capacity_readback

Correctable Public-Record Capacity

Exact finite contracts, controls, and archived candidate audits for the OPH capacity map. The canonical Pro5 producer is

N = log M_0(U_N),
F_set,r,epsilon(D) = {M_epsilon(q): q in Omega_tilde(r,D)},
M_0(q) = alpha(G_q).

The first line is the direct global proposal; the next two lines are its typed finite implementation.

The operational resolution, electroweak/Higgs bridge, and measured cosmological constant are independent downstream comparisons. They never define the direct map. The bounded counterfamily has verdict NOT_EVALUABLE_INCOMPLETE_CAPACITY_SOURCE_ANTECEDENT. The generation-register packet supplies exact all-rung capacity arithmetic and finite source-contract checks on rungs one through six. Universal all-rung membership in the complete A1--A3 source contract and the executable-to-Lean membership bridge are open. No universe-level physical N is emitted.

The independent finite (A_5) control is summarized in A5_FINITE_CONTROL_STATUS_2026-07-20.md. It proves (M_0=60) on a complete bounded software packet and (D_{\rm raw}=60k), but its publicly inert multiplicity makes raw dimension implementation-dependent. It is a no-go control for physical promotion, not a second physical packet or a cosmic selector.

Canonical lane

• CHECKPOINT_CHANNEL_AUDIT.md records the exact channel contract, decoder-search correctness argument, independent replay and exhaustive controls, and downstream receipt comparison for the public-checkpoint numerical repair under issue #1033.
• F_READBACK_SPEC.md is the Pro5 acceptance contract: complete terminal fiber, atom readouts, endogenous reachability, frozen publicness, global joint kernels, compound confusability graph, exact and approximate correctable capacity, carrier saturation, scalarization, refinement, finite-size slack, receipts, and controls.
• correctable_public_record_capacity.py evaluates finite public checkpoint packets. It computes global sections, exact maximum independent sets, receipt-scale worst-input approximate capacities, support semigroups, carrier bounds, terminal-fiber scalarization, no-new-confusability, greatest fixed points, and unique slack-zero certificates.
• public_record_csp.py is the exact constraint-propagating global-section backend. It is extensionally equivalent to Cartesian enumeration, but it model-counts the connected twelve-observer, twenty-four-atom source packet without exploring 24^12 assignments.
• PUBLIC_RECORD_GLUING_AUDIT.md records the exact finite gluing repair, injective record identifiers, independent completeness replay, and unchanged downstream source receipts under #1033.
• test_correctable_public_record_capacity.py covers saturation, cyclic permutation, joint-coupling nonidentifiability, approximate capacity, ambiguous fibers, order countermodels, target taint, and carrier failures.
• reversible_public_checkpoint_packet.py retains the finite twelve-port icosahedral reference control. It verifies every checkpoint generator is a permutation, certifies M_0(q)=|X_reach(q)|, and emits exact rank-one saturation. It remains explicitly nonphysical.
• test_reversible_public_checkpoint_packet.py checks the 12 vertices, 30 interfaces, exact reversible capacity identity, noninjective failure, and target-taint failure.
• source_derived_public_checkpoint_packet.py defines the first source-derived fixed-cutoff physical packet for issue #548. It freezes the carrier to the twelve edge-center ports with reversible write/check orientation, so D=|P_12 x {write,check}|=24. The producer emits:
  • a complete 67-world structural one-fault trial manifest with one terminal world, fully materialized candidates, SHA-256 completeness receipts, and an output-blind membership predicate;
  • total observer/interface atom readouts and 24 endogenous reachability histories;
  • a frozen universal publicness policy;
  • the complete 40-element D5 x C2 x C2 joint checkpoint family, all 1,600 support-relation compositions, and independently checked local marginals;
  • an empty compound graph, a 24-record independent set, inverse decoders, worst-input and total-variation receipts;
  • 24 orthogonal rank-one carrier projections, proving M_epsilon <= 24 and exact saturation M_0=24;
  • empty/incomplete/ambiguous/singleton fiber controls; isomorphism, cyclic, alternative-coupling, tiny-noise, circular-definition, taint, identity, erasure, and finite-suffix controls; and separate extension/refinement no-new-confusability injections with negative controls.
• test_source_derived_public_checkpoint_packet.py checks the full issue #548 acceptance surface.
• test_source_checkpoint_source_binding.py checks the supplied operations against an independent port-coordinate oracle. The source certificate binds every named continuation to its action on the actual public sections, replays every declared composition on those sections, and compares complete local marginals exactly. Capacity, invertibility and an abstract group table alone do not establish that binding: reversing the named rotations preserves all three while changing the source operation. Reordered maps and explicit zero probabilities remain valid.
• test_source_checkpoint_evidence.py, test_source_checkpoint_provenance.py and test_source_checkpoint_generators.py challenge the actual diagram, carrier projections, complete terminal-trial census, endogenous event histories and declared generating set. Rehashed false evidence is refused through the public certificate and its serialized direct-N consumer. Equivalent presentations and alternative valid executions remain admissible; a content hash alone is not a source proof.
• ISSUE_548_SOLUTION.md maps every acceptance item to the executable receipt.
• capacity_indexed_source_family.py generates four target-clean continuation completions at every positive rung. Reversible identity, copy collapse, a two-class cap, and hidden spectator multiplicity have different exact slack-zero sets while sharing the declared bounded antecedent.
• ISSUE_551_RESULT.md states the bounded counterfamily theorem and its boundary. The all-rung fixed-set disagreement is also proved in Lean/ObserverPatchHolography/CapacityNonidentifiability.lean.
• direct_n_closure_verdict.py consumes that result in the direct global proposal and records that no numeric cosmic value or cosmological comparison is permitted.
• public_record_capacity.py and its tests retain the superseded Pro4 checkpoint-fixed projection branch as a control. A cyclic permutation proves that it is not the canonical capacity definition.

The D=24 artifact is a source-derived fixed-cutoff packet in the declared simulator category. The all-rung counterfamily proves nonidentifiability for the base-agreement, positivity, and carrier-bound completion class. The generation-register audit transports terminal-fiber, A2, A3, sewing, extension, and refinement controls across six finite rungs. Its exact capacity formulas extend to every positive rung, while membership of the executable family in the complete source contract has no all-rung theorem. The bounded receipt lives in complete_packet_capacity_lift.py with the no-producer-import replay in verify_complete_packet_lift_independent.py, the consuming issue #505 verdict is NOT_EVALUABLE_INCOMPLETE_CAPACITY_SOURCE_ANTECEDENT, and the issue #589 horizon exit NOT_EVALUABLE_NO_HORIZON_RECORD_ATTACHMENT is recorded by horizon_record_attachment_verdict.py. The screen value 24 is not a cosmic result.

Generate the issue #548 receipts

python3 source_derived_public_checkpoint_packet.py --output-dir runtime

This writes the complete terminal-fiber manifest, public checkpoint packet, and certificate as canonical JSON.

Independent geometric estimator

• operational_readback_contract.py evaluates frozen scale-discrimination errors, preserves pre/post-checkpoint accounting, requires complete-fiber agreement for rho_op, and compares log M_0 with pi/rho_op^2 only after direct capacity exists.
• test_operational_readback_contract.py checks discrimination endpoints, all-coarser thresholding, capped error accounting, complete-fiber agreement, and independence controls.

The diagnostic residual is

R_rho = log M_0 - pi/rho_op^2.

Defining rho_op from M_0, or using capacity to select its protocol, invalidates the comparison.

Downstream bridges

These bridges consume a unique-zero direct closure. The bounded generation-register packet does not supply one, so they remain contracts with no evaluable capacity-side carrier:

• identifying the correctable record carrier with the de Sitter horizon may identify log D_star with A/(4 ell_star^2) and yield Lambda ell_star^2=3*pi/N_star;
• a positive refinement-natural carrier map may identify the source-normalized screen load with the four-copy weak load and test N_bridge=pi*exp(6*pi/(P*alpha_U(P))).

Neither bridge constructs the direct map. The exterior package proves the weak multiplicity four, but that integer alone does not identify a physical load carrier.

Archived lanes

The dated construction notes and F_candidate_*.py files preserve historical count, affine, and Banach candidates. They have diagnostic value only. The CP* and G2_GAP_1 notes likewise do not supply the exact finite-size selector.

Required completion work

• prove every transported source-contract control for every positive rung;
• bind the executable generation-register packet and capacity evaluator to the Lean completion used by the all-rung arithmetic theorem;
• independently replay that universal membership theorem;
• if the complete source class retains the identity completion, record the resulting complete-class nonidentifiability theorem; otherwise construct a target-clean source selector and prove its admissibility;
• prove an exact finite-size slack law with one regulator-stable physical zero for any proposed selector;
• independently certify the horizon-record, EW/Higgs load-carrier, and operational-resolution bridges;
• supply public hardware-realization evidence if a carrier implementation is claimed.

Usage

python3 -m pytest test_correctable_public_record_capacity.py -q
python3 -m pytest test_public_record_csp.py -q
python3 -m pytest test_reversible_public_checkpoint_packet.py -q
python3 -m pytest test_source_derived_public_checkpoint_packet.py -q
python3 -m pytest test_capacity_indexed_source_family.py -q
python3 -m pytest test_complete_packet_capacity_lift.py -q
python3 -m pytest test_direct_n_closure_verdict.py -q
python3 -m pytest test_horizon_record_attachment_verdict.py -q
python3 -m pytest test_operational_readback_contract.py -q
python3 -m pytest test_public_record_capacity.py -q