All-Possibility Hierarchical Filtering Simulation (APHFS): A Resource-Bounded Framework from Executable Rules toward Multiscale Biology, Aging, and Rejuvenation Research

Preprint; not a peer-reviewed journal article. This record archives the author's APHFS manuscript (version 1.0, dated 3 September 2026) and supporting reproducibility materials. Scientific models begin as imagined explanations; mathematics makes them executable, whereas evidence determines which remain tenable. All-Possibility Hierarchical Filtering Simulation (APHFS) joins bottom-up candidate generation to top-down reality constraints. "All" is local: within a declared finite grammar, resource envelope, and precision policy, every address receives a terminal account. The framework preserves observational ambiguity, rejects an inadequate class only under complete simultaneous bounds, and otherwise permits an indeterminate result. It withholds coarse-graining until specified upper-level observables and decisions remain stable under refinement. Future fault-tolerant quantum computers could enlarge executable dynamics, but would be an execution substrate rather than evidence. We evaluated this machinery once in the complete 256-rule elementary-cellular-automaton grammar. The generating rule and operational signature class were retained in all 64 A1 blocks, although five blocks retained two rules and three retained two classes. A deliberately distinguishable radius-two construction returned model-class inadequacy in all 64 B1 blocks. D2-CERT used the provably density-preserving periodic Rule 170 shift: 0/64 calibration and 0/64 held-out violations checked certificate-pipeline conformance for a constructed closed map, not uncertain coarse-law accuracy. Four resource policies reached identical decisions. One prespecified salted pseudo-random early-stopping order was least costly on this constructed workload (75,080 units), versus 89,816 for the other order, 92,592 for adaptive fidelity, and 132,096 for exhaustive search. These results establish bounded, failure-capable behavior only for this synthetic finite grammar, not physical or biological validation. A separately prespecified enzyme-kinetics study is the next empirical bridge toward multiscale biology; aging and safety-constrained rejuvenation remain longer-term destinations. Availability and limitations: The public repository and accompanying archive provide scientific source code, configurations, schemas, aggregate/source data, existing figures, and read-only reconstruction tools. Raw protected role values and the protected result container are not included; this is not an original-run replay or independent replication. Standalone rendering-font dependencies are external and documented in the archive. Licensing: CC BY 4.0 applies to the manuscript, documentation, figures and released data; MIT applies to software. Existing third-party notices remain applicable. The licenses apply by component, not as alternative licenses for the entire deposit. Authorship and AI assistance: Mingyuan Chen originated the central APHFS research direction. Substantive AI-assisted methodological, software and editorial work, and the author's review scope, are disclosed in section 12. AI outputs are not evidence or independent validation. Related repository: https://github.com/cmy515006420/APHFS