2-D, 3-D, n-D data interpolation - Is it possible to extrapolate?

Hello,

I'm having the following situation, I have a 2-D Look-up Table in Simulink using the below data.

row_index = [5 25 35]

col_index = [-20 -3 0 5 40]

table_data =

    3.0000    1.0000         0    1.5000    2.0000
    3.0000    1.0000         0    1.5000    2.0000
    3.0000    1.0000         0    1.5000    2.0000

What I'm trying to do is to have the same result (as the 2-D Look-up Table output from Simulink) using Matlab code/functions. I've tried using interp2, but if the values are extrapolated/out of rage, I get a NaN. I'm using version 7.13 (Matlab 2011) and I don't know if is there a solution for this.

For example interp2(col_index,row_index,table_data,-20,5) returns 3 witch is correct.

But if I use interp2(col_index,row_index,table_data,5,-20) returns NaN witch is also correct according to the function description, but in Simulink the 2-D Look-up Table returns 1.5

I'm asking you if is there a solution to get the same result as the 2-D look-up table from simulink. And the same for 3-D or n-D.

Thanks,

Alex

Answer 1

[Sean de Wolski]

You can set the 'ExtrapVal' to be non-nan:

doc interp2

Of course this is dangerous:

http://mrkwr.wordpress.com/2006/11/08/mark-twain-on-the-perils-of-extrapolation/

You may also like one fo the FEX extrapolation tools.