Complete formula

The capacity inequality is not a claim that an arbitrarily small fixed laboratory can support this example. If it fails, the example is not admitted on that instance.

With per-operation complete-channel allowances, a valid comparison budget is, for example,

Theorem thm:error bounds the complete output difference by its sum capped at one. The actual failure probability then lies within that allowance of $1/4$, provided the failure flag remains part of the compared output.