ExpressionSolve cant't solve cast

[Willi Braun]

The model Modelica.Electrical.Machines.Examples.SynchronousInductionMachines.SMEE_LoadDump fails, because we can't solve for for a cast expression.
See following model:

loadString("
model Test2
  parameter Real a = sin(3.14)*2;
  parameter Real x = 10;
  Real y;
  Integer p;
equation
  der(y) = y + p;
  der(y) = a*sin(time);
end Test2;
");
getErrorString();

• It should be possible to fix that, or vitalij?

• The tearing module should generate an error in such situations.

[Martin Sjölund]

It is an error to solve for p in any of the given equations. You need to have only discrete-time expressions in an equation if you want to solve for a discrete-time variable.

[Lennart Ochel]

The model is invalid. I will introduce better error messages when I split the simulation in continuous and discrete parts.

[Willi Braun]

I'm really not sure that this model is invalid, since let's take that model:

model A
  Real y;
  Integer p;
equation
  //der(y) = y + p;
   p = integer(y - der(y));
  der(y) = y;
end A;

This one is clear valid it contains just a fix to Martins argument with the discrete-time expressions. So why can't we do that implicit in the compiler, since it's an equation and somehow
the kind of writing shouldn't change the semantic:

der(y) = y + p;
<=>
p = y - der(y);

Okay, that perhaps a discussion for the Modelica Trac.

But the reason I reopen that the ticket again, is the following model and that has the same issue as the model above.

• In the initialization are all variables discrete-time, so that equation is valid now.
• We need to detect such situation while the tearing and that's quite easy.
• Also we could solve it that also for p and put the integer operator around the rhs.

model B
  Real y;
  parameter Integer p(fixed = false);
initial equation 
  der(y) = y + p;
equation 
  der(y) = sin(time);
end B;

[Lennart Ochel]

What is this ticket about? I cannot see your point.

BTW: Introducing the integer-operator changes the equation fundamentally. That is not just a simple transformation. Hence, that cannot be done implicitly by the compiler.

[Willi Braun]

Replying to lochel:

What is this ticket about? I cannot see your point.

Please check model B, which has the same issue as
Modelica.Electrical.Machines.Examples.SynchronousInductionMachines.SMEE_LoadDump
and which - in my opinion - is valid.

[Adrian Pop]

Seems that the SMEE_LoadDump issue was solved by ​m:r8044. See also: ​m:#1570.

[Willi Braun]

Replying to lochel:

What is this ticket about? I cannot see your point.

However, I changed the model B a bit and it produces the original issue now, that we all had
seen before the fix ​m:r8044 in Modelica.Electrical.Machines.Examples.SynchronousInductionMachines.SMEE_LoadDump.
It's a C compilation error which should catched up by the compiler with a reasonable error message, that's this ticket about.