Possible bug in af.approx1()

Question

Possible bug in af.approx1()

shyams2 opened this issue 7 years ago · comments

Shyam Sundar Sankaran commented 7 years ago

I've written a file comparing the results of the interpolate function in scipy libraries, and arrayfire:

http://nbviewer.jupyter.org/gist/ShyamSS-95/9267f7f860f4ce5aa5ea785c937bbfda

I don't seem to be getting the same result for both the functions.
af.approx1() seems to be quite off from the expected curve.

Is there a bug here?

Pavan Yalamanchili · Answer 1 · Tue Feb 28 2017 05:04:46 GMT+0800 (China Standard Time)

@ShyamSS-95 can you plot asin(interpolated - actual) ?

Pavan Yalamanchili · Answer 2 · Tue Feb 28 2017 05:05:49 GMT+0800 (China Standard Time)

@ShyamSS-95 wait that wouldn't work. I keep forgetting my trigonometric formulas :) It looks like the two curves are phase shifted. I'll investigate it once I get home.