| Re: How to speed up KRIG2D by 30x [message #86139 is a reply to message #86129] |
Thu, 10 October 2013 09:28   |
chris_torrence@NOSPAM
Messages: 528
Registered: March 2007
|
Senior Member |
|
|
On Thursday, October 10, 2013 5:44:05 AM UTC-6, Mike wrote:
> Hi Chris,
>
>
>
> Thank you for your prompt response.
>
>
>
> You are right, I did notice the covariances were not right. I rewrote them as variograms instead, like below:
>
>
>
> FUNCTION Krig_sphere, rad, c ;Return Spherical variogram Fcn
>
> r = rad/c[0] < 1.0
>
> return, c[1] + c[2] * (r*(1.5 - 0.5*r^2))
>
> end
>
> FUNCTION Krig_expon, rad, c ;Return Exponential variogram Fcn
>
> return, c[1] + c[2] * (1.0 - exp(-3.0*rad/c[0]))
>
> end
>
>
>
> Another small change I made was to use the two lines below instead of the double loop to fill the array coefficients
>
>
>
> a[0,0] = distance_measure( transpose([[x],[y]]), /MATRIX )
>
> a = call_function(fname, a, t) ; Get coefficient matrix
>
>
Hey Mike,
So when you are using the variograms, what does the rest of the code look like? I just want to make sure that we get the same answer.
-Chris
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
