jcalculus - limitations and compatibilities

[More Rice]

In j701, we can do:

   *: d. _1

to get 1st integral of x; but jcalculus would block:

|syntax error in pderiv_jcalculus_, unexecutable fragment (adv noun)

|       error 3;'Integrals not allowed'

Press ENTER to inspect

Q1. Any plan to support?

In j701, the following will give "domain error;" but pderiv_jcalculus_ will run forever:

   ^ d. _

Q2. Does it make sense to make it behave like ^:_ (stops and returns once converged)?

thanks

Maurice

[Henry Rich]

Long ago we realized that it was a mistake to implement calculus in C.  C is not the right language.  J will do.  We moved most of the C code into a J addon.

Jsoftware has no plans to augment the addon.  It is in the hands of the users, who have already made many improvements.  If you want the features you mention, you are encouraged to add them to the addon, or persuade someone else to.

Henry Rich

To unsubscribe from this group and stop receiving emails from it, send an email to forum+un...@jsoftware.com.

[More Rice]

Thanks Henry. Noted. Though, it isn't news - when the removal was 1st announced, I think you already explained.

I was asking the J community.  A key thing to note is that I didn't ask for a timeline (no intention to pressure anyone or entity), as I don't have mine either.  Free to ignore; I'll move on.

thanks.

[Raul Miller]

On Mon, Dec 2, 2024 at 9:17 PM More Rice <mrmor...@gmail.com> wrote:
> Q1. Any plan to support?

Would you like to provide support?

Support tends to be frustrating, as problems tend to become visible
and sometimes feel insurmountable. This is very much an issue with
generalized math, as it involves an unbounded number of unbounded
problems.

> Q2. Does it make sense to make it behave like ^:_ (stops and returns once converged)?

Yes, where possible.

Remember: Two functions are equal if they have the same domain and
codomain and their values are the same for all elements of the domain.
And while we treat the integer, real and complex domains, ultimately
we're only dealing with finite subsets of them. And this requires
judgement calls - issues which we are forced to neglect while we try
to address higher priorities.

Meanwhile, though, as a community, one of our priorities also has to
be bringing in new people and especially new potential contributors.

Welcome,

--
Raul

[More Rice]

Thank Raul.

>>   *: pderiv _1

>>|syntax error in pderiv_jcalculus_, unexecutable fragment (adv noun)

>>|       error 3;'Integrals not allowed'

>>Press ENTER to inspect

>> Q1. Any plan to support?
> Would you like to provide support?

I had a confusion - I was misled by the pro's documentation (html):

"u d. n is like u D. n except that u is treated as a rank-0 function."

I automatically thought since d. _1 is supported, so should D. _1 .  It turns out pderiv _1 's failure is consistent with D. :

   ^ ("0) D. _1  NB. j701

|nonce error

|       ^("0)D._1

So, no. I supposed it is good the way it is.  My apology for the confusion.

>>   ^ deriv _

>> Q2. Does it make sense to make it behave like ^:_ (stops and returns once converged)?

> Yes, where possible.

I don't see mathematical use of deriv _ , but only to make runtime more stable.  Perhaps it is why d. _ blocks by issuing domain error right from the get-go (?) - certainly an efficient approach.

> Meanwhile, though, as a community, one of our priorities also has to

> be bringing in new people and especially new potential contributors.

I don't hang around because I've nothing to do. I learned quite a bit from helpful folks in the forum (I can easily name a bunch; Raul, you amoung them. Thank you!) and will gladly reciprocate the same kindness.

(I hope to find time to dig into calculus.ijs over the holiday and look for small/achievable goals a journeyman can do; but want to stay out of the way of other J masters who may be doing some facelifting to the file.  Thus, I posted the questions to start a conversation.)

thanks, and happy holidays!

[Raul Miller]

On Fri, Dec 6, 2024 at 9:40 PM More Rice <mrmor...@gmail.com> wrote:
> I don't see mathematical use of deriv _ , but only to make runtime more stable. Perhaps it is why d. _ blocks by issuing domain error right from the get-go (?) - certainly an efficient approach.

Maybe https://math.libretexts.org/Bookshelves/Differential_Equations/Introduction_to_Partial_Differential_Equations_(Herman)/07%3A_Green's_Functions/7.06%3A_Method_of_Eigenfunction_Expansions
would be of interest? (Though that page is rather slow to come up.
Fortunately, if it breaks from overload, web.archive.org has a
snapshot of it.)

--
Raul