[grjmell]

I have following expression in my funkSetFieldsDict:

expressions
(
    setInternalField
    {
        field Ub;
        expression  "pos().z >= 0.1 ? vector(log(pos().z-0.08),0,0) : vector(0,0,0) ";
        keepPatches true;
    }
);

I always get a floating point error, unless i remove the "-0.08" bit. due to the condition to only evaluate that equation if pos().z >= 0.1 there should be no log of zero... so i don't understand why i always get that error?? in my mesh pos().z goes from 0 to 0.5.

[gschaider]

Originally Posted by grjmell

I have following expression in my funkSetFieldsDict:

expressions
(
    setInternalField
    {
        field Ub;
        expression  "pos().z >= 0.1 ? vector(log(pos().z-0.08),0,0) : vector(0,0,0) ";
        keepPatches true;
    }
);

I always get a floating point error, unless i remove the "-0.08" bit. due to the condition to only evaluate that equation if pos().z >= 0.1 there should be no log of zero... so i don't understand why i always get that error?? in my mesh pos().z goes from 0 to 0.5.

The problem is that for "a ? b : c" FSF first evaluates b and c and then picks from them according to a. Avoiding this would mean reprogramming almost all field operations.

The best solution I can think of is "vector(log(pos().z >= 0.1 ? pos().z-0.08 : 1),0,0)" (untested - hope there is no syntax error)

[grjmell]

Thanks for the explanation. It is a clever work around!