Re: puzzle

In article <1118820818.822263.140030@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
spinoza1111@xxxxxxxxx says...
> Gerry Quinn wrote:

> > K is dependent on the precise details of how the code is arranged, how
> > it suits the machine hardware, etc.
> >
> > That's precisely why it is discounted in big-O analysis.
>
> This is self-contradictory. Precisely because K is small relative to O,
> you can change it by changing code details.

Small relative to O? K represents a positive real number. Exactly how
do you compare that with the name of a (quasi-) function?

> K isn't discounted in big O analysis. In fact, if I were on the rag and
> in a mood for negative hermeneutics, I could say that ooo, Quinn
> doesn't know jack if he thinks K is discounted.

> What K is is a constant below which execution time doesn't sink for
> n>=1. For n<=10e1, k matters but then becomes unimportant for n>10e1.

Big-O analysis is predicated precisely on the notion that as n goes to
infinity, constant factors become asymptotically unimportant and can be
ignored.

> The psychological implication you've taken from big-O analysis, which
> had its origins in productive labor and as such is not some Platonic
> ideal accessible only to adepts, is that when one does studly big-O,
> one gets to wear some sort of cool robe or toga, and discount the
> concerns of the *canaille*.

Why would anyone think that? Big-O is a tool for analysing qualities
of algorithms. Indeed, it is a tool that is more useful in simple
informal cases accessible to all (or most!), than in any complex
mathematical theory, in which it has to be hemmed around with
restrictions that can be ignored in simple cases where 'infinity' is
reduced in some aspects to the bounds of particular hardware. (E.g.
where we do not bother with the fact that array pointers must become
larger as the number of array elements increases beyond a 'standard'
value.)

[Rambling nonsense deleted.]

> In a similar fashion, the compsci student is lured by unprincipled
> head-hunters from a shortlist of "good" (eg., white) schools into jobs
> in which he learns to work in a crappy fashion with crap while talking
> a good game about "good" algorithms such as de white peoples use, and
> in effect short-circuiting any conversation about what's actually going
> on with a personal attack.

Racism and personal attacks seem to be entirely your department in this
discussion so far.

- Gerry Quinn







.