Current sensor statistics: to polar to decomposed statistics #1150
Conversation
Signed-off-by: Martijn Govers <[email protected]>
| // the following terms break the symmetry between | ||
| // - polar -> decomposed -> uniform | ||
| // - polar -> independent -> uniform | ||
| // note that that is OK because decomposed is a 2D transformation of polar, while | ||
| // independent is a radial-only transformation of polar, so it is more stable. |
There was a problem hiding this comment.
This also means that because we use DecomposedComplexRandomVariable in our CurrentSensorCalcParams, our IterativeLinear solver actually uses a slightly larger variance than strictly necessary (because it uses UniformComplexRandomVariable, not DecomposedComplexRandomVariable).
This can be resolved by using PolarComplexRandomVariable in our CurrentSensorCalcParams and deferring the conversion either to DecomposedComplexRandomVariable (NRSE) or directly to UniformComplexRandomVariable (ILSE) where needed. MeasuredValues may then require some updates as well, as it would require different accumulations for the ILSE and the NRSE.
Out of scope of this PR, though.
Signed-off-by: Martijn Govers <[email protected]>
|
| // // first order approximation | ||
| // return std::make_pair(magnitude_variance * cos2_angle + magnitude2 * angle_variance * sin2_angle, | ||
| // magnitude_variance * sin2_angle + magnitude2 * angle_variance * cos2_angle); |
There was a problem hiding this comment.
do we want to remove this?
1 participant
Fixes #1068