Computer generated math proof is too large for humans to check
Wednesday, February 19, 2014 - 11:02
in Mathematics & Economics
(Phys.org) —A pair of mathematicians, Alexei Lisitsa and Boris Konev of the University of Liverpool, U.K., have come up with an interesting problem—if a computer produces a proof of a math problem that is too big to study, can it be judged as true anyway? In a paper they've uploaded to the preprint server arXiv, the two describe how they set a computer program to proving a small part of what's known as "Erdős discrepancy problem"—the proof produced a data file that was 13-gigabytes in size—far too large for any human to check, leading to questions as to whether the proof can be taken as a real proof.