Computational Cost of dsearchn

What is the computational cost of the dsearchn function?

Answer 1

[Sean de Wolski]

Are you looking for number of flops?

I don't think you're going to have much luck finding this. dsearchn relies on mex and qhull to do most of the work.

You could use tic/toc to time it, if that would also be sufficient.

[Arielle]

Tic/Toc helps, but the systems I am working with vary in size considerably.

In general, is dsearchn considered cheap/expensive? Or is that, again, a hard question to answer?

(Do you happen to know anything about the cost of mex and hull?)

[Arielle]

qhull*

[Sean de Wolski]

That's really an existential question and it really depends on your application. If dsearchn takes a few minutes for one person that might be extremely expensive where a few minutes for another person would be light speed.

Mex and qhull are used because they're fast!

Why do you need to know this computational complexity? Is there something else that's time-dependant and relies on it being complete?

I'm sorry I don't know much else about dsearchn. You could also use the profiler to identify other bottle necks in your code to see if it is the biggest slowdown.

[Arielle]

To make a long story short, my code is part of a greater project and needs to work as fast as possible. I'm trying to identify any areas of weakness, and I didn't know much about the cost of dsearchn. Thanks for your help!