Hi Oleg,
That's funny: in the Discord you say you have trouble parsing my code
and that you "know almost nothing about approximation", next thing I
know, you send an improvement on my work, written in 2 languages! :)
On a more serious note: I have no idea how to get this working on NixOS,
sorry.
I have also noticed the bug around the edge values with .cub, and have
mentioned it in the docs. Have you compared the two implementations
while staying away from the edges?
Don't leave us hanging! :)
I have never heard of Taylor, so I looked it up.
Is this what you mean?
https://brilliant.org/wiki/taylor-series-approximation/
I don't understand any of that yet, sorry.
Speaking of understanding: is there anything I can do to help you
understand tabulateNd?
Do you have any questions, would you like to chat on a medium of your
choice? (right here, irc, telephone, whatever)
I'd love to get some more feedback on it as I have no idea if it's
correct what I'm doing, I just made it all up. :)
Cheers,
Bart.
---
On 2023-05-09 22:46, Oleg Nesterov wrote:
On 05/09, Oleg Nesterov wrote:
See the attached test-case. If you want to compile it, you need
fpp (https://github.com/oleg-nesterov/fpp), maxima, and Linux.
Compiled with '-a plot.cpp'
$ ./test-plot -n 100000 | tail -n 1
0.00285841338 0.000623551081
As you can see, taylor() uses less memory and it is more accurate
at least in this particular case.
On second thought...
I recalled that ba.tabulate().cub works badly near r0 and r1 and I
guess
tabulateNd().cub inherits this problem.
So perhaps bicubic will work better than taylor if this problem is
fixed.
Oleg.
_______________________________________________
Faudiostream-users mailing list
Faudiostream-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/faudiostream-users
