[Jprogramming] Fwd: Re: extended precision essay

['Mike Day' via Programming]

Yes, I agree...  I had hoped to draw Roger out on what he had in mind 
when including those

statements in _his_ sinh function which I quoted below.  The cosh 
function also has non-productive

if statements.

I suspect he included them to be instructive in some way; maybe the 
essay had included comments

about them which he later removed.

Thanks, both,

Mike


On 29/04/2019 21:47, Devon McCormick wrote:
> As an experiment, I compared three versions of "sinh": the one we've been
> discussing, then a version with the first "if." removed and another with
> both removed.  For about 2000 random arguments, they all return the same
> value.  Not conclusive but indicative that the lines have no effect, as one
> might guess from looking at the code.
>
> On Mon, Apr 29, 2019 at 4:17 PM Raul Miller <rauldmiller at gmail.com> wrote:
>
>> Um... sure?
>>
>>      sinh=: DP&$: : (4 : 0) " 0
>>         if. x>10^.|y do. (1r2*_1^0>y)*x exp |y end.
>>         if. 1<|y do. -:-/x exp y,-y end.
>>         e=. 0.1^x
>>         n=. e (>i.1:) y (^%!@]) 1+2*i.>.0.5*y^.e
>>         +/(x:!.0 y) (^%!@]) 1+2x*i.n
>>      )
>>
>> Your if statements (when triggered) calculate values and then discards
>> them.
>>
>> If, for example, you changed that to:
>>
>>      sinh=: DP&$: : (4 : 0) " 0
>>         if. x>10^.|y do. (1r2*_1^0>y)*x exp |y return.end.
>>         if. 1<|y do. -:-/x exp y,-y return.end.
>>         e=. 0.1^x
>>         n=. e (>i.1:) y (^%!@]) 1+2*i.>.0.5*y^.e
>>         +/(x:!.0 y) (^%!@]) 1+2x*i.n
>>      )
>>
>> Then your if statements would be computing values which are the result
>> of your sinh.
>>
>> Or, if you just removed those two lines, your code would produce the
>> same results it currently produces (with the slight possibility of
>> eliminating some runtime error cases).
>>
>> (I've not attempted to look at your code for numerical accuracy, I am
>> just commenting on its structure.)
>>
>> I hope this makes sense?
>>
>> Thanks,
>>
>>
>> --
>> Raul
>>
>> On Mon, Apr 29, 2019 at 1:33 PM 'Mike Day' via Programming
>> <programming at jsoftware.com> wrote:
>>> As may be seen,  Roger Hui felt unable to advise re my private enquiry
>>> on one of his essays
>>> in the J Wiki.  Anyone else feel like commenting on the apparently
>>> unproductive lines of
>>> code?
>>> Thanks,
>>> Mike
>>>
>>>
>>> -------- Forwarded Message --------
>>> Subject:        Re: extended precision essay
>>> Date:   Wed, 24 Apr 2019 19:00:47 -0700
>>> From:   Roger Hui <rogerhui.canada at gmail.com>
>>> To:     Mike Day <mike_liz.day at tiscali.co.uk>
>>>
>>>
>>>
>>> Hi Mike.  Thanks for you msg. I am heavily engaged in other matters now,
>>> so if you want to have a more timely answer I suggest you query
>>> programming at jsoftware.com <mailto:programming at jsoftware.com>.
>>>
>>>
>>>
>>> On Tue, Apr 23, 2019 at 6:58 AM Mike Day <mike_liz.day at tiscali.co.uk
>>> <mailto:mike_liz.day at tiscali.co.uk>> wrote:
>>>
>>>      Dear Roger,
>>>
>>>      I've been trying to use material from your Essay on extended
>> precision
>>>      functions at
>>>
>>>
>> https://code.jsoftware.com/wiki/Essays/Extended_Precision_Functions#Utilities
>>>      and am puzzled by apparently unproductive code in the hyperbolic
>>>      functions,
>>>
>>>      eg
>>>
>>>      sinh=: DP&$: : (4 : 0) " 0
>>>         if. x>10^.|y do. (1r2*_1^0>y)*x exp |y end.
>>>         if. 1<|y do. -:-/x exp y,-y end.
>>>         e=. 0.1^x
>>>         n=. e (>i.1:) y (^%!@]) 1+2*i.>.0.5*y^.e
>>>         +/(x:!.0 y) (^%!@]) 1+2x*i.n
>>>      )
>>>
>>>      I would expect the if statements either
>>>
>>>           to modify or produce values for use later in the function
>>>
>>>      or
>>>
>>>          to have a "return." before the "end."
>>>
>>>      Not at all important,  but I'd welcome your ideas when you have time.
>>>
>>>
>>>      Very sad about John Scholes.  I'd guess you enjoyed your
>> collaboration
>>>      at Dyalog,
>>>
>>>      Best Wishes
>>>
>>>      Mike
>>>
>>>
>>>
>>>      ---
>>>      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
>> ----------------------------------------------------------------------
>> For information about J forums see http://www.jsoftware.com/forums.htm
>>
>