Hi Peter,
Then Christoph's behaviour makes sense. Because I recently fixed this to
actually follow the specification (since around r8921 OMC always
type-converts to Real arguments; code generation makes the calculation
efficient if the exponent is an real literal that can be converted into
an integer without loss of precision).
So these things work properly in the trunk, i.e.:
>>> 3+6*2^(-3)
3.75
--
Martin Sjölund
On 2011-05-30 11:41, Peter Aronsson wrote:
I just checked on our branch and there it is zero (i.e. in MathModelica it is
currently zero). This a type conversion feature/bug.
The expression 8 * 10^-5 is actually an integer expression, 8 * 1/10000, which
is then type converted at the top level to a Real, which gives 0.0;
This is somewhat bad in the specification. It reads: " Exponentiation "a^b" is defined as pow(double a,double b) in the ANSI C library
if both "a" and "b" are Real scalars. A Real scalar value is returned. If "a" or "b" are Integer scalars,
they are automatically promoted to "Real"."
However, Dymola (v7.4) gives the following:
a:=2^3
Declaring variable: Integer a ;
b:=2^(-3)
Declaring variable: Real b ;
That is, they do not follow the specification, instead they declare a Real if
the value is negative!
What if you want to do integer arithmetic's like 3 +6*2^(-3) ?
Dymola does:
c:=3 +6*2^(-3)
Declaring variable: Real c ;
Which is wrong. It should be 15/4
If you however write
c:=3 +16*2^(-3)
Declaring variable: Integer c ;
IMO this is really bad!
/Peter
-----Original Message-----
From: [email protected] [mailto:owner-
[email protected]] On Behalf Of Martin Sjölund
Sent: den 30 maj 2011 10:46
To: [email protected]
Cc: Christoph Höger
Subject: Re: Strange bug
Hi,
I don't see any problem with the code running in OpenModelica...
loadString("
model Test
Real x;
parameter Real p = 8 * 10^(-5); //becomes 0?
equation
x = time * p;
end Test;
");
simulate(Test);
val(x,0:0.1:1);
=>
{0.0,8e-06,1.6e-05,2.4e-05,3.2e-05,4e-05,4.8e-05,5.6e-05,6.4e-05,7.2e-05,8e-
05}
--
Martin Sjölund
On 2011-05-30 09:49, Christoph Höger wrote:
Hi,
what is wrong with that?
model Test
Real x;
parameter Real p = 8 * 10^(-5); //becomes 0?
equation
x = time * p;
end Test;
best,
Christoph
