The Universal Prior Does Not Select a Factorization通用先验不选出分解

Proves, via a constant-complexity carry-increment recoding witness, that the universal (Solomonoff–Levin) algorithmic prior selects no preferred tensor factorization — and therefore cannot, by itself, supply locality.借一个常数复杂度的进位递增重编码作为可审计见证,证明通用(Solomonoff–Levin)算法先验不选出任何优先的张量分解——因而其自身无法提供局域性。