TFPT scientific source excerpt Original repository path: experiments/theory-contracts/foundational-process-classification-20261016/OBSERVABILITY.txt Original file SHA-256: 5253c7df5f3347129e2eb5d8e6bb161d09d9fa19ae64a456ca967eebc7439468 Inclusive original lines: 236-315 Line numbers below refer to the original source. This bounded excerpt is source evidence; it is not a new proof/checker run. Conversation metadata and personal execution details are omitted where indicated. 000236 | 8. Finite sequential data admit a classical history realization 000237 | 000238 | THEOREM (standard constructive representation, no novelty claim). Given 000239 | finite preparation, action and report sets, horizon H, and nonnegative 000240 | normalized sequential probability tables p(a_1...a_n | q_1...q_n), n<=H, 000241 | suppose summing over the latest report recovers the prefix table and earlier 000242 | marginals do not depend on later choices. Then a finite classical stochastic 000243 | history machine realizes every table, including adaptive protocols. 000244 | 000245 | Proof: a machine state is the preparation label and full history 000246 | h=(q_1,a_1,...,q_k,a_k). When q is chosen, output a with conditional weight 000247 | p(h,q,a)/p(h), then append (q,a) to h. For unreachable zero-probability h, 000248 | choose any normalized conditional. At horizon H halt. Induction telescopes 000249 | the product of conditionals to the given table. There are finitely many 000250 | histories, bounded by |preparations| sum_(k=0)^H (|actions||reports|)^k. 000251 | The model may depend on the entire history and preparation context. For 000252 | support data use permitted branches in place of positive probabilities, 000253 | assuming analogous prefix extension consistency. 000254 | 000255 | This gives more than the initial definitions: arbitrary finite tables in 000256 | this class have an explicit classical representation. It explains an 000257 | identification LIMIT, not the origin of the physical carrier. It is not a 000258 | claim of an efficient simulator, bounded stationary memory, reversible 000259 | dynamics, preparation noncontextuality or spacelike locality. The prefix 000260 | condition supplies a direction to the experiment; it does not derive 000261 | physical causality or forbid feedback with memory. Constraints on resources, 000262 | independent preparations, causal access or operational equivalences can 000263 | make classical competitors testable; do not drop those constraints when 000264 | interpreting such a test. Nothing here overturns Bell or contextuality tests. 000265 | 000266 | Finite sequential data can distinguish I from D, yet cannot on these 000267 | assumptions exclude ALL classical hidden-memory implementations of D. 000268 | Those are different identification questions. Finite precision also cannot 000269 | prove exact all-context equality of two empirical preparation procedures. 000270 | 000271 | 9. What can select a remaining model, and what is actually necessary? 000272 | 000273 | Discriminators are predictions, not new measurements performed this round: 000274 | - R versus quantitative extensions: reproducible nontrivial weights under 000275 | a specified preparation, conditional on its operational controls. 000276 | - Specified I versus D: X,Z,X match rate1 versus1/2; disturbance can be 000277 | classical too, so this selects between instruments, not ontologies. 000278 | - I versus D correlation access: distinguish the two diagonal mixtures by 000279 | an accessible parity/joint test, or certify their operational equivalence 000280 | relative to a justified family. Mere absence of a found test is no proof 000281 | that no such test is physically possible. 000282 | - Unrestricted classical history models versus stricter alternatives: 000283 | finite sequential disturbance alone does not suffice. A justified memory, 000284 | causal-access or preparation-equivalence restriction is additional 000285 | empirical/operational information. No restriction is adopted just to 000286 | force the desired winner. 000287 | 000288 | All three models share the declared repeatable distinction, nonempty 000289 | composable transcripts, and at least two residual answer classes. Those 000290 | shared properties are largely the benchmark and its representational 000291 | consequences, not a nontrivial new physical necessity. Neither probability 000292 | weights, global information conservation, noncommuting controls, simplex 000293 | structure, compatible joint readout nor a quantum/Hilbert structure is 000294 | forced by that benchmark. The history-realization theorem adds a broad 000295 | conditional underidentification statement, not a chosen fundamental model. 000296 | 000297 | No selection of a physical world has been achieved. The next useful input 000298 | is a concrete reproducible phenomenon WITH justified operational/resource 000299 | constraints that separates candidates. Do not continue by designing a 000300 | protected memory or choosing a new axiom merely to remove the multiplicity. 000301 | Keep the multiplicity and its discriminators as the current research result. 000302 | 000303 | Primary comparison sources (not premises imported as physical necessities): 000304 | Barrett, Information processing in generalized probabilistic theories, 000305 | https://arxiv.org/html/quant-ph/0508211, sectionsII.1 andIV.3: probability, 000306 | convex closure and operational state specification are explicit assumptions; 000307 | the simplest two-binary-question normalized response space is a square. 000308 | Spekkens, Contextuality for preparations, transformations and unsharp 000309 | measurements, https://arxiv.org/html/quant-ph/0406166, sectionsI--II: 000310 | operational equivalence and ontological equivalence must be distinguished; 000311 | equating them is an additional noncontextuality assumption. 000312 | Hardy, Quantum Theory From Five Reasonable Axioms, 000313 | https://arxiv.org/abs/quant-ph/0101012: quantum reconstruction uses a package 000314 | of additional axioms. No reconstruction theorem is invoked here to select 000315 | a quantum theory from observation alone.