| Re: svdfit-measure_errors [message #72098] |
Tue, 10 August 2010 08:47 |
Wout De Nolf
Messages: 194
Registered: October 2008
|
Senior Member |
|
|
On Sun, 8 Aug 2010 10:48:00 -0700 (PDT), sid <gunvicsin11@gmail.com>
wrote:
> Hi,
> I am fitting my data with 3 degree polynomial, so im using svdfit,
> in that im setting the measure_errors keyword to a named variable, say
> suppose,
> idl>n=n_elements(y)
> idl>e=dblarr(n)
> idl>res=svdfit(x,y,4,measure_errors=e,yfit=y1)
> but if I print e an array is displayed with full of zeros and also I
> get warning message like this, '% SVDFIT: Warning: 3 singular values
> found.hy this happens'.
> Any of you please explain me why it happens. Moreover if I don't set
> this keyword then im not getting any warning like the above.
> what should I do to get rid of this.
> regards
> sid
MEASURE_ERRORS is something you give to svdfit as input, not something
you get as output. It should contain the standard deviation on each
y-value (e.g. measurement error). In the code-snip above, you tell
svdfit that all errors are zero. Apparently svdfit doesn't like that.
So ask yourself, what does y mean? Should some y-values be considered
"more important" or "more reliable" than others? If not, just omit the
MEASURE_ERRORS keyword.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
