@Carl Love Unless there is some weird branch cut they implicitly assume this looks like a bug to me. [try assuming x::complex]

@Rouben Rostamian  This should give the Dirac function as was evaluated above [this is the spherical Bessel closure relation]. It's not undefined.

oddOne.mw @tomleslie 

Here it is... Thanks for your help

@tomleslie 

I'm running on a mac [version 11.2.1, intel i9] and the output on your sheet is

"Standard Worksheet Interface, Maple 2020.2, Mac OS X, November 11 2020 Build ID 1502365

The "Physics Updates" package is not installed"

Note that RealDomain:-solve works if the numbers are altered slightly, it's just this particular one has led somewhere to calling assuming over a range of length zero. 

@chrisc realdomain.mw

@tomleslie I'm using the same version. Here's a worksheet with it. Weird!

@acer No, I always have them closed. 

@acer I meant Worksheet rather than workbook -- but yeah the standard GUI

@Axel Vogt yes but that is the point! the analytical answer is wrong

Great, thanks for confirming. Anyone know how to take it up with Maple?

Thanks again for your help.

Thanks again for your help.

Thanks for this. You have inspired me to use:

A_:=unapply(Spikes,Z):
A:=proc(zz)
if not zz::numeric then return 'procname'(args) fi;
return evalhf(A_(zz));
end proc: 

I understand what you're saying about not needing to define new procedures like this for things like this for use with plot, but I've always found that shielding what's in a procedure starting with "if not zz::numeric then return 'procname'(args) fi;" to be robust - for example, with some more if statements in the procedure, built in NAG integration routines only work with this in it. Perhaps you know a neater solution? 

Thanks for this. You have inspired me to use:

A_:=unapply(Spikes,Z):
A:=proc(zz)
if not zz::numeric then return 'procname'(args) fi;
return evalhf(A_(zz));
end proc: 

I understand what you're saying about not needing to define new procedures like this for things like this for use with plot, but I've always found that shielding what's in a procedure starting with "if not zz::numeric then return 'procname'(args) fi;" to be robust - for example, with some more if statements in the procedure, built in NAG integration routines only work with this in it. Perhaps you know a neater solution?