For interest, I had a look at the changes in the script.
I tried both derivatives, and both failed in different ways:
a) polynomials using p. :
(1 2 3 & p. ) deriv 1 NB. result should be 2 6&p.
0&p.
(0 & p. ) deriv 1 NB. ok!
0&p.
b) f() ^ h()
(*: ^ *:) deriv 1
|domain error: deriv_jcalculus_
| 13!:8(3)
After some puzzlement, I traced the difficulties with each:
a) line 170
if. coeffs=. -: '' do. '0&p.' return.
should of course be
if. coeffs -: '' do. '0&p.' return.
b) I couldn't see what was wrong until I realised that fexp is undefined!
So adding a definition along the lines of the other fops removes the
error condition.
fexp =: 4 : 0
'((',x,')^(',y,'))'
)
I wouldn't dare try altering the online source code!
Cheers,
Mike
On 05/02/2021 04:34, 'Jon Hough' via Programming wrote:
I wrote a bug report and fix in the source chat a couple of weeks ago. There
was no response, so I guess I wrote in the wrong chat.
The calculus addon script has one glaring issue, and one small issue with an
easy fix.
The glaring issue is the inability to differentiate the constant zero function
0&p.
that gives an error.
The fix is here: https://github.com/jsoftware/math_calculus/pull/2
I think this should be fixed asap, as it also causes an error on double
differentiation of a linear polynomial.
1 2&p.
The other issue is differentiating a general f^g function, which currently
gives an error, but is easily doable.
Thanks,
Jon
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
--
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
