The first claim is that given a non-trivial real interval [a,b], then I_[a,b] evaluated over the set of all polynomials with Galois group S_p is unbounded.
The core idea behind the proof is a bit trivial: with something non-zero, one can multiply it to boost it as big as one desires (cue Archimedes).
But where does the bootstrap come from?
Answer: the Osofsky polynomial cited as [1]. But the author forgets that the interval [a,b] is the adversary's choice, not theirs. So said adversary promptly proves there's a non-trivial [a,b] that zeroizes Osofsky.
Thank you for reading. To your point, I probably should have added that if the Osofsky polynomial integrates to zero on the chosen interval, we can translate it by a suitable rational number and the integral will be non-zero and will still have the same Galois group. I thought that was clear but maybe not. Either way, the claim is valid.
Some research related to questions I had which I find interesting.
The first claim is that given a non-trivial real interval [a,b], then I_[a,b] evaluated over the set of all polynomials with Galois group S_p is unbounded.
The core idea behind the proof is a bit trivial: with something non-zero, one can multiply it to boost it as big as one desires (cue Archimedes).
But where does the bootstrap come from?
Answer: the Osofsky polynomial cited as [1]. But the author forgets that the interval [a,b] is the adversary's choice, not theirs. So said adversary promptly proves there's a non-trivial [a,b] that zeroizes Osofsky.
Poof.
Thank you for reading. To your point, I probably should have added that if the Osofsky polynomial integrates to zero on the chosen interval, we can translate it by a suitable rational number and the integral will be non-zero and will still have the same Galois group. I thought that was clear but maybe not. Either way, the claim is valid.
> we can translate it by a suitable rational number and the integral will be non-zero
You're correct. The main theorem is true, but your write-up has defects.