exponentiation of fixed point or decimal

exponentiation of fixed point or decimal

[Matthias Teege]

Moin,

I have a question about exponentiation of fixed point typs. The
following code didnt compile because of the '**' operator is not
defined for fixed point typs. So I made a lot of type conversions from
my fixed point type to float but the result for an Var_B of 10000
differs 0.01 of what my desk calculator says.

I have also made test with the numerics packages. If I use reals there
will be a better result but my "application" is for financial
transactions and I think fixed point or decimal is the best choice.

Is there any way for an exponentiation of fixed types?
 
procedure test is

   Max_Delta  : constant := 0.001;
   Max_Digits : constant := 15;

   type MyDec is delta Max_Delta digits Max_Digits;

   package MyDec_IO is new Ada.Text_IO.Decimal_IO(MyDec);

   Var_A: MyDec := 5.00;
   Var_B: MyDec;

   function myfunc(K: MyDec; Z: MyDec; Count: Positive ) return MyDec is
   begin
      return K * (1.00 + Z / 100) ** Count;
   -- return MyDec(Float(K) * Float((1.00 + Z / 100)) ** Count);
   end myfunc;

Begin
   MyDec_IO.Get(Var_B);
   MyDec_IO.Put(myfunc(Var_B, 5.00, 5), EXP=>0, FORE=>2, AFT=>2);
end test;

Many thanks
Matthias

-- 
Matthias Teege -- matthias@mteege.de -- http://emugs.de
make world not war
PGP-Key auf Anfrage

Re: exponentiation of fixed point or decimal

[Robert Dewar]

In article <87hf46uah6.fsf@moon.mteege.de>,
  Matthias Teege <matthias@mteege.de> wrote:
>
> Moin,
>
> I have a question about exponentiation of fixed
point typs. The
> following code didnt compile because of the '**'
operator is not
> defined for fixed point typs. So I made a lot of
type conversions from
> my fixed point type to float


Why on earth did you chose type Float, is 6 digits
*really* enough accuracy for your "financial
application"?


Sent via Deja.com
http://www.deja.com/

Re: exponentiation of fixed point or decimal

[Mats Weber]

In my compoenent library
   http://lglwww.epfl.ch/Team/MW/mw_components.html
there is a package called exponentiation_functions that will do what you
want, using the multiplication defined for the fixed point type. You
will have to be careful about roundoff errors, fixed point types can be tricky.