25 #include <queso/LogNormalJointPdf.h>
26 #include <queso/GslVector.h>
27 #include <queso/GslMatrix.h>
32 template<
class V,
class M>
36 const V& lawExpVector,
37 const V& lawVarVector)
39 BaseJointPdf<V,M>(((std::string)(prefix)+
"gau").c_str(),domainSet),
40 m_lawExpVector (new V(lawExpVector)),
41 m_lawVarVector (new V(lawVarVector)),
42 m_diagonalCovMatrix(true)
65 template<
class V,
class M>
68 delete m_lawVarVector;
69 delete m_lawExpVector;
72 template <
class V,
class M>
76 return *m_lawExpVector;
79 template <
class V,
class M>
83 return *m_lawVarVector;
86 template<
class V,
class M>
89 const V& domainVector,
90 const V* domainDirection,
93 V* hessianEffect)
const
95 if ((m_env.subDisplayFile()) && (m_env.displayVerbosity() >= 55)) {
96 *m_env.subDisplayFile() <<
"Entering LogNormalJointPdf<V,M>::actualValue()"
97 <<
", meanVector = " << *m_lawExpVector
98 <<
": domainVector = " << domainVector
99 <<
", domainVector.sizeLocal() = " << domainVector.sizeLocal()
100 <<
", this->m_domainSet.vectorSpace().dimLocal() = " << this->m_domainSet.vectorSpace().dimLocal()
106 queso_require_msg(!(hessianMatrix || hessianEffect),
"incomplete code for gradVector, hessianMatrix and hessianEffect calculations");
108 double returnValue = 0.;
110 V zeroVector(domainVector);
111 zeroVector.cwSet(0.);
112 if (domainVector.atLeastOneComponentSmallerOrEqualThan(zeroVector)) {
116 else if (this->m_domainSet.contains(domainVector) ==
false) {
122 returnValue = std::exp(this->lnValue(domainVector,domainDirection,gradVector,hessianMatrix,hessianEffect));
125 (*gradVector) *= returnValue;
129 if ((m_env.subDisplayFile()) && (m_env.displayVerbosity() >= 55)) {
130 *m_env.subDisplayFile() <<
"Leaving LogNormalJointPdf<V,M>::actualValue()"
131 <<
", meanVector = " << *m_lawExpVector
132 <<
": domainVector = " << domainVector
133 <<
", returnValue = " << returnValue
140 template<
class V,
class M>
143 const V& domainVector,
144 const V* domainDirection,
147 V* hessianEffect)
const
149 if ((m_env.subDisplayFile()) && (m_env.displayVerbosity() >= 55)) {
150 *m_env.subDisplayFile() <<
"Entering LogNormalJointPdf<V,M>::lnValue()"
151 <<
", meanVector = " << *m_lawExpVector
152 <<
": domainVector = " << domainVector
156 queso_require_msg(!(domainDirection || hessianMatrix || hessianEffect),
"incomplete code for gradVector, hessianMatrix and hessianEffect calculations");
158 double returnValue = 0.;
160 V zeroVector(domainVector);
161 zeroVector.cwSet(0.);
162 if (domainVector.atLeastOneComponentSmallerOrEqualThan(zeroVector)) {
164 returnValue = -INFINITY;
166 else if (this->m_domainSet.contains(domainVector) ==
false) {
168 returnValue = -INFINITY;
171 if (m_diagonalCovMatrix) {
172 V diffVec(zeroVector);
173 for (
unsigned int i = 0; i < domainVector.sizeLocal(); ++i) {
174 diffVec[i] = std::log(domainVector[i]) - this->lawExpVector()[i];
182 (*gradVector)[i] = -(1.0 / domainVector[i]) -
183 diffVec[i] / (domainVector[i] * this->lawVarVector()[i]);
186 returnValue = ((diffVec*diffVec)/this->lawVarVector()).sumOfComponents();
189 for (
unsigned int i = 0; i < domainVector.sizeLocal(); ++i) {
190 returnValue -= std::log(domainVector[i]);
191 if (m_normalizationStyle == 0) {
192 returnValue -= std::log(std::sqrt(2. * M_PI * this->lawVarVector()[i]));
197 queso_error_msg(
"situation with a non-diagonal covariance matrix makes no sense");
199 returnValue += m_logOfNormalizationFactor;
202 if ((m_env.subDisplayFile()) && (m_env.displayVerbosity() >= 55)) {
203 *m_env.subDisplayFile() <<
"Leaving LogNormalJointPdf<V,M>::lnValue()"
204 <<
", meanVector = " << *m_lawExpVector
205 <<
": domainVector = " << domainVector
206 <<
", returnValue = " << returnValue
213 template<
class V,
class M>
220 if (m_diagonalCovMatrix) {
221 unsigned int n_params = meanVector.sizeLocal();
222 queso_assert_equal_to (n_params, this->lawExpVector().sizeLocal());
224 for (
unsigned int i = 0; i < n_params; ++i) {
225 meanVector[i] = std::exp(this->lawExpVector()[i] + 0.5*this->lawVarVector()[i]);
229 queso_error_msg(
"situation with a non-diagonal covariance matrix makes no sense");
233 template<
class V,
class M>
240 if (m_diagonalCovMatrix) {
241 unsigned int n_params = this->lawExpVector().sizeLocal();
242 queso_assert_equal_to (n_params, this->lawVarVector().sizeLocal());
243 queso_assert_equal_to (n_params, covMatrix.numCols());
244 queso_assert_equal_to (covMatrix.numCols(), covMatrix.numRowsGlobal());
246 covMatrix.zeroLower();
247 covMatrix.zeroUpper();
249 for (
unsigned int i = 0; i < n_params; ++i) {
250 covMatrix(i,i) = (std::exp(this->lawVarVector()[i]) - 1) *
251 std::exp(2*this->lawExpVector()[i] + this->lawVarVector()[i]);
255 queso_error_msg(
"situation with a non-diagonal covariance matrix makes no sense");
259 template<
class V,
class M>
265 if ((m_env.subDisplayFile()) && (m_env.displayVerbosity() >= 2)) {
266 *m_env.subDisplayFile() <<
"Entering LogNormalJointPdf<V,M>::computeLogOfNormalizationFactor()"
270 if ((m_env.subDisplayFile()) && (m_env.displayVerbosity() >= 2)) {
271 *m_env.subDisplayFile() <<
"Leaving LogNormalJointPdf<V,M>::computeLogOfNormalizationFactor()"
272 <<
", m_logOfNormalizationFactor = " << m_logOfNormalizationFactor
double computeLogOfNormalizationFactor(unsigned int numSamples, bool updateFactorInternally) const
Computes the logarithm of the normalization factor.
virtual void distributionMean(V &meanVector) const
Mean value of the underlying random variable.
LogNormalJointPdf(const char *prefix, const VectorSet< V, M > &domainSet, const V &lawExpVector, const V &lawVarVector)
Constructor.
A templated class for handling sets.
~LogNormalJointPdf()
Destructor.
double commonComputeLogOfNormalizationFactor(unsigned int numSamples, bool updateFactorInternally) const
Common method (to the derived classes) to compute the logarithm of the normalization factor...
const BaseEnvironment & m_env
virtual void distributionVariance(M &covMatrix) const
Covariance matrix of the underlying random variable.
A templated (base) class for handling joint PDFs.
double actualValue(const V &domainVector, const V *domainDirection, V *gradVector, M *hessianMatrix, V *hessianEffect) const
Actual value of the Log-Normal PDF (scalar function).
A class for handling Log-Normal joint PDFs.
double lnValue(const V &domainVector, const V *domainDirection, V *gradVector, M *hessianMatrix, V *hessianEffect) const
Logarithm of the value of the Log-Normal PDF (scalar function).
MonteCarloSGOptions::MonteCarloSGOptions(const BaseEnvironment &env, const char *prefix, const McOptionsValues &alternativeOptionsValues queso_require_equal_to_msg)(m_env.optionsInputFileName(), std::string(""), std::string("this constructor is incompatible with the existence of an options input file"))
unsigned int displayVerbosity() const
const V & lawExpVector() const
Access to the vector of mean values and private attribute: m_lawExpVector.
const V & lawVarVector() const
Access to the vector of variance values and private attribute: m_lawVarVector.
std::ofstream * subDisplayFile() const
Access function for m_subDisplayFile (displays file on stream).