comp.lang.idl-pvwave archive: archive » Re: The IDL way: Find last non-zero value

Home » Public Forums » archive » Re: The IDL way: Find last non-zero value

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

Re: The IDL way: Find last non-zero value [message #67727 is a reply to message #67726]

Mon, 24 August 2009 19:43

Jeremy Bailin
Messages: 618
Registered: April 2008

Senior Member

On Aug 24, 8:12 pm, Jeremy Bailin <astroco...@gmail.com> wrote:
> On Aug 24, 4:42 pm, Eric Hudson <ehud...@mit.edu> wrote:
>
>> Thanks to all of you for the suggestions. I timed them all on several
>> of the larger matrices and found that the last suggestion (using
>> cumulative totals) was about twice as fast as the previous ones, and
>> about 5 times faster than the brute force loop. It did depend on how
>> many zeros ended each line on average. When the number was small the
>> loop could win, but for my typical matrices the cumulative total is
>> the clear winner.
>
>> Thanks again,
>> Eric
>
> I'm shocked that no one's suggested histogram! Here's my take, which
> I'd wager does at least as well as the cumulative totals:
>
> sz=size(array,/dimen)
> ncol=sz[0]
> nrow=sz[1]
> evenrows = 2*lindgen(nrow)
> h = histogram( (array ne 0) + rebin(reform(evenrows,
> 1,nrow),ncol,nrow), min=0, max=2*nrow-1, $
> reverse_indices=ri)
> result = ri[ri[evenrows+2]-1] mod ncol
> w = where(h[evenrows+1] eq 0, nw)
> if nw gt 0 then result[evenrows[w]]=-1
>
> -Jeremy.

As usual, stupid typo in the last line, which should be:

if nw gt 0 then result[w]=-1

It looks to me like it doesn't beat the cumulative total version after
all, at least on my test case. Oh well! Eric, if you have the timings
on them, I'd be very curious to see how they stack up in detail!

-Jeremy.

Report message to a moderator

[Message index]

		Re: The IDL way: Find last non-zero value By: Jeremy Bailin on Mon, 24 August 2009 19:53
		Re: The IDL way: Find last non-zero value By: Jeremy Bailin on Mon, 24 August 2009 19:43
		Re: The IDL way: Find last non-zero value By: penteado on Mon, 24 August 2009 18:35
		Re: The IDL way: Find last non-zero value By: Jeremy Bailin on Mon, 24 August 2009 17:12
		Re: The IDL way: Find last non-zero value By: Eric Hudson on Mon, 24 August 2009 13:42
		Re: The IDL way: Find last non-zero value By: cgguido on Sun, 23 August 2009 20:39
		Re: The IDL way: Find last non-zero value By: penteado on Sun, 23 August 2009 19:10
		Re: The IDL way: Find last non-zero value By: Chris[6] on Sun, 23 August 2009 17:32

Previous Topic:	Re: Histogram Too Many Array Elements
Next Topic:	Integer - decimal help

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

] [

]

Current Time: Sat Oct 11 16:16:42 PDT 2025

Total time taken to generate the page: 0.79969 seconds