Re: [Maxima-discuss] Other number forms.

[Gunter K. <gu...@pe...>]

On 09.02.2016 18:35, Raymond Toy wrote:
>>>>>> "Michael" == Michael Soegtrop <Soegtrop> writes:
> 
>     Michael> Dear Stavros,
> 
>     Michael> I think the problem is that when numerators and
>     Michael> denominators are taken apart, e.g. all coefficients of a
>     Michael> polynom are brought to a common denominator and this is
>     Michael> factored out, information about the original scaling of
>     Michael> numbers is lost. Imagine what happens when this is done
>     Michael> to a rational function: all the coefficients are large
>     Michael> integers, and the two common denominators end up as
>     Michael> factored out fraction. In this situation it is almost
>     Michael> impossible to recover the original order of magnitude of
>     Michael> the coefficients. You end up with very large coefficients
>     Michael> in the numerator and denominator polynom. Of cause one
>     Michael> could make the coefficient of the largest power of
>     Michael> numerator and denominator 1, but this is not always
>     Michael> desirable in physical applications. The original numbers
> 
> Wouldn't that be true even if maxima supported rationals the way you
> want?  Maxima doesn't understand what is physically desirable so the
> coefficients would get mangled in some way unless you, as the user,
> impose the constraints.
In many of my calculations I know that I want a few µV, Volts,
Kilivolts, Amperes, Kiloamperes or µA. And normally I want 1-3 digits to
be in front of the comma. But telling this from

2000 V*(20 Ohms)/(2 Ohms)

would be hard for any heuristics.

Also atan(0.01*10^3A/5*!0^3*A) would not really tell what I am up to...

but after a

load("engineering-format");

float(%);

brings most results into a meaningful format: 1-3 digits in front of a
comma and  an exponent dividable by 3.

Are you sure that a auto detection for the number format would create
easier-to-interpret numbers?

Kind regards,

   Gunter.