Finding algorithms

[Christopher Wong]

What would make a good resource for algorithms? There are ample references
APIs, languages, structure and numerical work. But some applications need a
little more thinking. For example: a program that sets up matches for a
competition, maximizing matchups, randomness and competitiveness. That's a
combinatorics and optimization problem. Where would you go to when you want
to solve a practical problem rather than spend the next few weeks working
through theory?

Chris

[Greg P.]

"Christopher Wong" <cwong@world.std.com> wrote in message
news:bga4hs$ijc$1@bob.news.rcn.net...

> What would make a good resource for algorithms? There are ample references
> APIs, languages, structure and numerical work. But some applications need

a

> little more thinking. For example: a program that sets up matches for a
> competition, maximizing matchups, randomness and competitiveness. That's a
> combinatorics and optimization problem. Where would you go to when you

want

> to solve a practical problem rather than spend the next few weeks working
> through theory?

A whiteboard, black marker, cold coffee, and a cigarette =)

[Kevin Easton]

In comp.os.linux.development.apps Christopher Wong <cwong@world.std.com> wrote:

> What would make a good resource for algorithms? There are ample references
> APIs, languages, structure and numerical work. But some applications need a
> little more thinking. For example: a program that sets up matches for a
> competition, maximizing matchups, randomness and competitiveness. That's a
> combinatorics and optimization problem. Where would you go to when you want
> to solve a practical problem rather than spend the next few weeks working
> through theory?

Knuth's TAOCP?

- Kevin.

[Greg P.]

"Kevin Easton" <kevin@-nospam-pcug.org.au> wrote in message
news:newscache$u0mvih$p5p$1@tomato.pcug.org.au...

> In comp.os.linux.development.apps Christopher Wong <cwong@world.std.com>

wrote:

> > What would make a good resource for algorithms? There are ample

references

> > APIs, languages, structure and numerical work. But some applications

need a

> > little more thinking. For example: a program that sets up matches for a
> > competition, maximizing matchups, randomness and competitiveness. That's

a

> > combinatorics and optimization problem. Where would you go to when you

want

> > to solve a practical problem rather than spend the next few weeks

working

> > through theory?

>
> Knuth's TAOCP?

Eeek, don't you think Knuth is a bit much for this? I guess I agree with you
though.

Have you read the Algorithms in <insert language here> by R. Sedgewick? It
pretty much covers everything from simple connectivity-problems to graphs:
[url]http://www.amazon.com/exec/obidos/tg/detail/-/020172684X/qid=1059633925/sr=1-2/ref=sr_1_2/103-5750351-7153416?v=glance&s=books[/url]

However, I own his C++ versions of Parts1-4 and part 5...and they aren't
ANSI/ISO standardized (at least my printing). =/ Good books though to peruse
while sitting on the john. Each theory has a code example, so less of the
ideals, more of the how-to.

[Michael B Allen]

On Thu, 31 Jul 2003 00:05:09 -0400, Christopher Wong wrote:

> What would make a good resource for algorithms? There are ample
> references APIs, languages, structure and numerical work. But some
> applications need a little more thinking. For example: a program that
> sets up matches for a competition, maximizing matchups, randomness and
> competitiveness. That's a combinatorics and optimization problem. Where
> would you go to when you want to solve a practical problem rather than
> spend the next few weeks working through theory?

Well this is good start:

[url]http://www.nist.gov/dads/[/url]

but I have to admit I have never really used anything from this site
directly. The Coreman book on the otherhand is something I have used
on several occations. Although I don't think either of these resources
will be sufficient to cover advanced topics like you describe. I think
you're destined for a "few weeks working through theory". Even if you
found something that hits the nail on the head the topic is sophisticated
enough that you would probably need to understand the theory to use it
anyway.

Mike

[e1p1s]

Christopher Wong wrote:

> What would make a good resource for algorithms? There are ample references
> APIs, languages, structure and numerical work. But some applications need a
> little more thinking. For example: a program that sets up matches for a
> competition, maximizing matchups, randomness and competitiveness. That's a
> combinatorics and optimization problem. Where would you go to when you want
> to solve a practical problem rather than spend the next few weeks working
> through theory?
>
> Chris
>

Not always but sometimes useful: [url]http://www.nist.gov/dads/[/url]

[rossb]

Christopher Wong <cwong@world.std.com> wrote in message news:<bga4hs$ijc$1@bob.news.rcn.net>...

> What would make a good resource for algorithms? There are ample references
> APIs, languages, structure and numerical work. But some applications need a
> little more thinking. For example: a program that sets up matches for a
> competition, maximizing matchups, randomness and competitiveness. That's a
> combinatorics and optimization problem. Where would you go to when you want
> to solve a practical problem rather than spend the next few weeks working
> through theory?
>
> Chris

I often find that "Numerical Recipes" by Press, Flannery, Teukolsky,
and Vetterling has some good answers, and includes an explanation of
why other ways of doing it might not be so good. Its saved me a few
times when things got too "statistical".

Brenton

[Nils Petter Vaskinn]

On Thu, 31 Jul 2003 04:08:48 +0000, Greg P. wrote:

> "Christopher Wong" <cwong@world.std.com> wrote in message
> news:bga4hs$ijc$1@bob.news.rcn.net...

>> What would make a good resource for algorithms? There are ample
>> references APIs, languages, structure and numerical work. But some
>> applications need

> a

>> little more thinking. For example: a program that sets up matches for a
>> competition, maximizing matchups, randomness and competitiveness.
>> That's a combinatorics and optimization problem. Where would you go to
>> when you want to solve a practical problem rather than spend the next
>> few weeks working through theory?

Do you need perfect or good?
First shot at an algorithm:
1. assign randomly
2. Do a few checks to identify if the alternative sucks
3. If it does suck goto 1
4. done

Then depending on how badly it performs improve the placing in step 1
until performance is acceptable.

The thing is it depends on how much the program is used, if it's used for
one specific competition once a year it's ok if it takes 10 minutes to
find a good setup, compared to an unknown amount of programmer time. If
it's to be used a lot it may be time to get out those combinatorics books.

> A whiteboard, black marker, cold coffee, and a cigarette =)

Cold coffee? ouch! I'd have to be pretty desperate to resort to that.

Whiteboard, several colors markers, paper, pen, hot coffee (but not hot
enough to burn your tounge), cigarettes.

regards
NPV

[Sam Zoghaib]

Greg P. wrote in article <WX2Wa.399$jg7.121@newsread3.news.pas.earthlink.ne t>
on Thursday 31 July 2003 08:47 in comp.unix.programmer:

>> Knuth's TAOCP?

> Eeek, don't you think Knuth is a bit much for this? I guess I agree with you
> though.

Well, Knuth's TAOCP has all the theory, that's true, but it's well written ,
so if you just ned to give a shot at an algorithm, you can just pick one.
Algorithm are first described in English, then coded in MIX, and assembly
language for a computer of Knuth's design.
And when you need some more information about the algorithm, well, you just
have to read the section more thoroughly.

Sam
--
"Don't be afraid, I'm gonna give you the choice I never had..."

- Lestat in "Interview with the Vampire" (Ann Rice, 1976)

[Sam Zoghaib]

rossb wrote in article <66728128.0307310037.15bf1e18@posting.google.com > on
Thursday 31 July 2003 10:37 in comp.unix.programmer:

> I often find that "Numerical Recipes" by Press, Flannery, Teukolsky,
> and Vetterling has some good answers, and includes an explanation of
> why other ways of doing it might not be so good.

I don't have the book, so I can't give my personal opinion, but there's a
thread on comp.lang.c about it, and this link was given; you may want to read
that first:

[url]http://www.lysator.liu.se/c/num-recipes-in-c.html[/url]

Sam
--
"Creativity can be a social contribution, but only in so far as society is
free to use the results."

- Richard Stallman

[Bjorn Reese]

Christopher Wong wrote:

> competition, maximizing matchups, randomness and competitiveness. That's a
> combinatorics and optimization problem. Where would you go to when you want
> to solve a practical problem rather than spend the next few weeks working
> through theory?

One possibility is to ask on the sci.op-research newsgroup.

[Lee Sau Dan]

>>>>> "Nils" == Nils Petter Vaskinn <no@spam.for.me.invalid> writes:

Nils> The thing is it depends on how much the program is used, if
Nils> it's used for one specific competition once a year it's ok
Nils> if it takes 10 minutes to find a good setup, compared to an
Nils> unknown amount of programmer time. If it's to be used a lot
Nils> it may be time to get out those combinatorics books.

There are applications with real-time requirements, even if it is to
be used only once per year. Having to wait for 10 minutes for an
answer may not be acceptable.


--
Lee Sau Dan §õ¦u´°(Big5) ~{@nJX6X~}(HZ)

E-mail: [email]danlee@informatik.uni-freiburg.de[/email]
Home page: [url]http://www.informatik.uni-freiburg.de/~danlee[/url]