[Date Prev] [Date Next] [Thread Prev] [Thread Next] Date Index Thread Index Search archive:
Date:Mon Jan 1 08:59:59 2004 
Subject:Re: Comparing Garbage Collectors 
From:Jonathan L Cunningham 
Volume-ID:1040101.01 

On Thu, 1 Jan 2004 00:15:20 +0000 (UTC), A.Sloman@cs.bham.ac.uk wrote:

>However, if deciding how an arbitrary program should be optimised is as
>difficult as deciding whether an arbitrary procedure will ever halt,

It is.

>then we have problems.

Happy New Year ;-).

But it depends on what you mean by "optimised" -- would optimising
the depth first search in TEACH TOWER into a polynomial time
algorithm, if it is possible[1], count as an "optimisation"?

How about optimising a bubble sort into, say, quicksort[2] or
one of the better sorts?

[1] It's attempting to solve the knapsack problem, which is one
of the familiar NP-complete problems, so we don't *know* if it
is possible, although it is strongly believed that it isn't.

[2] Which has bad "worst case" behaviour, so your optimiser
would have to know something about the likely distribution of
data to be sorted: it could, in some case, be *worse* than
bubblesort, IIRC. Therefore, your optimiser needs to be
psychic, and psychic computers have yet to be developed ...
so it's actually a *harder* problem than the halting problem.

Jonathan

-- 
    Use jlc at address, not spam.