Canonical-shaped, versioned experiment proposals for criticism. Every card is explicitly unadmitted: it is not a run, result, evidence record, ranking, or production plan.
illustrative-unadmitted · revision 1 · physical · falsification
Prepare the same typed local pair and boundary repeatedly, blind the operator to remote graph context, apply the declared global drive, and test whether the output topology is invariant to remote context.
Strongest falsifier
Output class changes systematically with remote context while local pair and boundary are held fixed.
Equipment capabilities
- required: chiral-nematic sample cell with controlled electric drive (any-one, quantity 1)
- required: calibrated multiview polarized microscopy (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · physical · discrimination
Encode two finite matrices into complementary optical frequencies, measure the nonlinear mixing output, and compare against an independently computed dense product under randomized phase and amplitude controls.
Strongest falsifier
The same product signature remains when the nonlinear interaction is ablated.
Equipment capabilities
- required: phase-stable complementary-frequency source (any-one, quantity 1)
- required: complex spectral readout with calibrated dynamic range (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · physical · characterization
Prepare known input OAM and transverse modes, apply a declared interaction, and compare input and output modal charge and energy balances with phase-resolved readout.
Strongest falsifier
Residual modal charge changes beyond uncertainty or leakage tracks an unmodeled interaction channel.
Equipment capabilities
- required: phase-resolved OAM and multimode sorter (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · physical · discrimination
Collide two independently prepared OAM/multimode wave packets with controlled relative phase and angle, and discriminate a local angular-interference/contraction rule from linear superposition and detector artifacts.
Strongest falsifier
The signature is fully explained by linear interference, alignment, or detector cross-talk.
Equipment capabilities
- required: phase-resolved multimode field analyzer (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · physical · calibration
Write the same finite optical circuit separately in perovskite, LiNbO3, and chalcogenide variants, inject enumerated basis states, and compare decoded outputs against an exact symbolic oracle with calibrated phase and loss accounting.
Strongest falsifier
A basis-state output cannot be distinguished from a bypass or crosstalk model at the declared precision.
Equipment capabilities
- required: femtosecond direct-write platform with metrology (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · hybrid · falsification
Address separated written paths across a finite 3-D volume, apply a declared optical operator, and test whether state readout remains separable and repeatable across depth and recurrence passes.
Strongest falsifier
Feedback-off or planar controls reproduce the proposed volumetric recurrence signature.
Equipment capabilities
- required: 3-D femtosecond writing and depth metrology (any-one, quantity 1)
- alternative: depth-resolved complex optical readout (any-one, quantity 1)
- alternative: phase-resolved modal volume readout (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · hybrid · characterization
Apply a fixed sequence of optical inputs to a mixed-halide perovskite loop while an explicit controller adapts one declared parameter, then distinguish recurrent state adaptation from photochemical drift.
Strongest falsifier
A drift-only model explains the same sequence and state dependence without adaptive updates.
Equipment capabilities
- required: time-resolved calibrated spectral readout (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · physical · discrimination
Drive a multimode optical medium with independently logged RF and acoustic controls, reconstruct the complex mode-to-mode operator, and compare it with a field-off and single-control model.
Strongest falsifier
An ablated or linear model predicts the joint output within the same tolerance.
Equipment capabilities
- required: independently phase-logged RF and acoustic controller (any-one, quantity 1)
- optional: independent RF phase pickup (at-least-n, quantity 2)
- optional: independent acoustic delay pickup (at-least-n, quantity 2)
Agent-readable JSON
illustrative-unadmitted · revision 1 · simulation · replication
Generate a finite tensor workload, route operands through the declared 3-D coincidence address, and compare FPGA outputs and resource traces with an exact software oracle and a randomized address control.
Strongest falsifier
A declared coincidence address produces a dropped, duplicated, or order-dependent tensor contribution.
Equipment capabilities
- required: FPGA with deterministic cycle/event trace (any-one, quantity 1)
Agent-readable JSON
illustrative-unadmitted · revision 1 · analytical · discrimination
Enumerate finite amplitude expressions in a Rust exact oracle, reduce the corresponding HVM symbolic terms, and compare normalized limits and failure cases without treating a rewrite trace as a physical result.
Strongest falsifier
The symbolic system returns a finite amplitude where the exact oracle proves divergence or undefined normalization.
Equipment capabilities
- required: reproducible Rust exact arithmetic toolchain (any-one, quantity 1)
Agent-readable JSON