Adjust limits for Guinier-Porod model.

[pkienzle]

Limit s to the range [0, 3].

Fix limiting values for I(q) when Rg or Porod exponent m go to zero. Add limiting value for s=3.

[gonzalezma]

All clear.
Perhaps remove unneeded parenthesis around exp(-ms) * (n*ms/rg**2)**ms?

[pkienzle]

Perhaps remove unneeded parenthesis around exp(-ms) * (n*ms/rg**2)**ms?

The parenthesis are there to force python to evaluate the expression as vector * (scalar * scalar) rather than (vector * scalar) * scalar. The second form costs twice as much.

[gonzalezma]

Thanks. I missed that.

[krzywon]

@yunliu01 and @smk78 - A discussion from todays meeting asked if these limits will effect any science cases. The model itself says any value outside the range of 0-3 is unstable, but that doesn't mean there aren't any science cases where that isn't true. Please check when you are able.

[smk78]

I will be honest: I have very little experience of this model.

But I note the text under Eq 4 in the Hammouda paper. If s were >3 then the 'dimensionality parameter' would be negative and that instinctively seems bad?

I also note @mdoucet imposed an upper limit of 3 when he wrote the GP fit function for Mantid (https://github.com/mantidproject/mantid/blob/400d30c758e6bd9f5441c1e3e5287f6c9164785e/Framework/PythonInterface/plugins/functions/GuinierPorod.py).

So on this basis I see no reason not to approve.

[smk78]

Have not reviewed the code, but approve the limits imposed on parameter s.