make sure harmonic average particle interpolator does not average neg…

[jdannberg]

…ative values

This is based on a discussion with @anne-glerum this morning.

For all pull requests:

• I have followed the instructions for indenting my code.

[jdannberg]

Looks like the one test that is failing (particle_interpolator_harmonic_average_solkz) triggers this exact problem: It averages a compositional field that is negative, in this case, it is the a field that is used to compute the density which is initialized as -1 * sin(2z) * cos(3pi*x) in a 1x1 box, so it varies between plus and minus 1.

Any suggestions what to do with that? If this test is just for averaging, I could just add +1 to that whole function and it should be fine. Does that seem reasonable?

[gassmoeller]

I could just add +1 to that whole function and it should be fine. Does that seem reasonable?

Seems reasonable to me since this test is not meant to test the accuracy of SolKz. Just add a comment to the parameter file that explains why you changed it.

[jdannberg]

Thank you! I have adapted the test (and while the pressure error is meaningless now, this actually improved the velocity error quite a bit).

[gassmoeller]

The improved velocity error probably comes from the fact we no longer compute wrong averages for the negative values.
Anyway, thanks for the fix.