|
Monado OpenXR Runtime
|
How well one kind of sensor was fit, so that a bad solve can be attributed to the sensor that could not be explained rather than guessed at from the parameter values. More...
#include </builds/monado/monado/src/xrt/tracking/constellation/optimizer/math.hpp>
Public Member Functions | |
| double | meanChiSquared () const |
| Chi squared per residual row. Around 1 when the data matches the model and its noise model. | |
| double | rmsSigma () const |
| The average size of a residual, in standard deviations. | |
Data Fields | |
| size_t | num_residuals |
| How many residual rows this kind of factor contributes. | |
| double | chi_squared |
Sum of the squared whitened residuals, 2 * cost. | |
How well one kind of sensor was fit, so that a bad solve can be attributed to the sensor that could not be explained rather than guessed at from the parameter values.
Both factors are whitened, so their residuals are in standard deviations and these numbers are comparable to each other and to 1. A mean_chi_squared far above 1 means the model cannot explain that sensor's data.
|
inline |
Chi squared per residual row. Around 1 when the data matches the model and its noise model.
References num_residuals.
Referenced by xrt::tracking::constellation::optimizer::rig_calibration::RigCalibration::optimize(), and rmsSigma().
|
inline |
The average size of a residual, in standard deviations.
References meanChiSquared().
Referenced by xrt::tracking::constellation::optimizer::rig_calibration::RigCalibration::optimize().
| double xrt::tracking::constellation::optimizer::FactorFitness::chi_squared |
Sum of the squared whitened residuals, 2 * cost.
Referenced by xrt::tracking::constellation::optimizer::rig_calibration::RigCalibration::optimize().
| size_t xrt::tracking::constellation::optimizer::FactorFitness::num_residuals |
How many residual rows this kind of factor contributes.
Referenced by xrt::tracking::constellation::optimizer::evaluateFactorFitness(), meanChiSquared(), and xrt::tracking::constellation::optimizer::rig_calibration::RigCalibration::optimize().