|
|
| Re: Strange WHERE issue? [message #76286 is a reply to message #76272] |
Thu, 26 May 2011 19:40  |
Fabzou
Messages: 76
Registered: November 2010
|
Member |
|
|
Ah, and the first line :
> c = WHERE(chisqarr_ph[1,*] EQ MIN(chisqarr_ph[1,*]))
could be replaced by:
dummy = MIN(chisqarr_ph[1,*], c)
But this is just detail.
On 05/27/2011 02:52 AM, polystethylene wrote:
> Wonder if anybody can help me with this...
>
> Essentially I'm just varying a parameter (1st column) and testing chi
> squared (2nd column), and I want to find chi squared minimum + 1 in
> order find my 1 sigma error bars...
>
> I have the following code:
>
> c = WHERE(chisqarr_ph[1,*] EQ MIN(chisqarr_ph[1,*]))
>
>
> d = WHERE(chisqarr_ph[1,*] LE (chisqarr_ph[1,c]+1))
>
>
>
> I have a problem with the line d = WHERE(chisqarr_ph[1,*] LE
> (chisqarr_ph[1,c]+1)), which returns -1.
>
> What I don't understand is that when I print chisqarr_ph[1,c]+1, it
> correctly returns 1045.4516.
> Here is the section of chisqarr_ph[*,*] where chisqarr_ph[1,*] is LE
> (chisqarr_ph[1,c]+1):
>
>
> 0.0030200630 1045.4305 6.5339409
> 0.0030400632 1045.3898 6.5336863
> 0.0030600633 1045.3513 6.5334455
> 0.0030800635 1045.3134 6.5332085
> 0.0031000637 1045.2760 6.5329750
> 0.0031200638 1045.2392 6.5327453
> 0.0031400640 1045.2011 6.5325071
> 0.0031600641 1045.1639 6.5322746
> 0.0031800643 1045.1274 6.5320465
> 0.0032000644 1045.0914 6.5318213
> 0.0032200646 1045.0560 6.5315999
> 0.0032400647 1045.0214 6.5313835
> 0.0032600649 1044.9917 6.5311983
> 0.0032800650 1044.9634 6.5310215
> 0.0033000652 1044.9358 6.5308485
> 0.0033200653 1044.9088 6.5306800
> 0.0033400655 1044.8827 6.5305167
> 0.0033600656 1044.8573 6.5303578
> 0.0033800658 1044.8333 6.5302083
> 0.0034000659 1044.8113 6.5300706
> 0.0034200661 1044.7896 6.5299349
> 0.0034400662 1044.7685 6.5298032
> 0.0034600664 1044.7481 6.5296754
> 0.0034800665 1044.7274 6.5295459
> 0.0035000667 1044.7072 6.5294197
> 0.0035200668 1044.6876 6.5292974
> 0.0035400670 1044.6687 6.5291792
> 0.0035600672 1044.6505 6.5290653
> 0.0035800673 1044.6329 6.5289559
> 0.0036000675 1044.6161 6.5288504
> 0.0036200676 1044.5997 6.5287478
> 0.0036400678 1044.5836 6.5286474
> 0.0036600679 1044.5683 6.5285520
> 0.0036800681 1044.5540 6.5284624
> 0.0037000682 1044.5403 6.5283766
> 0.0037200684 1044.5282 6.5283011
> 0.0037400685 1044.5186 6.5282411
> 0.0037600687 1044.5096 6.5281852
> 0.0037800688 1044.5013 6.5281333
> 0.0038000690 1044.4936 6.5280852
> 0.0038200691 1044.4866 6.5280414
> 0.0038400693 1044.4803 6.5280016
> 0.0038600694 1044.4746 6.5279665
> 0.0038800696 1044.4697 6.5279355
> 0.0039000697 1044.4653 6.5279080
> 0.0039200699 1044.4614 6.5278838
> 0.0039400700 1044.4581 6.5278632
> 0.0039600702 1044.4554 6.5278460
> 0.0039800704 1044.4532 6.5278327
> 0.0040000705 1044.4519 6.5278244
> 0.0040200707 1044.4516 6.5278223
> 0.0040400708 1044.4532 6.5278322
> 0.0040600710 1044.4574 6.5278584
> 0.0040800711 1044.4621 6.5278881
> 0.0041000713 1044.4674 6.5279212
> 0.0041200714 1044.4733 6.5279582
> 0.0041400716 1044.4787 6.5279918
> 0.0041600717 1044.4849 6.5280308
> 0.0041800719 1044.4919 6.5280741
> 0.0042000720 1044.4992 6.5281198
> 0.0042200722 1044.5071 6.5281691
> 0.0042400723 1044.5157 6.5282233
> 0.0042600725 1044.5272 6.5282952
> 0.0042800726 1044.5398 6.5283739
> 0.0043000728 1044.5530 6.5284564
> 0.0043200729 1044.5669 6.5285431
> 0.0043400731 1044.5815 6.5286344
> 0.0043600732 1044.5970 6.5287310
> 0.0043800734 1044.6138 6.5288366
> 0.0044000736 1044.6328 6.5289551
> 0.0044200737 1044.6521 6.5290758
> 0.0044400739 1044.6721 6.5292004
> 0.0044600740 1044.6926 6.5293290
> 0.0044800742 1044.7136 6.5294597
> 0.0045000743 1044.7351 6.5295941
> 0.0045200745 1044.7572 6.5297325
> 0.0045400746 1044.7800 6.5298750
> 0.0045600748 1044.8035 6.5300219
> 0.0045800749 1044.8277 6.5301733
> 0.0046000751 1044.8526 6.5303288
> 0.0046200752 1044.8779 6.5304871
> 0.0046400754 1044.9036 6.5306472
> 0.0046600755 1044.9299 6.5308120
> 0.0046800757 1044.9571 6.5309819
> 0.0047000758 1044.9849 6.5311559
> 0.0047200760 1045.0144 6.5313397
> 0.0047400761 1045.0466 6.5315415
> 0.0047600763 1045.0796 6.5317475
> 0.0047800764 1045.1132 6.5319574
> 0.0048000766 1045.1474 6.5321712
>
> Given that there are rows in the array for which the condition is
> true, can anybody see why WHERE is returning -1?
>
> It's probably me have a dim moment, but I can't for the life of me
> understand what's going on...
>
> Many thanks for any help !
>
> Stef
|
|
|
|
| Re: Strange WHERE issue? [message #76287 is a reply to message #76286] |
Thu, 26 May 2011 19:34 |
polystethylene
Messages: 28
Registered: February 2009
|
Junior Member |
|
|
Hi guys,
Thanks for the replies, that's fixed the issue; took the expression
out of the boolean evaluation and set it to another variable, then
used 'variable[0]' in the where clause.
I was having a dim moment of sorts though; I've run into this before
and I'm fairly sure now that I even asked the same question here 2
years ago.
You were all quick to respond and equally helpful back then, too :)
Cheers,
Stef
|
|
|
|
| Re: Strange WHERE issue? [message #76288 is a reply to message #76287] |
Thu, 26 May 2011 19:24 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Ben Tupper writes:
> It really isn't a WHERE issue, though. The meat of the issue is how IDL
> handles the mismatched array lengths.
Yeah, what Ben said, which I am going to write
down, because that was as good an explanation
as we are ever going to get! :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.idlcoyote.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Strange WHERE issue? [message #76289 is a reply to message #76288] |
Thu, 26 May 2011 19:05 |
ben.bighair
Messages: 221
Registered: April 2007
|
Senior Member |
|
|
Hi,
On 5/26/11 8:52 PM, polystethylene wrote:
> Wonder if anybody can help me with this...
>
> Essentially I'm just varying a parameter (1st column) and testing chi
> squared (2nd column), and I want to find chi squared minimum + 1 in
> order find my 1 sigma error bars...
>
> I have the following code:
>
> c = WHERE(chisqarr_ph[1,*] EQ MIN(chisqarr_ph[1,*]))
>
>
> d = WHERE(chisqarr_ph[1,*] LE (chisqarr_ph[1,c]+1))
>
>
>
> I have a problem with the line d = WHERE(chisqarr_ph[1,*] LE
> (chisqarr_ph[1,c]+1)), which returns -1.
>
> What I don't understand is that when I print chisqarr_ph[1,c]+1, it
> correctly returns 1045.4516.
THis has to do with how IDL handles arrays of differing lengths. It
turns out that your variable 'd' is an array (of length 1, so it looks
like a scalar but it isn't). WHERE always returns a vector except when
all of the values in the input argument are zero. (I know, it isn't
consistent.) You want to make sure that you are using a scalar in the
comparison as part of the comparison argument to WHERE. David Fanning
has the scoop here...
http://www.idlcoyote.com/misc_tips/brokenwhere.html
It really isn't a WHERE issue, though. The meat of the issue is how IDL
handles the mismatched array lengths.
Here, IDL 'knows' to scan the entire length of the first argument
IDL> print, [5,6,7,8] EQ 5
1 0 0 0
IDL> print, [5,6,7,8] EQ 7
0 0 1 0
Change the second argument to an array (of length 1) and things get
trickier. In this case, IDL 'knows' to scan only the length of the
shorter of the two arguments - comparing element-by-element.
IDL> print, [5,6,7,8] EQ [5]
1
IDL> print, [5,6,7,8] EQ [7]
0
It makes more sense when you add a second element to the second argument.
IDL> print, [5,6,7,8] EQ [5,7]
1 0
So, the 5s matched but the 6 and 7 did not. Bummer. This is why I
often write variables that I think should be scalar as 'x[0]' - it looks
silly but it prevents gotchas like this. Of course, there are times
when one could exploit IDL's behavior - but I can't think of one right
now.
Cheers,
Ben
> Here is the section of chisqarr_ph[*,*] where chisqarr_ph[1,*] is LE
> (chisqarr_ph[1,c]+1):
>
>
> 0.0030200630 1045.4305 6.5339409
> 0.0030400632 1045.3898 6.5336863
> 0.0030600633 1045.3513 6.5334455
> 0.0030800635 1045.3134 6.5332085
> 0.0031000637 1045.2760 6.5329750
> 0.0031200638 1045.2392 6.5327453
> 0.0031400640 1045.2011 6.5325071
> 0.0031600641 1045.1639 6.5322746
> 0.0031800643 1045.1274 6.5320465
> 0.0032000644 1045.0914 6.5318213
> 0.0032200646 1045.0560 6.5315999
> 0.0032400647 1045.0214 6.5313835
> 0.0032600649 1044.9917 6.5311983
> 0.0032800650 1044.9634 6.5310215
> 0.0033000652 1044.9358 6.5308485
> 0.0033200653 1044.9088 6.5306800
> 0.0033400655 1044.8827 6.5305167
> 0.0033600656 1044.8573 6.5303578
> 0.0033800658 1044.8333 6.5302083
> 0.0034000659 1044.8113 6.5300706
> 0.0034200661 1044.7896 6.5299349
> 0.0034400662 1044.7685 6.5298032
> 0.0034600664 1044.7481 6.5296754
> 0.0034800665 1044.7274 6.5295459
> 0.0035000667 1044.7072 6.5294197
> 0.0035200668 1044.6876 6.5292974
> 0.0035400670 1044.6687 6.5291792
> 0.0035600672 1044.6505 6.5290653
> 0.0035800673 1044.6329 6.5289559
> 0.0036000675 1044.6161 6.5288504
> 0.0036200676 1044.5997 6.5287478
> 0.0036400678 1044.5836 6.5286474
> 0.0036600679 1044.5683 6.5285520
> 0.0036800681 1044.5540 6.5284624
> 0.0037000682 1044.5403 6.5283766
> 0.0037200684 1044.5282 6.5283011
> 0.0037400685 1044.5186 6.5282411
> 0.0037600687 1044.5096 6.5281852
> 0.0037800688 1044.5013 6.5281333
> 0.0038000690 1044.4936 6.5280852
> 0.0038200691 1044.4866 6.5280414
> 0.0038400693 1044.4803 6.5280016
> 0.0038600694 1044.4746 6.5279665
> 0.0038800696 1044.4697 6.5279355
> 0.0039000697 1044.4653 6.5279080
> 0.0039200699 1044.4614 6.5278838
> 0.0039400700 1044.4581 6.5278632
> 0.0039600702 1044.4554 6.5278460
> 0.0039800704 1044.4532 6.5278327
> 0.0040000705 1044.4519 6.5278244
> 0.0040200707 1044.4516 6.5278223
> 0.0040400708 1044.4532 6.5278322
> 0.0040600710 1044.4574 6.5278584
> 0.0040800711 1044.4621 6.5278881
> 0.0041000713 1044.4674 6.5279212
> 0.0041200714 1044.4733 6.5279582
> 0.0041400716 1044.4787 6.5279918
> 0.0041600717 1044.4849 6.5280308
> 0.0041800719 1044.4919 6.5280741
> 0.0042000720 1044.4992 6.5281198
> 0.0042200722 1044.5071 6.5281691
> 0.0042400723 1044.5157 6.5282233
> 0.0042600725 1044.5272 6.5282952
> 0.0042800726 1044.5398 6.5283739
> 0.0043000728 1044.5530 6.5284564
> 0.0043200729 1044.5669 6.5285431
> 0.0043400731 1044.5815 6.5286344
> 0.0043600732 1044.5970 6.5287310
> 0.0043800734 1044.6138 6.5288366
> 0.0044000736 1044.6328 6.5289551
> 0.0044200737 1044.6521 6.5290758
> 0.0044400739 1044.6721 6.5292004
> 0.0044600740 1044.6926 6.5293290
> 0.0044800742 1044.7136 6.5294597
> 0.0045000743 1044.7351 6.5295941
> 0.0045200745 1044.7572 6.5297325
> 0.0045400746 1044.7800 6.5298750
> 0.0045600748 1044.8035 6.5300219
> 0.0045800749 1044.8277 6.5301733
> 0.0046000751 1044.8526 6.5303288
> 0.0046200752 1044.8779 6.5304871
> 0.0046400754 1044.9036 6.5306472
> 0.0046600755 1044.9299 6.5308120
> 0.0046800757 1044.9571 6.5309819
> 0.0047000758 1044.9849 6.5311559
> 0.0047200760 1045.0144 6.5313397
> 0.0047400761 1045.0466 6.5315415
> 0.0047600763 1045.0796 6.5317475
> 0.0047800764 1045.1132 6.5319574
> 0.0048000766 1045.1474 6.5321712
>
> Given that there are rows in the array for which the condition is
> true, can anybody see why WHERE is returning -1?
>
> It's probably me have a dim moment, but I can't for the life of me
> understand what's going on...
>
> Many thanks for any help !
>
> Stef
|
|
|
|
| Re: Strange WHERE issue? [message #76290 is a reply to message #76289] |
Thu, 26 May 2011 19:01 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
polystethylene writes:
>
> Wonder if anybody can help me with this...
>
> Essentially I'm just varying a parameter (1st column) and testing chi
> squared (2nd column), and I want to find chi squared minimum + 1 in
> order find my 1 sigma error bars...
>
> I have the following code:
>
> c = WHERE(chisqarr_ph[1,*] EQ MIN(chisqarr_ph[1,*]))
>
>
> d = WHERE(chisqarr_ph[1,*] LE (chisqarr_ph[1,c]+1))
>
>
>
> I have a problem with the line d = WHERE(chisqarr_ph[1,*] LE
> (chisqarr_ph[1,c]+1)), which returns -1.
This is one of the classic Where gotchas:
http://www.idlcoyote.com/misc_tips/noidea.html
You are only comparing the *first* value in your
array, because the value on the right of the Boolean
expression is a vector of one element.
And, you should be VERY careful using these WHERE
expressions with floating values. You will not get
the results you expect at least some of the time.
See, for example, this article:
http://www.idlcoyote.com/math_tips/razoredge.html
You might want to have a look at the program Floats_Equal
in the Coyote Library:
http://www.idlcoyote.com/programs/floats_equal.pro
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.idlcoyote.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
