A weighting is applied to the RMSD in the RootMeanSquareDeviation job. Looking at the code the weight technically does not follow the procedure described in https://mdanse.readthedocs.io/en/protos/pages/parameters.html#weights. However the current way it is used seems to make some sense.
It seems like when you used something other then equal you are doing some sort of "root weighted mean square deviation"
Suggested fix
Update the documentation and make it clear what MDANSE is doing with RMSD and the weight setting. Or remove the setting so that it follows the documentation.
A weighting is applied to the RMSD in the RootMeanSquareDeviation job. Looking at the code the weight technically does not follow the procedure described in https://mdanse.readthedocs.io/en/protos/pages/parameters.html#weights. However the current way it is used seems to make some sense.
It seems like when you used something other then equal you are doing some sort of "root weighted mean square deviation"
Also the documentation describes the standard RMSD https://mdanse.readthedocs.io/en/protos/pages/structure.html#root-mean-square-deviation.
Suggested fix Update the documentation and make it clear what MDANSE is doing with RMSD and the weight setting. Or remove the setting so that it follows the documentation.