>From: Davide Rocchesso <rocchesso@sci.univr.it>
Hmm.. I send this mail to saol-dev only and removed 4-5 personal addresses
from the header. Hope everybody are still on board.
>nonlinear map coupled with a dynamic linear system:
>y = f(x);
>x = dyn(y);
>In the saol semantics, there is always a unit delay between the two x's.
>Am I right? This might be source of instabilities and inaccuracy.
>I should be thinking about these issues in the near future, so I will be
>happy to discuss with you if there's interest.
I have understood that in the flow system, the delays are natural but
if we want turn the SAOL to physical simulator, then it is ok to me.
What if we want both systems to SAOL? Then I think we need something
more than
y = f(x);
x = dyn(y);
to make the two systems distinct.
How about
y(0) = f(x(0));
x(0) = dyn(y(0));
where 0 tells the time of the used sample.
The zero has nothing to do with how many samples are gone through
the obcode. It is only for making minor time differences clear.
For current SAOL system, the code would be
y(0) = f(x(0));
x(1) = dyn(y(0));
or
y(0) = f(x(-1));
x(0) = dyn(y(0));
or such. This could be default interpretation.
The previous example
w = y + z;
x = x + y;
x = x + z;
would be written as
w(0) = y(0) + z(0);
x(0) = x(-1) + y(0);
x(0) = x(0) + z(0);
or such.
What new comes with the new system: at least the physical accuracy.
Even it is now difficult to come up with a working system, we don't
have to change the language later.
A FIR filter could be made with
y = x(-1)*a + x(0)*b + x(1)*c;
but I don't know if that is already possible. I'm not able to verify
it from SAOL specs now.
Yours,
Juhana
This archive was generated by hypermail 2b29 : Wed May 10 2000 - 12:14:10 EDT