Belle II Software development
CDCRobustSZFitter.cc
1/**************************************************************************
2 * basf2 (Belle II Analysis Software Framework) *
3 * Author: The Belle II Collaboration *
4 * *
5 * See git log for contributors and copyright holders. *
6 * This file is licensed under LGPL-3.0, see LICENSE.md. *
7 **************************************************************************/
8#include <tracking/trackFindingCDC/fitting/CDCRobustSZFitter.h>
9
10#include <tracking/trackFindingCDC/fitting/CDCSZFitter.h>
11
12#include <tracking/trackFindingCDC/fitting/CDCSZObservations.h>
13
14#include <tracking/trackFindingCDC/eventdata/trajectories/CDCTrajectorySZ.h>
15
16#include <tracking/trackFindingCDC/numerics/Median.h>
17#include <tracking/trackFindingCDC/numerics/WithWeight.h>
18
19using namespace Belle2;
20using namespace TrackFindingCDC;
21
23{
24 // This seems to be some other algorithm
25
26
27 CDCTrajectorySZ trajectorySZ;
28 CDCSZFitter szFitter;
29
30 if (observationsSZ.size() > 4) {
31 CDCSZObservations observationsSZFiltered;
32
33 double meanTanLambda = 0;
34 double meanStartZ = 0;
35
36 for (unsigned int i = 0; i < observationsSZ.size(); i++) {
37 for (unsigned int j = 0; j < observationsSZ.size(); j++) {
38 if (i != j) {
39 observationsSZFiltered.fill(observationsSZ.getS(j),
40 observationsSZ.getZ(j),
41 observationsSZ.getWeight(j));
42 }
43 } // for j
44
45 szFitter.update(trajectorySZ, observationsSZFiltered);
46 meanTanLambda += trajectorySZ.getTanLambda();
47 meanStartZ += trajectorySZ.getZ0();
48 } // for i
49
50 return CDCTrajectorySZ(meanTanLambda / observationsSZ.size(),
51 meanStartZ / observationsSZ.size());
52 } else {
54 }
55}
56
58{
59 std::vector<double> tanLambdas;
60 tanLambdas.reserve(szObservations.size() * (szObservations.size() - 1) / 2);
61 for (unsigned int i = 0; i < szObservations.size(); i++) {
62 for (unsigned int j = i + 1; j < szObservations.size(); j++) {
63 double s1 = szObservations.getS(i);
64 double s2 = szObservations.getS(j);
65 if (s1 == s2) continue;
66 double z1 = szObservations.getZ(i);
67 double z2 = szObservations.getZ(j);
68 tanLambdas.push_back((z2 - z1) / (s2 - s1));
69 }
70 }
71
72 const double tanLambda = median(std::move(tanLambdas));
73 const double z0 = getMedianZ0(szObservations, tanLambda);
74
75 CDCTrajectorySZ trajectorySZ(tanLambda, z0);
76 trajectorySZ.setNDF(szObservations.size());
77 return trajectorySZ;
78}
79
81{
82 std::vector<WithWeight<double> > weightedTanLambdas;
83 Weight totalWeight = 0;
84 weightedTanLambdas.reserve(szObservations.size() * (szObservations.size() - 1) / 2);
85 for (unsigned int i = 0; i < szObservations.size(); i++) {
86 for (unsigned int j = i + 1; j < szObservations.size(); j++) {
87 double s1 = szObservations.getS(i);
88 double s2 = szObservations.getS(j);
89 if (s1 == s2) continue;
90 double z1 = szObservations.getZ(i);
91 double z2 = szObservations.getZ(j);
92
93 double w1 = szObservations.getWeight(i);
94 double w2 = szObservations.getWeight(j);
95
96 // Longer legs receive proportionally longer weight.
97 Weight weight = std::abs(s2 - s1) * hypot2(w1, w2);
98 weightedTanLambdas.emplace_back((z2 - z1) / (s2 - s1), weight);
99 totalWeight += weight;
100 }
101 }
102
103 for (WithWeight<double>& weightedTanLambda : weightedTanLambdas) {
104 weightedTanLambda.weight() /= totalWeight;
105 }
106
107 const double tanLambda = weightedMedian(std::move(weightedTanLambdas));
108 const double z0 = getMedianZ0(szObservations, tanLambda);
109
110 CDCTrajectorySZ trajectorySZ(tanLambda, z0);
111 trajectorySZ.setNDF(szObservations.size());
112 return trajectorySZ;
113}
114
115double CDCRobustSZFitter::getMedianZ0(const CDCSZObservations& szObservations, double tanLambda) const
116{
117 std::vector<double> z0s;
118 z0s.reserve(szObservations.size());
119 for (unsigned int i = 0; i < szObservations.size(); i++) {
120 double s = szObservations.getS(i);
121 double z = szObservations.getZ(i);
122 z0s.push_back(z - s * tanLambda);
123 }
124 const double z0 = median(std::move(z0s));
125 return z0;
126}
CDCTrajectorySZ fitWeightedTheilSen(const CDCSZObservations &szObservations) const
Implements the weighted Theil-Sen line fit algorithm.
CDCTrajectorySZ fitTheilSen(const CDCSZObservations &szObservations) const
Implements the original Theil-Sen line fit algorithm.
double getMedianZ0(const CDCSZObservations &szObservations, double tanLambda) const
Compute the median z0 intercept from the given observations and an estimated slope.
CDCTrajectorySZ fitUsingSimplifiedTheilSen(const CDCSZObservations &szObservations) const
Fit a linear sz trajectory to the reconstructed stereo segment.
Class implementing the z coordinate over travel distance line fit.
Definition: CDCSZFitter.h:27
void update(const CDCSegmentPair &segmentPair) const
Updates the trajectory of the axial stereo segment pair inplace.
Definition: CDCSZFitter.cc:163
Class serving as a storage of observed sz positions to present to the sz line fitters.
std::size_t fill(double s, double z, double weight=1.0)
Appends the observed position.
double getWeight(int iObservation) const
Getter for the weight / inverse variance of the observation at the given index.
double getZ(int iObservation) const
Getter for the z value of the observation at the given index.
std::size_t size() const
Returns the number of observations stored.
double getS(int iObservation) const
Getter for the arc length value of the observation at the given index.
Linear trajectory in sz space.
static CDCTrajectorySZ basicAssumption()
Constructs a basic assumption, what the z0 start position and the sz slope are, including some broad ...
double getTanLambda() const
Getter for the slope over the travel distance coordinate.
void setNDF(std::size_t ndf)
Setter for the number of degrees of freedom of the line fit.
double getZ0() const
Getter for the z coordinate at zero travel distance.
A mixin class to attach a weight to an object.
Definition: WithWeight.h:24
Abstract base class for different kinds of events.