template<typename indices_t, size_t kSize>
A measurement of a fixed-size subspace of the full parameters.
The measurement intentionally does not store a pointer/reference to the reference object in the geometry hierarchy, i.e. the surface or volume. The reference object can already be identified via the geometry identifier provided by the source link. Since a measurement must be anchored within the geometry hierarchy, all measurement surfaces and volumes must provide valid geometry identifiers. In all use-cases, e.g. Kalman filtering, a pointer/reference to the reference object is available before the measurement is accessed; e.g. the propagator provides the surface pointer during navigation, which is then used to lookup possible measurements.
The pointed-to geometry object would differ depending on the parameter type. This means either, that there needs to be an additional variable type or that a pointer to a base object is stored (requiring a
dynamic_castlater on). Both variants add additional complications. Since the geometry object is not required anyways (as discussed above), not storing it removes all these complications altogether.
- Template Parameters
indices_t – Parameter index type, determines the full parameter space
kSize – Size of the parameter subset.
using FullParametersVector = ActsVector<kFullSize>
Vector type containing all parameters in the same space.
Measurement() = delete
A measurement can only be constructed with valid parameters.
template<typename parameters_t, typename covariance_t>
inline Measurement(SourceLink source, const std::array<indices_t, kSize> &indices, const Eigen::MatrixBase<parameters_t> ¶ms, const Eigen::MatrixBase<covariance_t> &cov)
Construct from source link, subset indices, and measured data.
The indices must be ordered and must describe/match the content of parameters and covariance.
- Template Parameters
parameters_t – Input parameters vector type
covariance_t – Input covariance matrix type
source – The link that connects to the underlying detector readout
indices – Which parameters are measured
params – Measured parameters values
cov – Measured parameters covariance
~Measurement() = default
inline ExpansionMatrix expander() const
Expansion matrix from the measured subspace into the full space.
This is equivalent to the transpose of the projection matrix only in the case of a trivial projection matrix. While this is the case here, it is still recommended to use the expansion matrix directly in cases where it is explicitly used.
inline std::ostream &operator<<(std::ostream &os) const
inline ProjectionMatrix projector() const
Projection matrix from the full space into the measured subspace.
inline ParametersVector residuals(const FullParametersVector &reference) const
Compute residuals in the measured subspace.
This computes the difference
measured - referencetaking into account the allowed parameter ranges. Only the reference values in the measured subspace are used for the computation.
reference – Reference parameters in the full space.
Public Static Functions
static inline constexpr size_t size()
Number of measured parameters.