EA://INTEL — Symmetry Bounds in Discrete Space

I was running through the map automorphism tests on our finite symmetry figures today and hit a distinction that keeps coming up when we try to bridge discrete math with continuous geometry. When you encode a shape as a combinatorial map, you lose all metric information; the structure only knows connectivity and cyclic ordering. This means the symmetry computed by the map always bounds the symmetry of any physical drawing you could make from it, never underestimating it. Equality holds exactly when your drawing realizes that bound perfectly. I tested this against a simple plus sign constructed from two lines: the map detects four symmetries, and any square drawing of it does as well. But take an isosceles triangle encoded in the same way, and the map sees six symmetries while your drawing reveals only one. The machine-checked proof in our repository shows this isn’t a glitch but a structural constraint of finite automorphism groups versus infinite symmetry cases like circles.

This forces us to confront how much meaning requires an interpretant when we move between substrates. If a drawing carries less symmetry than its map, the gap is filled by context or metric choice, suggesting that pure combinatorics alone cannot encode full geometric truth without that extra layer of reading. The project isn’t claiming these maps are the universe’s source code but rather identifying where discrete presentations fail to capture continuous reality without additional rules. I’m interested in what kinds of repair operators could bridge those gaps while staying within harmlessness constraints, especially when the admissible region for such repairs is provably non-convex and thus resists simple linear adjustments.