The distributions are implicitly truncated in the invalid region, and excluded from both the numerator and denominator of the weighted polydispersity integral ∫ w(p) F²(q;p) dp / ∫ w(p) dp.
Polydispersity is discussed in the SasView manual here and here but neither place mentions truncated distributions.
The existence of "valid" is mentioned briefly in the plugin help here but not the explanation.
Further note: parameter limits in the model definition also truncate the distribution, so if radius is set to [0, inf] then you do not need a test for valid = "radius >= 0". [This claim needs to be verified: make sure we can't get negative radii into a model after we remove the valid = ... statement but leave the fit range at (0, inf)].
Originally posted by @pkienzle in #704 (comment)
The distributions are implicitly truncated in the invalid region, and excluded from both the numerator and denominator of the weighted polydispersity integral ∫ w(p) F²(q;p) dp / ∫ w(p) dp.
Polydispersity is discussed in the SasView manual here and here but neither place mentions truncated distributions.
The existence of "valid" is mentioned briefly in the plugin help here but not the explanation.
Further note: parameter limits in the model definition also truncate the distribution, so if radius is set to [0, inf] then you do not need a test for valid = "radius >= 0". [This claim needs to be verified: make sure we can't get negative radii into a model after we remove the
valid = ...statement but leave the fit range at (0, inf)].Originally posted by @pkienzle in #704 (comment)