Setting Divide by Zero Equal to Zero in Maxima

Dear Professor,

Are you suggesting that Isabelle/HOL, Moore and Penrose have gone " "... 
off the tracks without notice". " ?

Where is the wreckage?

Regards,

Jalaluddin

---------------------------------------------------------------------------------------------

The following correspondence is in reverse chronological order:

---------------------------------------------------------------------------------------------
On 23/02/11 23:16 PM, Richard Fateman wrote:
> On 2/23/2011 4:28 AM, Jalaluddin Morris wrote:
>> Dear Professor,
>>
>> I deny saying that "... the following theorem a/b + c/d = (a ? d + b 
>> ? c)/(b ? d) becomes false." - in fact I used word "inoperative".
> Substituting the word "inoperative" for "false" in my statement does 
> not change the fact that this is quite a loss.
>
>> As for "quite a loss", where is the relative gain from (1 / 0 
>> undefined) causing Maxima to stop dead in its tracks, or (1 / 0 = 
>> infinity) - 
> If you wish to redefine division so that 1/0 becomes infinity, that 
> can be tolerated in some models of arithmetic.  (See for example,
> discussion of the IEEE floating-point system!) That does not imply 0/0 
> is infinity.   To change x/0 to 0 is to provide an alternative
> to "stop dead in its tracks"   that is  "to go off the tracks without 
> notice".
>
>> useful if we wish to compactify the reals
> If you said x/0 is changed to infinity for x>0,    or perhaps any 
> non-zero x,  that's possibly ok.  for x/0 -> 0, that's news to me.
>> (but does Maxima need to?)?
> Who knows.  One can also carry around (x/0) in one's calculations  
> like a raisin in a cookie, at least for a while.
> If someone multiplies it by zero, it will probably go away.
> RJF
>>
>> Regards,
>>
>> Jalaluddin
>>
>> ************************************************************************** 
>>
>>
>> On 22/02/11 23:19 PM, Richard Fateman wrote:
>>> On 2/22/2011 4:25 AM, Jalaluddin Morris wrote:
>>>> Dear List Members,
>>>>
>>>> I have attached a pdf addressing the question as to why one might wish
>>>> to have the option of x / 0 = 0 in Maxima.
>>>>
>>>> On the subject of Isabelle/Hol, members may wish to download the
>>>> software from
>>>> http://www.cl.cam.ac.uk/research/hvg/Isabelle/
>>>> and familiarize themselves with the GUI, despite the steep learning 
>>>> curve.
>>>>
>>>> Regards,
>>>>
>>>> Jalaluddin
>>>>
>>>
>>> As your paper says, the following theorem
>>>
>>> a/b + c/d = (a ? d + b ? c)/(b ? d)
>>>
>>> becomes false. That is quite a loss. My suggestion is that you change
>>> your computation and not use "/" but some other notation, and then
>>> define what you means by it through programs and perhaps tellsimp.
>>>
>>> RJF
>>>
>>>
>> ************************************************************************** 
>>
>
>
--------------------------------------------------------------------
from: Jalaluddin Morris
23/02/11 20:28 PM

Re: [Maxima] Setting Divide by Zero Equal to Zero in Maxima

Dear Professor,

I deny saying that "... the following theorem a/b + c/d = (a ? d + b ? 
c)/(b ? d) becomes false." - in fact I used word "inoperative". As for 
"quite a loss", where is the relative gain from (1 / 0 undefined) 
causing Maxima to stop dead in its tracks, or (1 / 0 = infinity) - 
useful if we wish to compactify the reals (but does Maxima need to?)?

Regards,

Jalaluddin

--------------------------------------------------------------------
from: Richard Fateman
22/02/11 23:19 PM

subject: Re: [Maxima] Setting Divide by Zero Equal to Zero in Maxima


On 2/22/2011 4:25 AM, Jalaluddin Morris wrote:
> Dear List Members,
>
> I have attached a pdf addressing the question as to why one might wish
> to have the option of x / 0 = 0 in Maxima.
>
> On the subject of Isabelle/Hol, members may wish to download the
> software from
> http://www.cl.cam.ac.uk/research/hvg/Isabelle/
> and familiarize themselves with the GUI, despite the steep learning 
> curve.
>
> Regards,
>
> Jalaluddin
>

As your paper says, the following theorem

a/b + c/d = (a ? d + b ? c)/(b ? d)

becomes false. That is quite a loss. My suggestion is that you change
your computation and not use "/" but some other notation, and then
define what you means by it through programs and perhaps tellsimp.

RJF