rmk wrote: > > In article <33785E6F.2B8B@bnr.co.uk>, Ian Woollard <email@example.com> wrote: > <Bzzzt. Wrong. > < > <Heuristics aren't algorithms. The definition of an algorithm is > <something that is guaranteed to give you the right answer. > < > > Ahem, I recall nothing in the design of an algorithm that guarantees a > *right* answer.
Some algorithms have been proven to be correct. I once read an interesting book called 'Provably Correct Programs'. I remember that they suggested that you write programs in a way that makes it easy to prove them correct, rather than write a program and then try to prove that it is correct.
You need to specify pre and post conditions, and the sequences of states of the variables mathematically. Can be done. (Sometimes.)
Of course you can't ever know whether the implementation has been done correctly. Still this IS sci.maths right? Practical matters are not important ;-)
> I can, in fact, imagine several algorithms, mostly those > designed by former students but some of my own, that almost never give the > right answer.
Then by definition, they weren't algorithms, they were heuristics.
Wow! Two pedant points in two postings. Next! ;-)
> ...rmk > > -- > Hamlet is the tragedy of tackling a family problem > too soon after college. > - Tom Masson (1866-1934)
-- -Ian |-A day for firm decisions|-Put off procrastination| |Or is it? |as long as possible. | -Predestination was |-Of course I'm sane. |-Remove freedom of | doomed from the start |The voices tell me so. |speech for censors, now!|