Isaac's Condition (TU) is false
In 2006 Isaac reduced the simple normality to base 2 of square roots of non-squares to a property he called Condition (TU). It states that the upper frequency of the digit 1 in the dyadic expansion of sqrt(ν) depends only on the tail of ν's expansion. Whether Condition (TU) holds determines whether Isaac's claimed proof is valid.
The first proof is elementary. It turns Isaac's own reduction into a bootstrap that builds a sparse dyadic series x for which the pair (x^2, x^2+1/4) would violate (TU). The second proof uses the Mosquera–Shmerkin Fourier-decay theorem and the Davenport–Erdős–LeVeque criterion on a ratio-1/8 Cantor measure. This yields explicit counterexamples for almost every point of the measure. Everything is formalized in Lean 4 with Mathlib in the file BB47/Isaac.lean, with the two cited theorems stated as axioms.
Condition (TU) is false, already for the full sequence n_i = i, so the simple normality of sqrt(2) to base 2 remains open. The first proof is verified using only Lean's standard axioms. The second proof additionally relies on exactly the two cited theorems.
| Statement | Lean declaration |
|---|---|
| Definition 2.1 | TU, TUfull, SameTail, freq |
| Proposition 3.1 | frequently_freq_ge |
| Theorem 1.1, first proof | not_TUfull_bootstrap, not_TU_bootstrap |
| Theorem 4.1 (axiom) | MS.fourierDecay_smoothImage |
| Theorem 1.1, second proof | not_TUfull_fourier |
