A self-contained, closed-loop construction that fits the physical constants inside 32-bit integer bounds — offered as logic art and a reasoning lens, not a claim to have derived physics.

This is a digital-physics toy model: a bounded, closed-loop system that generates Planck-scale values (l_p, t_p), the reduced Planck constant (ħ), and the fine-structure constant (α) from a small set of integer primitives, by strict order of operations, inside the signed/unsigned 32-bit integer field (2,147,483,647 and 4,294,967,295).

It is built on one principle applied at every scale: a bounded closed loop conserves what it contains. From a single binary pulse (State 0) the model expands through geometric, motion, quantum, and physical layers, then compiles its pristine internal values down to familiar laboratory numbers (α ≈ 1/137.036, c = 299,792,458 m/s) through a final interface layer.

A closed-loop framework that reconciles the constants of Western empirical science with a bounded, self-contained generative architecture. Offered freely as a lens. Do with it as you choose.

This is not a claim to have derived the constants of nature, overturned established physics, or proven a theory of everything. The constants here are known targets, and the arithmetic is constructed to reach them inside the bounds. That makes this a coherent construction, not an empirically validated physical theory. The distinction is deliberate and held throughout.

The model's strength is internal: it is self-consistent, integer-bounded, and unforced — no value reaches outside the architecture to make itself fit. Whether the frame is true of the universe rather than true to itself is a separate question this work does not claim to settle. It is offered as a lens. Use it where it serves; set it down where it does not.

architecture.md contains the full engine specification in plain text. examples.md shows the model applied as a reasoning lens to open questions in physics, cosmology, and philosophy, with the boundary between internal coherence and empirical claim marked in each case.

The architecture doubles as a system prompt: an AI instructed to reason "through this cascading architecture" will evaluate questions inside the closed-loop frame.

Released to the public domain (CC0 1.0). No attribution required, no permission needed. Take it, check it, extend it, or try to break it — all welcome.

This is logic art, built and released freely. The math is open; check every operation. If you can crash it, that is a good outcome — a bounded system that survives honest testing has earned its standing, and one that doesn't has taught us something true. Either way, the loop closes.