0
"""PTCNA — Prime Tensor Circled Neural Architecture.

One architecture, four layers. Each layer's tensors divide into the next; every
Expand All @@ -20,9 +20,14 @@

from . import neural, circle, seed, core # noqa: F401
from .ucns_integration import (
UCNS_PRODUCER_COMMIT,
UCNS_RECEIPT_SHA256,
UCNSIntegrationState,
UCNSIntegrationStatus,
UCNSIntegrationSuspended,
UCNSReceiptError,
consume_ucns_receipt,
load_bundled_ucns_receipt,
require_ucns_integration,
ucns_integration_status,
)
Expand Down Expand Up @@ -76,9 +81,14 @@
"circle",
"seed",
"core",
"UCNS_PRODUCER_COMMIT",
"UCNS_RECEIPT_SHA256",
"UCNSIntegrationState",
"UCNSIntegrationStatus",
"UCNSIntegrationSuspended",
"UCNSReceiptError",
"consume_ucns_receipt",
"load_bundled_ucns_receipt",
"ucns_integration_status",
"require_ucns_integration",
"PTCNAEngine",
Expand All @@ -90,4 +100,4 @@
"evaluate",
"__version__",
]
# ratios: loc_comments=31:51 imports_exports=4:1 calls_definitions=1:0
# ratios: loc_comments=41:51 imports_exports=4:1 calls_definitions=1:0
9 changes: 5 additions & 4 deletions ptcna/circle/compose.py
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
# ratios: loc_comments=33:56 imports_exports=5:3 calls_definitions=14:2
# ratios: loc_comments=33:57 imports_exports=5:3 calls_definitions=14:2
"""Compose neural payloads into a non-differentiating circle tensor.

Usage:
Expand All @@ -8,8 +8,9 @@
circle = compose_circle(["n0", "n1", "n2"], routing_step=2)
assert circle.at(circle.anchor_order[0]) == "n0"

The routing is local PTCNA structure. It is deliberately not labeled as UCNS
while the PTCNA-specific UCNS higher-gonol integration is suspended.
The routing is local PTCNA structure. The exact default aggregate may carry its
separate UCNS candidate-state receipt; this constructor does not grant that
provenance to arbitrary circles.
"""
from __future__ import annotations

Expand Down Expand Up @@ -106,4 +107,4 @@ def compose_circle(


__all__ = ["star_polygon_order", "compose_circle"]
# ratios: loc_comments=33:56 imports_exports=5:3 calls_definitions=14:2
# ratios: loc_comments=33:57 imports_exports=5:3 calls_definitions=14:2
11 changes: 5 additions & 6 deletions ptcna/core/prime_core/PROVENANCE.md
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@
> - **Current ownership:** `Scalar`, fixed seven-by-seven composition, and the
> UCNS carrier claims below are historical. Current code uses
> `ptcna.neural.NeuralScalar`, shared `ptcna.circle.CircleTensor`, variable
> composition counts, local identities, and a typed suspended UCNS boundary.
> composition counts, local identities, and an exact-shape UCNS receipt boundary.

# CLAUDE.md — prime_core (PTCA three-stratum core)

Expand Down Expand Up @@ -97,11 +97,10 @@ guard (5), routing steps (6). Stdlib `unittest` only — no pytest dependency.
- **Validator resolved.** PTCNA vendors the bounded skill-lib collection and
collects MODULE_BUILD, CONTRACTS, CHECKS, and BOUNDARIES declarations into
`ptcna_msdmd.ts`.
- **UCNS binding remains suspended.** The available UCNS source does not
expose the reviewed PTCNA-specific higher-gonol producer profile this layer
needs. Current code does not import archived `a0_safe`, `UCNSObject`, or
`factor_search` surfaces; it returns typed suspension and uses explicitly
local identities.
- **UCNS binding is candidate scoped.** The default `157x7x7x53` initialization
consumes an exact reviewed UCNS receipt. Current code does not import archived
`a0_safe`, `UCNSObject`, or `factor_search` surfaces; unmatched shapes return
typed suspension and use explicitly local identities.
- **Packaging resolved.** `prime_core` ships under
`ptcna.core.prime_core` in the `ptcna` distribution.
- **Seam under load.** The descend/ascend split is the whole experiment;
Expand Down
2 changes: 1 addition & 1 deletion ptcna/core/prime_core/__init__.py
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
Fiqs host payload vectors opaquely. The core, seeds, circles, and fiqs are all
non-differentiating. Neural-owned payloads may be supplied with
``build_core(..., payload_factory=...)`` without moving gradient ownership out
of ``ptcna.neural``. UCNS integration is explicitly suspended.
of ``ptcna.neural``. UCNS attribution is exact-shape and receipt scoped.
"""
from __future__ import annotations

Expand Down
Loading