The New Sudoku Players' Forum

[mistermac]

There is a theory of information reliabilty which a man called Kelly was famous for.

On a "noisy" wire the message (digital in this case) becomes suspect. He devised a method based on probability of sussing out the correct message.

In other words, the "puzzle" caused by noise could be solved by taking the "known", eventhat was uncertain, and deducing the whole message.

A Sudoku is a little like that, giving incomplete information, and enabling a solution, or at least a partial solution, or the possible solution.

Is there any theory on thei topic known to members of this forum, as the topic of incomplete information is of great interest to me.

Meanwhile, I will amuse myself by trying ti solve the "hard" puzzles here.

No slice and dice for you guys I am sure. The clues I feel sure are minimal.

Thanks for allowing me in, and please reply if you can help with my problem above.

[Bigtone53]

Mistermac,

Welcome to the forum.

The Kelly Criterion (derived from the noisy-wire work) is well-known to gamblers as a theoretical technique for calculating the optimum fraction of one's current bank to invest given the level of uncertainties and possible returns. You can read about it here . The difficulty in practice is that the true odds of gambles, especially horseracing ones, can only be approximated and are not known precisely.

I am not aware of any work on its relevance to sudoku, but at least on this forum, sudoku-solving is typically a logical process, rather than maximising probablities or (worse) pure trial and error.