mirror of
https://github.com/allaunthefox/Research-Stack.git
synced 2026-07-31 03:05:21 +00:00
439 lines
18 KiB
Python
439 lines
18 KiB
Python
#!/usr/bin/env python3
|
|
"""Re-evaluate DD compression targets against current implementation receipts.
|
|
|
|
This is a receipt-bearing target review. It does not recompress data. It
|
|
reads the local decision-diagram, topology, route-card, KV, citation, and
|
|
singular-chart receipts and classifies which configuration knobs are currently:
|
|
|
|
* promotion-backed by measured bytes,
|
|
* useful as pruning / control configuration, or
|
|
* still only proposal surface until an encode/decode/hash evaluator exists.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import hashlib
|
|
import json
|
|
from pathlib import Path
|
|
from typing import Any
|
|
|
|
|
|
REPO = Path(__file__).resolve().parents[2]
|
|
SHIM = REPO / "4-Infrastructure" / "shim"
|
|
OUT = SHIM / "dd_target_implementation_reevaluation_receipt.json"
|
|
CURRICULUM_OUT = SHIM / "dd_target_implementation_reevaluation_curriculum.jsonl"
|
|
|
|
GENERATED_AT = "2026-05-08T00:00:00+00:00"
|
|
HUTTER_ENWIK9_TARGET_BYTES = 109_685_197
|
|
|
|
RECEIPT_PATHS = {
|
|
"dimensional_shell_dd_probe": SHIM / "dimensional_shell_dd_probe_receipt.json",
|
|
"projectable_geometry_topology_model": SHIM
|
|
/ "projectable_geometry_topology_model_receipt.json",
|
|
"alphaevolve_dd_experiment_card_runner": SHIM
|
|
/ "alphaevolve_dd_experiment_card_receipt.json",
|
|
"non_euclidean_semantic_kv_prior": SHIM
|
|
/ "non_euclidean_semantic_kv_prior_receipt.json",
|
|
"citation_math_function_distillation": SHIM
|
|
/ "citation_math_function_distillation_receipt.json",
|
|
"singular_route_chart_equations": SHIM
|
|
/ "singular_route_chart_equations_receipt.json",
|
|
"compression_ratio_rederivation": SHIM
|
|
/ "compression_ratio_rederivation_receipt.json",
|
|
}
|
|
|
|
|
|
def stable_json(obj: Any) -> str:
|
|
return json.dumps(obj, sort_keys=True, separators=(",", ":"), ensure_ascii=True)
|
|
|
|
|
|
def sha256_bytes(data: bytes) -> str:
|
|
return hashlib.sha256(data).hexdigest()
|
|
|
|
|
|
def load_json(path: Path) -> dict[str, Any]:
|
|
return json.loads(path.read_text(encoding="utf-8"))
|
|
|
|
|
|
def rel(path: Path) -> str:
|
|
return str(path.relative_to(REPO))
|
|
|
|
|
|
def receipt_hash(path: Path, data: dict[str, Any]) -> str:
|
|
for key in (
|
|
"stable_topology_model_hash_sha256",
|
|
"stable_shell_dd_hash_sha256",
|
|
"receipt_hash",
|
|
):
|
|
value = data.get(key)
|
|
if isinstance(value, str):
|
|
return value
|
|
return sha256_bytes(path.read_bytes())
|
|
|
|
|
|
def source_receipts(receipts: dict[str, dict[str, Any]]) -> dict[str, Any]:
|
|
return {
|
|
name: {
|
|
"path": rel(RECEIPT_PATHS[name]),
|
|
"hash": receipt_hash(RECEIPT_PATHS[name], receipt),
|
|
"schema": receipt.get("schema", "unknown"),
|
|
}
|
|
for name, receipt in receipts.items()
|
|
}
|
|
|
|
|
|
def class_counts(cards: dict[str, Any]) -> dict[str, int]:
|
|
counts = cards.get("class_counts")
|
|
if isinstance(counts, dict):
|
|
return {str(key): int(value) for key, value in counts.items()}
|
|
derived: dict[str, int] = {}
|
|
for card in cards.get("cards", []):
|
|
cls = str(card.get("applicability_class", "unknown"))
|
|
derived[cls] = derived.get(cls, 0) + 1
|
|
return derived
|
|
|
|
|
|
def current_best(shell: dict[str, Any], topology: dict[str, Any]) -> dict[str, Any]:
|
|
shell_best = shell["summary"]["best_shell_adjusted_overall"]
|
|
topology_best = topology["best_approach"]["selected_route"]
|
|
return {
|
|
"slice": topology_best["slice"],
|
|
"route": topology["best_approach"]["route"],
|
|
"byte_transform": f"{topology_best['transform']} -> {topology_best['codec']}",
|
|
"source_bytes": topology_best["source_bytes"],
|
|
"raw_baseline_bytes": topology_best["raw_baseline_bytes"],
|
|
"xml_bz2_bytes": topology_best["compressed_bytes"],
|
|
"topology_or_shell_witness_bytes": topology_best["topology_witness_bytes"],
|
|
"modeled_total_bytes": topology_best["modeled_total_bytes"],
|
|
"modeled_ratio": topology_best["modeled_ratio"],
|
|
"remaining_margin_vs_raw_baseline_bytes": topology_best[
|
|
"gain_vs_raw_after_topology_bytes"
|
|
],
|
|
"witness_budget_before_losing_raw_bytes": topology_best[
|
|
"overhead_budget_before_losing_raw"
|
|
],
|
|
"projected_enwik9_total_bytes": shell_best[
|
|
"projected_enwik9_total_bytes"
|
|
],
|
|
"projected_hutter_gap_bytes": shell_best[
|
|
"hutter_target_gap_bytes_projected_enwik9"
|
|
],
|
|
"rehydration_status": shell_best["shell_status"],
|
|
"lawful": bool(topology_best["lawful"] and not shell_best["shell_status"]["nan0"]),
|
|
}
|
|
|
|
|
|
def implementation_status(receipts: dict[str, dict[str, Any]]) -> list[dict[str, Any]]:
|
|
shell = receipts["dimensional_shell_dd_probe"]
|
|
topology = receipts["projectable_geometry_topology_model"]
|
|
cards = receipts["alphaevolve_dd_experiment_card_runner"]
|
|
kv = receipts["non_euclidean_semantic_kv_prior"]
|
|
citation = receipts["citation_math_function_distillation"]
|
|
singular = receipts["singular_route_chart_equations"]
|
|
ratio = receipts["compression_ratio_rederivation"]
|
|
|
|
return [
|
|
{
|
|
"id": "dimensional_shell_dd_probe",
|
|
"status": "implemented_receipt_over_existing_measurements",
|
|
"promotion_authority": "measured route receipt inherited from reversible approach; shell only adds bounded witness",
|
|
"evidence": {
|
|
"route_count": shell["summary"]["route_count"],
|
|
"promoted_route_count": shell["summary"]["promoted_route_count"],
|
|
"lower_bound_pruned_count": shell["summary"][
|
|
"lower_bound_pruned_count"
|
|
],
|
|
"nan0_route_count": shell["summary"]["nan0_route_count"],
|
|
"shell_witness_bytes": shell["dd_policy"][
|
|
"shell_witness_bytes_per_non_raw_route"
|
|
],
|
|
},
|
|
"verdict": "promotable_when_margin_survives_witness",
|
|
},
|
|
{
|
|
"id": "projectable_geometry_topology_model",
|
|
"status": "implemented_topology_witness_model",
|
|
"promotion_authority": "existing xml_token -> bz2 byte measurement plus 16-byte witness budget",
|
|
"evidence": topology["best_approach"]["selected_route"],
|
|
"verdict": "current_best_control_plane",
|
|
},
|
|
{
|
|
"id": "alphaevolve_dd_experiment_card_runner",
|
|
"status": "implemented_search_card_prior",
|
|
"promotion_authority": "none until local candidate code runs through encode/decode/hash evaluator",
|
|
"evidence": {
|
|
"card_count": cards["card_count"],
|
|
"class_counts": class_counts(cards),
|
|
"full_program_collection": cards["validation"].get(
|
|
"full_program_collections", False
|
|
),
|
|
},
|
|
"verdict": "proposal_and_dashboard_surface",
|
|
},
|
|
{
|
|
"id": "citation_math_function_distillation",
|
|
"status": "implemented_operator_distillation",
|
|
"promotion_authority": "none; this is the route compiler algebra",
|
|
"evidence": {
|
|
"primary_function": citation["primary_function"]["name"],
|
|
"algebra_count": len(citation["dd_function_algebra"]),
|
|
"function_group_count": len(citation["function_groups"]),
|
|
},
|
|
"verdict": "configuration_basis",
|
|
},
|
|
{
|
|
"id": "singular_route_chart_equations",
|
|
"status": "implemented_equation_group",
|
|
"promotion_authority": "none unless a finite chart decodes and hashes exact bytes under measured byte count",
|
|
"evidence": {
|
|
"primary_function": singular["singular_route_chart"][
|
|
"primary_function"
|
|
],
|
|
"equation_count": len(singular["equations"]),
|
|
"dd_state_field_count": len(singular["dd_state_extension"]),
|
|
"receipt_hash": singular["receipt_hash"],
|
|
},
|
|
"verdict": "fail_closed_chart_layer",
|
|
},
|
|
{
|
|
"id": "non_euclidean_semantic_kv_prior",
|
|
"status": "implemented_prior_and_watchlist",
|
|
"promotion_authority": "none until byte-store / KV-cache routes have exact rehydration receipts",
|
|
"evidence": {
|
|
"prior_family_count": len(kv["prior_families"]),
|
|
"priority_watch_items": [
|
|
item["id"] for item in kv["priority_watch_items"]
|
|
],
|
|
"treefiddy_status": kv["local_treefiddy_status"]["status"],
|
|
},
|
|
"verdict": "proposal_surface_with_local_tree_bound",
|
|
},
|
|
{
|
|
"id": "compression_ratio_rederivation",
|
|
"status": "implemented_ratio_audit",
|
|
"promotion_authority": "ratio schema only; byte counts remain terminal authority",
|
|
"evidence": {
|
|
"claim_boundary": ratio["claim_boundary"],
|
|
"finance_claim_lut_lawful": ratio["finance_claim_lut"]["lawful"],
|
|
},
|
|
"verdict": "measurement_schema_guard",
|
|
},
|
|
]
|
|
|
|
|
|
def configuration_axes(best: dict[str, Any]) -> list[dict[str, Any]]:
|
|
remaining = int(best["remaining_margin_vs_raw_baseline_bytes"])
|
|
budget = int(best["witness_budget_before_losing_raw_bytes"])
|
|
return [
|
|
{
|
|
"id": "topology_witness_bytes",
|
|
"current_setting": 16,
|
|
"admissible_range_for_current_best_bytes": [0, budget],
|
|
"target_role": "bounded closure/control witness",
|
|
"reevaluation": "keep at 16 unless a new encoder earns more payload savings",
|
|
},
|
|
{
|
|
"id": "operator_basis_B_R_F_T_A_E_V",
|
|
"current_setting": "available_as_distilled_algebra",
|
|
"target_role": "configure pruning, routing, folding, transport, allocation, repair, and verification",
|
|
"reevaluation": "wire as evaluator stages; do not charge bytes unless an emitted receipt/witness is serialized",
|
|
},
|
|
{
|
|
"id": "singular_route_chart",
|
|
"current_setting": "available_not_wired_to_byte_evaluator",
|
|
"target_role": "contain singular/ambiguous/unbounded route regions",
|
|
"reevaluation": (
|
|
"use on failing or ambiguous branches first; applying extra chart "
|
|
f"bytes to the current best must stay under {remaining} bytes"
|
|
),
|
|
},
|
|
{
|
|
"id": "tinyenc_encrypted_kv_envelope",
|
|
"current_setting": "watch_item_only",
|
|
"target_role": "future compressed/encrypted byte-store route",
|
|
"reevaluation": "diagnostic until compressed bytes, query index overhead, and plaintext hash are receipted",
|
|
},
|
|
{
|
|
"id": "treekv_treefiddy_spine",
|
|
"current_setting": "local_treefiddy_found_prior_only",
|
|
"target_role": "bounded tree route spine and owner/depth guard",
|
|
"reevaluation": "proposal only until TreeKV merge preserves decode reachability and exact residual leaves",
|
|
},
|
|
{
|
|
"id": "alphaevolve_route_population",
|
|
"current_setting": "experiment_card_shape_only",
|
|
"target_role": "candidate generator and analysed/program counters",
|
|
"reevaluation": "feed candidates into the DD; never promote gallery score or novelty without local byte receipt",
|
|
},
|
|
]
|
|
|
|
|
|
def admissibility_verdicts(best: dict[str, Any]) -> list[dict[str, Any]]:
|
|
remaining = int(best["remaining_margin_vs_raw_baseline_bytes"])
|
|
return [
|
|
{
|
|
"route_or_overlay": "xml_token -> topology_witness_16b -> bz2",
|
|
"verdict": "promote_current_incumbent_for_small_slice",
|
|
"reason": "measured route survives exact shell/topology witness and beats raw+bz2 by 1105 bytes",
|
|
},
|
|
{
|
|
"route_or_overlay": "singular_chart_on_current_best",
|
|
"verdict": "prune_unless_chart_bytes_are_paid_by_new_savings",
|
|
"reason": f"any added chart receipt consumes the remaining {remaining}-byte margin",
|
|
},
|
|
{
|
|
"route_or_overlay": "B/R/F/T/A/E/V operator basis",
|
|
"verdict": "admissible_as_evaluator_configuration",
|
|
"reason": "operators organize route search and receipts, but do not themselves add byte evidence",
|
|
},
|
|
{
|
|
"route_or_overlay": "TreeKV + Tree Fiddy",
|
|
"verdict": "diagnostic_until_encoder_exists",
|
|
"reason": "local Tree Fiddy can bound tree depth, but no TreeKV byte route has been evaluated",
|
|
},
|
|
{
|
|
"route_or_overlay": "TinyEnc-style encrypted KV store",
|
|
"verdict": "diagnostic_until_store_receipt_exists",
|
|
"reason": "encryption, query index, and compression overhead must all be counted with plaintext rehydration",
|
|
},
|
|
{
|
|
"route_or_overlay": "AlphaEvolve route population",
|
|
"verdict": "proposal_generator_only",
|
|
"reason": "program and analysed counts are useful search controls, not compression evidence",
|
|
},
|
|
]
|
|
|
|
|
|
def next_target(best: dict[str, Any]) -> dict[str, Any]:
|
|
return {
|
|
"target_name": "configurable_bounded_route_evaluator",
|
|
"reframed_goal": "optimize route evaluation and pruning configuration before adding more serialized witness bytes",
|
|
"why": (
|
|
"The current best has only 1105 bytes of margin after the 16-byte "
|
|
"witness. New knobs should reject bad branches cheaply or propose "
|
|
"new byte-saving transforms; they should not be blindly appended to "
|
|
"the incumbent route."
|
|
),
|
|
"minimum_config_fields": [
|
|
"slice_id",
|
|
"candidate_route",
|
|
"backend_codec",
|
|
"topology_witness_bytes",
|
|
"singular_chart_enabled",
|
|
"singular_chart_bytes",
|
|
"treefiddy_spine_enabled",
|
|
"tinyenc_envelope_enabled",
|
|
"ratio_schema",
|
|
"incumbent_bytes",
|
|
"lower_bound_bytes",
|
|
"rehydration_hash",
|
|
],
|
|
"first_safe_implementation_step": (
|
|
"Add a configuration-matrix wrapper over existing receipts that "
|
|
"computes lower bounds and prunes overlays whose witness/chart/KV "
|
|
"cost would erase the incumbent margin before recompression."
|
|
),
|
|
"first_real_compression_step": (
|
|
"Evaluate one new payload-saving transform family at a time, starting "
|
|
"with bounded corpus-resolution/fascicle or tokenbook variants, then "
|
|
"charge any topology/singular/KV witness bytes in the terminal receipt."
|
|
),
|
|
"hard_target_bytes_enwik9": HUTTER_ENWIK9_TARGET_BYTES,
|
|
"diagnostic_target": "beat raw baseline per slice under explicit ratio_schema",
|
|
"current_gap_to_hard_target_projected_enwik9_bytes": best[
|
|
"projected_hutter_gap_bytes"
|
|
],
|
|
}
|
|
|
|
|
|
def curriculum_lines(receipt: dict[str, Any]) -> list[dict[str, Any]]:
|
|
lines: list[dict[str, Any]] = []
|
|
for item in receipt["implementation_status"]:
|
|
lines.append(
|
|
{
|
|
"type": "implementation_status",
|
|
"id": item["id"],
|
|
"status": item["status"],
|
|
"verdict": item["verdict"],
|
|
}
|
|
)
|
|
for item in receipt["configuration_axes"]:
|
|
lines.append(
|
|
{
|
|
"type": "configuration_axis",
|
|
"id": item["id"],
|
|
"current_setting": item["current_setting"],
|
|
"reevaluation": item["reevaluation"],
|
|
}
|
|
)
|
|
for item in receipt["admissibility_verdicts"]:
|
|
lines.append(
|
|
{
|
|
"type": "admissibility_verdict",
|
|
"id": item["route_or_overlay"],
|
|
"verdict": item["verdict"],
|
|
"reason": item["reason"],
|
|
}
|
|
)
|
|
return lines
|
|
|
|
|
|
def build_receipt() -> dict[str, Any]:
|
|
receipts = {name: load_json(path) for name, path in RECEIPT_PATHS.items()}
|
|
best = current_best(
|
|
receipts["dimensional_shell_dd_probe"],
|
|
receipts["projectable_geometry_topology_model"],
|
|
)
|
|
receipt: dict[str, Any] = {
|
|
"schema": "dd_target_implementation_reevaluation_v1",
|
|
"generated_at": GENERATED_AT,
|
|
"source_receipts": source_receipts(receipts),
|
|
"current_target": {
|
|
"primary_machine": "bounded_exact_route_compiler",
|
|
"hard_target_bytes_enwik9": HUTTER_ENWIK9_TARGET_BYTES,
|
|
"diagnostic_target": "beat raw baseline per slice under explicit ratio_schema",
|
|
"proof_surface": [
|
|
"exact decoded bytes",
|
|
"rehydration hash",
|
|
"measured compressed_total_bytes",
|
|
"explicit ratio_schema",
|
|
"bounded sidecar/witness/container/compute cost",
|
|
"fail-closed NaN0 metadata holds",
|
|
],
|
|
},
|
|
"current_best_implemented_route": best,
|
|
"implementation_status": implementation_status(receipts),
|
|
"configuration_axes": configuration_axes(best),
|
|
"admissibility_verdicts": admissibility_verdicts(best),
|
|
"reevaluated_next_target": next_target(best),
|
|
"claim_boundary": (
|
|
"This is a target and implementation reevaluation over local receipts. "
|
|
"It does not recompress data, prove optimality, or promote any route "
|
|
"without an encode/decode/hash/byte-count receipt."
|
|
),
|
|
}
|
|
preimage = {key: value for key, value in receipt.items() if key != "receipt_hash"}
|
|
receipt["receipt_hash"] = sha256_bytes(stable_json(preimage).encode("utf-8"))
|
|
return receipt
|
|
|
|
|
|
def main() -> None:
|
|
receipt = build_receipt()
|
|
OUT.write_text(json.dumps(receipt, indent=2, sort_keys=True) + "\n", encoding="utf-8")
|
|
lines = curriculum_lines(receipt)
|
|
CURRICULUM_OUT.write_text(
|
|
"".join(json.dumps(line, sort_keys=True) + "\n" for line in lines),
|
|
encoding="utf-8",
|
|
)
|
|
print(json.dumps({
|
|
"receipt": rel(OUT),
|
|
"curriculum": rel(CURRICULUM_OUT),
|
|
"receipt_hash": receipt["receipt_hash"],
|
|
"line_count": len(lines),
|
|
"current_best": receipt["current_best_implemented_route"]["route"],
|
|
"next_target": receipt["reevaluated_next_target"]["target_name"],
|
|
}, indent=2, sort_keys=True))
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|