normal_law.h
1.47 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
#ifndef PROBABILITE_NORMAL_LAW_H
#define PROBABILITE_NORMAL_LAW_H
#include "two_parameter_law.h"
namespace Probability {
class NormalLaw : public TwoParameterLaw {
public:
NormalLaw();
I18n::Message title() override;
Type type() const override;
bool isContinuous() const override;
float xMin() override;
float yMin() override;
float xMax() override;
float yMax() override;
I18n::Message parameterNameAtIndex(int index) override;
I18n::Message parameterDefinitionAtIndex(int index) override;
float evaluateAtAbscissa(float x) const override;
bool authorizedValueAtIndex(float x, int index) const override;
void setParameterAtIndex(float f, int index) override;
double cumulativeDistributiveFunctionAtAbscissa(double x) const override;
double cumulativeDistributiveInverseForProbability(double * probability) override;
private:
constexpr static double k_maxRatioMuSigma = 1000.0f;
/* For the standard norma law, P(X < y) > 0.9999995 with y >= 4.892 so the
* value displayed is 1. But this is dependent on the fact that we display
* only 7 decimal values! */
static_assert(Constant::LargeNumberOfSignificantDigits == 7, "k_maxProbability is ill-defined compared to LargeNumberOfSignificantDigits");
constexpr static double k_boundStandardNormalDistribution = 4.892;
double standardNormalCumulativeDistributiveFunctionAtAbscissa(double abscissa) const;
double standardNormalCumulativeDistributiveInverseForProbability(double probability);
};
}
#endif