On Tue, Jul 28, 2015 at 3:20 PM, Henry Baker <hbaker1@pipeline.com> wrote:
"Blinding", thanks!
Is there an algebraic theory of blinding that is sufficiently general to cover both crypto blinding and things like "pre-conditioning" in numerical matrix theory?
Blinding and preconditioning serve different purposes, so it would have to be a pretty general idea. The preconditioning sounds very similar to things like randomizing a list before quicksorting it or randomizing a hashtable's hash algorithm to prevent worst-case behaviors. http://web.archive.org/web/20070202204633/www.cs.rice.edu/~scrosby/hash/Cros... Random inputs are also useful for certain kinds of proof: http://www.scholarpedia.org/article/Applications_of_algorithmic_information_... Theorem [Chaitin-Schwartz, 1978] For almost all inputs, the probabilistic primality test is error-free in case it uses a long enough algorithmically random inputs. Theorem [Calude-Zimand, 1984] For almost all inputs, every “decent” Monte Carlo simulation algorithm is error-free in case it uses a long enough algorithmically random inputs. -- Mike Stay - metaweta@gmail.com http://www.cs.auckland.ac.nz/~mike http://reperiendi.wordpress.com