mathlesstraveled.com
More on Fermat witnesses and liars
In my previous post I stated, without proof, the following theorem: Theorem: if $latex n$ is composite and there exists at least one Fermat witness for $latex n$, then at least half of the numbers …