https://mooseframework.inl.gov
Public Types | Public Member Functions | Static Public Member Functions | Public Attributes | Static Public Attributes | Protected Member Functions | Protected Attributes | Private Attributes | List of all members
LMC Class Reference

Covariance function for multi-output Gaussian Processes based on the linear model of coregionalization (LMC) More...

#include <LMC.h>

Inheritance diagram for LMC:
[legend]

Public Types

using HyperParameterMap = std::unordered_map< std::string, torch::Tensor >
 
typedef DataFileName DataFileParameterType
 

Public Member Functions

 LMC (const InputParameters &parameters)
 
void computeCovarianceMatrix (torch::Tensor &K, const torch::Tensor &x, const torch::Tensor &xp, const bool is_self_covariance) const override
 Generates the Covariance Matrix given two sets of points in the parameter space. More...
 
bool computedKdhyper (torch::Tensor &dKdhp, const torch::Tensor &x, const std::string &hyper_param_name, unsigned int ind) const override
 Redirect dK/dhp for hyperparameter "hp". More...
 
void loadHyperParamMap (const HyperParameterMap &map)
 Load some hyperparameters into the local map contained in this object. More...
 
void buildHyperParamMap (HyperParameterMap &map) const
 Populates the input maps with the owned hyperparameters. More...
 
virtual bool getTuningData (const std::string &name, unsigned int &size, Real &min, Real &max) const
 Get the default minimum and maximum and size of a hyperparameter. More...
 
void dependentCovarianceTypes (std::map< UserObjectName, std::string > &name_type_map) const
 Populate a map with the names and types of the dependent covariance functions. More...
 
const std::vector< UserObjectName > & dependentCovarianceNames () const
 Get the names of the dependent covariances. More...
 
virtual bool isTunable (const std::string &name) const
 Check if a given parameter is tunable. More...
 
unsigned int numOutputs () const
 Return the number of outputs assumed for this covariance function. More...
 
virtual bool enabled () const
 
std::shared_ptr< MooseObject > getSharedPtr ()
 
std::shared_ptr< const MooseObject > getSharedPtr () const
 
bool isKokkosObject () const
 
MooseApp & getMooseApp () const
 
const std::string & type () const
 
const std::string & name () const
 
std::string typeAndName () const
 
MooseObjectParameterName uniqueParameterName (const std::string &parameter_name) const
 
MooseObjectName uniqueName () const
 
const InputParameters & parameters () const
 
const hit::Node * getHitNode () const
 
bool hasBase () const
 
const std::string & getBase () const
 
const T & getParam (const std::string &name) const
 
std::vector< std::pair< T1, T2 > > getParam (const std::string &param1, const std::string &param2) const
 
const T * queryParam (const std::string &name) const
 
const T & getRenamedParam (const std::string &old_name, const std::string &new_name) const
 
T getCheckedPointerParam (const std::string &name, const std::string &error_string="") const
 
bool isParamValid (const std::string &name) const
 
bool isParamSetByUser (const std::string &name) const
 
void connectControllableParams (const std::string &parameter, const std::string &object_type, const std::string &object_name, const std::string &object_parameter) const
 
void paramError (const std::string &param, Args... args) const
 
void paramWarning (const std::string &param, Args... args) const
 
void paramWarning (const std::string &param, Args... args) const
 
void paramInfo (const std::string &param, Args... args) const
 
std::string messagePrefix (const bool hit_prefix=true) const
 
std::string errorPrefix (const std::string &) const
 
void mooseError (Args &&... args) const
 
void mooseDocumentedError (const std::string &repo_name, const unsigned int issue_num, Args &&... args) const
 
void mooseErrorNonPrefixed (Args &&... args) const
 
void mooseWarning (Args &&... args) const
 
void mooseWarning (Args &&... args) const
 
void mooseWarningNonPrefixed (Args &&... args) const
 
void mooseWarningNonPrefixed (Args &&... args) const
 
void mooseDeprecated (Args &&... args) const
 
void mooseDeprecated (Args &&... args) const
 
void mooseDeprecatedNoTrace (Args &&... args) const
 
void mooseInfo (Args &&... args) const
 
void callMooseError (std::string msg, const bool with_prefix, const hit::Node *node=nullptr, const bool show_trace=true) const
 
std::string getDataFileName (const std::string &param) const
 
std::string getDataFileNameByName (const std::string &relative_path) const
 
std::string getDataFilePath (const std::string &relative_path) const
 
const Parallel::Communicator & comm () const
 
processor_id_type n_processors () const
 
processor_id_type processor_id () const
 

Static Public Member Functions

static InputParameters validParams ()
 
static bool isScalarHyperParameter (const torch::Tensor &tensor)
 Return true if a hyperparameter tensor stores one scalar value. More...
 
static bool isVectorHyperParameter (const torch::Tensor &tensor)
 Return true if a hyperparameter tensor stores a vector of values. More...
 
static void callMooseError (MooseApp *const app, const InputParameters &params, std::string msg, const bool with_prefix, const hit::Node *node, const bool show_trace=true)
 

Public Attributes

 usingCombinedWarningSolutionWarnings
 
const ConsoleStream _console
 

Static Public Attributes

static const std::string type_param
 
static const std::string name_param
 
static const std::string unique_name_param
 
static const std::string app_param
 
static const std::string moose_base_param
 
static const std::string kokkos_object_param
 

Protected Member Functions

void computeBMatrix (torch::Tensor &Bmat, const unsigned int exp_i) const
 Computes the covariance matrix for the outputs (using the latent coefficients) We use a $B = a_i a_i^T + diag(lambda_i)$ expansion here where $a_i$ and $lambda_i$ are vectors. More...
 
void computeAGradient (torch::Tensor &grad, const unsigned int exp_i, const unsigned int index) const
 Computes the gradient of $B$ with respect to the entries in $a_i$ in the following expression: $B = a_i a_i^T + diag(lambda_i)$. More...
 
void computeLambdaGradient (torch::Tensor &grad, const unsigned int exp_i, const unsigned int index) const
 Computes the gradient of $B$ with respect to the entries in $lambda_i$ in the following expression: $B = a_i a_i^T + diag(lambda_i)$. More...
 
torch::Tensor & addRealHyperParameter (const std::string &name, const Real value, const bool is_tunable)
 Register a scalar hyperparameter to this covariance function. More...
 
torch::Tensor & addVectorRealHyperParameter (const std::string &name, const std::vector< Real > &value, const bool is_tunable)
 Register a vector hyperparameter to this covariance function. More...
 
void flagInvalidSolutionInternal (const InvalidSolutionID invalid_solution_id) const
 
InvalidSolutionID registerInvalidSolutionInternal (const std::string &message, const bool warning) const
 
CovarianceFunctionBase * getCovarianceFunctionByName (const UserObjectName &name) const
 Lookup a CovarianceFunction object by name and return pointer. More...
 

Protected Attributes

const unsigned int _num_expansion_terms
 The number of expansion terms in the output ovariance matrix. More...
 
HyperParameterMap _hyperparameters
 Map of hyperparameters stored as rank-0 or rank-1 tensors. More...
 
std::unordered_set< std::string > _tunable_hp
 list of tunable hyper-parameters More...
 
const unsigned int _num_outputs
 The number of outputs this covariance function is used to describe. More...
 
const std::vector< UserObjectName > _dependent_covariance_names
 The names of the dependent covariance functions. More...
 
std::vector< std::string > _dependent_covariance_types
 The types of the dependent covariance functions. More...
 
std::vector< CovarianceFunctionBase * > _covariance_functions
 Vector of pointers to the dependent covariance functions. More...
 
const bool & _enabled
 
MooseApp & _app
 
Factory & _factory
 
ActionFactory & _action_factory
 
const std::string & _type
 
const std::string & _name
 
const InputParameters & _pars
 
const Parallel::Communicator & _communicator
 

Private Attributes

std::vector< const torch::Tensor * > _a_coeffs
 The vectors in the $B = a_i a_i^T + diag(lambda_i)$ expansion. More...
 
std::vector< const torch::Tensor * > _lambdas
 

Detailed Description

Covariance function for multi-output Gaussian Processes based on the linear model of coregionalization (LMC)

Definition at line 19 of file LMC.h.

Member Typedef Documentation

◆ HyperParameterMap

using CovarianceFunctionBase::HyperParameterMap = std::unordered_map<std::string, torch::Tensor>
inherited

Definition at line 23 of file CovarianceFunctionBase.h.

Constructor & Destructor Documentation

◆ LMC()

LMC::LMC ( const InputParameters &  parameters)

Definition at line 30 of file LMC.C.

31  : CovarianceFunctionBase(parameters),
32  _num_expansion_terms(getParam<unsigned int>("num_latent_funcs"))
33 {
34  // We use a random number generator to obtain the initial guess for the
35  // hyperparams
36  MooseRandom generator_latent;
37  generator_latent.seed(0, 1980);
38 
39  // First add and initialize the a A coefficients in the (aa^T+lambda*I) matrix
40  for (const auto exp_i : make_range(_num_expansion_terms))
41  {
42  const std::string a_coeff_name = "acoeff_" + std::to_string(exp_i);
43  std::vector<Real> acoeff_values(_num_outputs);
44  for (const auto out_i : make_range(_num_outputs))
45  acoeff_values[out_i] = 3.0 * generator_latent.rand(0) + 1.0;
46  auto & acoeff_vector = addVectorRealHyperParameter(a_coeff_name, acoeff_values, true);
47  _a_coeffs.push_back(&acoeff_vector);
48  }
49 
50  // Then add and initialize the lambda coefficients in the (aa^T+lambda*I) matrix
51  for (const auto exp_i : make_range(_num_expansion_terms))
52  {
53  const std::string lambda_name = "lambda_" + std::to_string(exp_i);
54  std::vector<Real> lambda_values(_num_outputs);
55  for (const auto out_i : make_range(_num_outputs))
56  lambda_values[out_i] = 3.0 * generator_latent.rand(0) + 1.0;
57  auto & lambda_vector = addVectorRealHyperParameter(lambda_name, lambda_values, true);
58  _lambdas.push_back(&lambda_vector);
59  }
60 }
LMC::_num_expansion_terms
const unsigned int _num_expansion_terms
The number of expansion terms in the output ovariance matrix.
Definition: LMC.h:67
MooseObject::parameters
const InputParameters & parameters() const
MooseRandom::seed
void seed(std::size_t i, unsigned int seed)
CovarianceFunctionBase::CovarianceFunctionBase
CovarianceFunctionBase(const InputParameters &parameters)
Definition: CovarianceFunctionBase.C:69
LMC::_lambdas
std::vector< const torch::Tensor * > _lambdas
Definition: LMC.h:73
CovarianceFunctionBase::_num_outputs
const unsigned int _num_outputs
The number of outputs this covariance function is used to describe.
Definition: CovarianceFunctionBase.h:115
make_range
IntRange< T > make_range(T beg, T end)
CovarianceFunctionBase::addVectorRealHyperParameter
torch::Tensor & addVectorRealHyperParameter(const std::string &name, const std::vector< Real > &value, const bool is_tunable)
Register a vector hyperparameter to this covariance function.
Definition: CovarianceFunctionBase.C:105
MooseRandom::rand
Real rand(std::size_t i)
LMC::_a_coeffs
std::vector< const torch::Tensor * > _a_coeffs
The vectors in the $B = a_i a_i^T + diag(lambda_i)$ expansion.
Definition: LMC.h:72

Member Function Documentation

◆ addRealHyperParameter()

torch::Tensor & CovarianceFunctionBase::addRealHyperParameter ( const std::string &  name,
const Real  value,
const bool  is_tunable 
)
protectedinherited

Register a scalar hyperparameter to this covariance function.

Parameters
nameThe name of the parameter
valueThe initial value of the parameter
is_tunableIf the parameter is tunable during optimization

Definition at line 95 of file CovarianceFunctionBase.C.

98 {
99  const auto prefixed_name = _name + ":" + name;
100  return insertHyperParameter(
101  _hyperparameters, _tunable_hp, prefixed_name, makeScalarHyperParameter(value), is_tunable);
102 }
CovarianceFunctionBase::_hyperparameters
HyperParameterMap _hyperparameters
Map of hyperparameters stored as rank-0 or rank-1 tensors.
Definition: CovarianceFunctionBase.h:109
CovarianceFunctionBase::_tunable_hp
std::unordered_set< std::string > _tunable_hp
list of tunable hyper-parameters
Definition: CovarianceFunctionBase.h:112
MooseObject::_name
const std::string & _name
MooseObject::name
const std::string & name() const

◆ addVectorRealHyperParameter()

torch::Tensor & CovarianceFunctionBase::addVectorRealHyperParameter ( const std::string &  name,
const std::vector< Real > &  value,
const bool  is_tunable 
)
protectedinherited

Register a vector hyperparameter to this covariance function.

Parameters
nameThe name of the parameter
valueThe initial value of the parameter
is_tunableIf the parameter is tunable during optimization

Definition at line 105 of file CovarianceFunctionBase.C.

Referenced by LMC().

108 {
109  const auto prefixed_name = _name + ":" + name;
110  return insertHyperParameter(
111  _hyperparameters, _tunable_hp, prefixed_name, makeVectorHyperParameter(value), is_tunable);
112 }
CovarianceFunctionBase::_hyperparameters
HyperParameterMap _hyperparameters
Map of hyperparameters stored as rank-0 or rank-1 tensors.
Definition: CovarianceFunctionBase.h:109
CovarianceFunctionBase::_tunable_hp
std::unordered_set< std::string > _tunable_hp
list of tunable hyper-parameters
Definition: CovarianceFunctionBase.h:112
MooseObject::_name
const std::string & _name
MooseObject::name
const std::string & name() const

◆ buildHyperParamMap()

void CovarianceFunctionBase::buildHyperParamMap ( HyperParameterMap &  map) const
inherited

Populates the input maps with the owned hyperparameters.

Parameters
mapMap of hyperparameters that should be populated

Definition at line 153 of file CovarianceFunctionBase.C.

Referenced by StochasticTools::GaussianProcess::setupCovarianceMatrix().

154 {
155  // First, add the hyperparameters of the dependent covariance functions
156  for (const auto dependent_covar : _covariance_functions)
157  dependent_covar->buildHyperParamMap(map);
158 
159  // At the end we just append the hyperparameters this object owns
160  for (const auto & iter : _hyperparameters)
161  if (!isScalarHyperParameter(iter.second) && !isVectorHyperParameter(iter.second))
162  mooseError("Unsupported hyperparameter rank ", iter.second.dim(), " for ", iter.first, ".");
163  else
164  map[iter.first] = iter.second.clone();
165 }
CovarianceFunctionBase::_hyperparameters
HyperParameterMap _hyperparameters
Map of hyperparameters stored as rank-0 or rank-1 tensors.
Definition: CovarianceFunctionBase.h:109
CovarianceFunctionBase::isVectorHyperParameter
static bool isVectorHyperParameter(const torch::Tensor &tensor)
Return true if a hyperparameter tensor stores a vector of values.
Definition: CovarianceFunctionBase.C:50
CovarianceFunctionBase::isScalarHyperParameter
static bool isScalarHyperParameter(const torch::Tensor &tensor)
Return true if a hyperparameter tensor stores one scalar value.
Definition: CovarianceFunctionBase.C:44
CovarianceFunctionBase::_covariance_functions
std::vector< CovarianceFunctionBase * > _covariance_functions
Vector of pointers to the dependent covariance functions.
Definition: CovarianceFunctionBase.h:124
MooseObject::mooseError
void mooseError(Args &&... args) const

◆ computeAGradient()

void LMC::computeAGradient ( torch::Tensor &  grad,
const unsigned int  exp_i,
const unsigned int  index 
) const
protected

Computes the gradient of $B$ with respect to the entries in $a_i$ in the following expression: $B = a_i a_i^T + diag(lambda_i)$.

Parameters
gradThe gradient matrix that should be populated
exp_iThe index of the expansion of B
indexThe index within the vector $a_i$

Definition at line 168 of file LMC.C.

Referenced by computedKdhyper().

171 {
172  const auto & a_coeffs = *_a_coeffs[exp_i];
173  mooseAssert(cast_int<int64_t>(index) < a_coeffs.numel(), "Incorrect LMC coefficient index.");
174  auto basis = torch::zeros_like(a_coeffs);
175  const auto index_tensor =
176  torch::tensor({cast_int<int64_t>(index)},
177  torch::TensorOptions().dtype(torch::kLong).device(a_coeffs.device()));
178  basis.index_fill_(0, index_tensor, 1.0);
179  grad = torch::outer(basis, a_coeffs) + torch::outer(a_coeffs, basis);
180 }
NS::grad
std::string grad(const std::string &var)
Definition: NS.h:92
LMC::_a_coeffs
std::vector< const torch::Tensor * > _a_coeffs
The vectors in the $B = a_i a_i^T + diag(lambda_i)$ expansion.
Definition: LMC.h:72

◆ computeBMatrix()

void LMC::computeBMatrix ( torch::Tensor &  Bmat,
const unsigned int  exp_i 
) const
protected

Computes the covariance matrix for the outputs (using the latent coefficients) We use a $B = a_i a_i^T + diag(lambda_i)$ expansion here where $a_i$ and $lambda_i$ are vectors.

Parameters
BmatThe matrix which should be populated by the covariance values
exp_iThe expansion index in the latent space

Definition at line 160 of file LMC.C.

Referenced by computeCovarianceMatrix(), and computedKdhyper().

161 {
162  const auto & a_coeffs = *_a_coeffs[exp_i];
163  const auto & lambda_coeffs = *_lambdas[exp_i];
164  Bmat = torch::outer(a_coeffs, a_coeffs) + torch::diag(lambda_coeffs);
165 }
LMC::_lambdas
std::vector< const torch::Tensor * > _lambdas
Definition: LMC.h:73
LMC::_a_coeffs
std::vector< const torch::Tensor * > _a_coeffs
The vectors in the $B = a_i a_i^T + diag(lambda_i)$ expansion.
Definition: LMC.h:72

◆ computeCovarianceMatrix()

void LMC::computeCovarianceMatrix ( torch::Tensor &  K,
const torch::Tensor &  x,
const torch::Tensor &  xp,
const bool  is_self_covariance 
) const
overridevirtual

Generates the Covariance Matrix given two sets of points in the parameter space.

Parameters
KReference to a matrix which should be populated by the covariance entries
xReference to the first set of points
xpReference to the second set of points
is_self_covarianceSwitch to enable adding the noise variance to the diagonal of the covariance matrix

Implements CovarianceFunctionBase.

Definition at line 63 of file LMC.C.

67 {
68  const auto options = x.options().dtype(at::kDouble);
69  // Create temporary vectors for constructing the covariance matrix
70  torch::Tensor K_params = torch::zeros({x.sizes()[0], xp.sizes()[0]}, options);
71  torch::Tensor B = torch::zeros({_num_outputs, _num_outputs}, options);
72  K = torch::zeros({x.sizes()[0] * _num_outputs, xp.sizes()[0] * _num_outputs}, options);
73  torch::Tensor K_working;
74 
75  // For every expansion term we add the contribution to the covariance matrix
76  for (const auto exp_i : make_range(_num_expansion_terms))
77  {
78  _covariance_functions[exp_i]->computeCovarianceMatrix(K_params, x, xp, is_self_covariance);
79  computeBMatrix(B, exp_i);
80  K_working = torch::kron(B, K_params);
81  K += K_working;
82  }
83 }
LMC::_num_expansion_terms
const unsigned int _num_expansion_terms
The number of expansion terms in the output ovariance matrix.
Definition: LMC.h:67
NS::K
static const std::string K
Definition: NS.h:174
CovarianceFunctionBase::_covariance_functions
std::vector< CovarianceFunctionBase * > _covariance_functions
Vector of pointers to the dependent covariance functions.
Definition: CovarianceFunctionBase.h:124
CovarianceFunctionBase::_num_outputs
const unsigned int _num_outputs
The number of outputs this covariance function is used to describe.
Definition: CovarianceFunctionBase.h:115
x
const std::vector< double > x
Definition: EquilibriumConstantFitTest.C:17
LMC::computeBMatrix
void computeBMatrix(torch::Tensor &Bmat, const unsigned int exp_i) const
Computes the covariance matrix for the outputs (using the latent coefficients) We use a $B = a_i a_i...
Definition: LMC.C:160
make_range
IntRange< T > make_range(T beg, T end)

◆ computedKdhyper()

bool LMC::computedKdhyper ( torch::Tensor &  dKdhp,
const torch::Tensor &  x,
const std::string &  hyper_param_name,
unsigned int  ind 
) const
overridevirtual

Redirect dK/dhp for hyperparameter "hp".

Returns false is the parameter has not been found in this covariance object.

Parameters
dKdhpThe matrix which should be populated with the derivatives
xThe input vector for which the derivatives of the covariance matrix is computed
hyper_param_nameThe name of the hyperparameter
indThe index within the hyperparameter. 0 if it is a scalar parameter. If it is a vector parameter, it should be the index within the vector.

Reimplemented from CovarianceFunctionBase.

Definition at line 86 of file LMC.C.

90 {
91  // Early return in the paramter name is longer than the expected [name] prefix.
92  // We prefix the parameter names with the name of the covariance function.
93  if (name().length() + 1 > hyper_param_name.length())
94  return false;
95 
96  // Strip the prefix from the given parameter name
97  const std::string name_without_prefix = hyper_param_name.substr(name().length() + 1);
98 
99  // Check if the parameter is tunable
100  if (_tunable_hp.find(hyper_param_name) != _tunable_hp.end())
101  {
102  const std::string acoeff_prefix = "acoeff_";
103  const std::string lambda_prefix = "lambda_";
104 
105  // Allocate storage for the factors of the total gradient matrix
106  const auto options = x.options().dtype(at::kDouble);
107  torch::Tensor dBdhp = torch::zeros({_num_outputs, _num_outputs}, options);
108  torch::Tensor K_params = torch::zeros({x.sizes()[0], x.sizes()[0]}, options);
109 
110  if (name_without_prefix.find(acoeff_prefix) != std::string::npos)
111  {
112  // Automatically grab the expansion index
113  const int number = std::stoi(name_without_prefix.substr(acoeff_prefix.length()));
114  computeAGradient(dBdhp, number, ind);
115  _covariance_functions[number]->computeCovarianceMatrix(K_params, x, x, true);
116  }
117  else if (name_without_prefix.find(lambda_prefix) != std::string::npos)
118  {
119  // Automatically grab the expansion index
120  const int number = std::stoi(name_without_prefix.substr(lambda_prefix.length()));
121  computeLambdaGradient(dBdhp, number, ind);
122  _covariance_functions[number]->computeCovarianceMatrix(K_params, x, x, true);
123  }
124  dKdhp = torch::kron(dBdhp, K_params);
125  return true;
126  }
127  else
128  {
129  // Allocate storage for the matrix factors
130  const auto options = x.options().dtype(at::kDouble);
131  torch::Tensor B_tmp = torch::zeros({_num_outputs, _num_outputs}, options);
132  torch::Tensor B = torch::zeros({_num_outputs, _num_outputs}, options);
133  torch::Tensor dKdhp_sub = torch::zeros({x.sizes()[0], x.sizes()[0]}, options);
134 
135  // First, check the dependent covariances
136  bool found = false;
137  for (const auto dependent_covar : _covariance_functions)
138  if (!found)
139  found = dependent_covar->computedKdhyper(dKdhp_sub, x, hyper_param_name, ind);
140 
141  if (!found)
142  mooseError("Hyperparameter ", hyper_param_name, "not found!");
143 
144  // Then we compute the output covariance
145  for (const auto exp_i : make_range(_num_expansion_terms))
146  {
147  computeBMatrix(B_tmp, exp_i);
148  B += B_tmp;
149  }
150 
151  dKdhp = torch::kron(B, dKdhp_sub);
152 
153  return true;
154  }
155 
156  return false;
157 }
CovarianceFunctionBase::_tunable_hp
std::unordered_set< std::string > _tunable_hp
list of tunable hyper-parameters
Definition: CovarianceFunctionBase.h:112
LMC::_num_expansion_terms
const unsigned int _num_expansion_terms
The number of expansion terms in the output ovariance matrix.
Definition: LMC.h:67
CovarianceFunctionBase::_covariance_functions
std::vector< CovarianceFunctionBase * > _covariance_functions
Vector of pointers to the dependent covariance functions.
Definition: CovarianceFunctionBase.h:124
MooseObject::name
const std::string & name() const
CovarianceFunctionBase::_num_outputs
const unsigned int _num_outputs
The number of outputs this covariance function is used to describe.
Definition: CovarianceFunctionBase.h:115
x
const std::vector< double > x
Definition: EquilibriumConstantFitTest.C:17
LMC::computeAGradient
void computeAGradient(torch::Tensor &grad, const unsigned int exp_i, const unsigned int index) const
Computes the gradient of $B$ with respect to the entries in $a_i$ in the following expression: $B = ...
Definition: LMC.C:168
LMC::computeBMatrix
void computeBMatrix(torch::Tensor &Bmat, const unsigned int exp_i) const
Computes the covariance matrix for the outputs (using the latent coefficients) We use a $B = a_i a_i...
Definition: LMC.C:160
LMC::computeLambdaGradient
void computeLambdaGradient(torch::Tensor &grad, const unsigned int exp_i, const unsigned int index) const
Computes the gradient of $B$ with respect to the entries in $lambda_i$ in the following expression: $...
Definition: LMC.C:183
make_range
IntRange< T > make_range(T beg, T end)
MooseObject::mooseError
void mooseError(Args &&... args) const

◆ computeLambdaGradient()

void LMC::computeLambdaGradient ( torch::Tensor &  grad,
const unsigned int  exp_i,
const unsigned int  index 
) const
protected

Computes the gradient of $B$ with respect to the entries in $lambda_i$ in the following expression: $B = a_i a_i^T + diag(lambda_i)$.

Parameters
gradThe gradient matrix that should be populated
exp_iThe index of the expansion of B
indexThe index within the vector $lambda_i$

Definition at line 183 of file LMC.C.

Referenced by computedKdhyper().

186 {
187  mooseAssert(index < _num_outputs, "Incorrect LMC lambda index.");
188  auto basis = torch::zeros_like(*_lambdas[exp_i]);
189  const auto index_tensor =
190  torch::tensor({cast_int<int64_t>(index)},
191  torch::TensorOptions().dtype(torch::kLong).device(_lambdas[exp_i]->device()));
192  basis.index_fill_(0, index_tensor, 1.0);
193  grad = torch::diag(basis);
194 }
LMC::_lambdas
std::vector< const torch::Tensor * > _lambdas
Definition: LMC.h:73
CovarianceFunctionBase::_num_outputs
const unsigned int _num_outputs
The number of outputs this covariance function is used to describe.
Definition: CovarianceFunctionBase.h:115
NS::grad
std::string grad(const std::string &var)
Definition: NS.h:92

◆ dependentCovarianceNames()

const std::vector<UserObjectName>& CovarianceFunctionBase::dependentCovarianceNames ( ) const
inlineinherited

Get the names of the dependent covariances.

Definition at line 67 of file CovarianceFunctionBase.h.

Referenced by StochasticTools::GaussianProcess::linkCovarianceFunction().

68  {
69  return _dependent_covariance_names;
70  }
CovarianceFunctionBase::_dependent_covariance_names
const std::vector< UserObjectName > _dependent_covariance_names
The names of the dependent covariance functions.
Definition: CovarianceFunctionBase.h:118

◆ dependentCovarianceTypes()

void CovarianceFunctionBase::dependentCovarianceTypes ( std::map< UserObjectName, std::string > &  name_type_map) const
inherited

Populate a map with the names and types of the dependent covariance functions.

Parameters
name_type_mapReference to the map which should be populated

Definition at line 204 of file CovarianceFunctionBase.C.

Referenced by StochasticTools::GaussianProcess::linkCovarianceFunction().

206 {
207  for (const auto dependent_covar : _covariance_functions)
208  {
209  dependent_covar->dependentCovarianceTypes(name_type_map);
210  name_type_map.insert(std::make_pair(dependent_covar->name(), dependent_covar->type()));
211  }
212 }
CovarianceFunctionBase::_covariance_functions
std::vector< CovarianceFunctionBase * > _covariance_functions
Vector of pointers to the dependent covariance functions.
Definition: CovarianceFunctionBase.h:124

◆ getCovarianceFunctionByName()

CovarianceFunctionBase * CovarianceInterface::getCovarianceFunctionByName ( const UserObjectName &  name) const
protectedinherited

Lookup a CovarianceFunction object by name and return pointer.

Definition at line 26 of file CovarianceInterface.C.

Referenced by ActiveLearningGaussianProcess::ActiveLearningGaussianProcess(), CovarianceFunctionBase::CovarianceFunctionBase(), GaussianProcessTrainer::GaussianProcessTrainer(), and GaussianProcessSurrogate::setupCovariance().

27 {
28  std::vector<CovarianceFunctionBase *> models;
29  _covar_feproblem.theWarehouse()
30  .query()
31  .condition<AttribName>(name)
32  .condition<AttribSystem>("CovarianceFunction")
33  .queryInto(models);
34  if (models.empty())
35  mooseError("Unable to find a CovarianceFunction object with the name '" + name + "'");
36  return models[0];
37 }
mooseError
void mooseError(Args &&... args)
FEProblemBase::theWarehouse
TheWarehouse & theWarehouse() const
name
const std::string name
Definition: Setup.h:21
TheWarehouse::query
Query query()
CovarianceInterface::_covar_feproblem
FEProblemBase & _covar_feproblem
Reference to FEProblemBase instance.
Definition: CovarianceInterface.h:31

◆ getTuningData()

bool CovarianceFunctionBase::getTuningData ( const std::string &  name,
unsigned int &  size,
Real &  min,
Real &  max 
) const
virtualinherited

Get the default minimum and maximum and size of a hyperparameter.

Returns false is the parameter has not been found in this covariance object.

Parameters
nameThe name of the hyperparameter
sizeReference to an unsigned int that will contain the size of the hyperparameter (will be populated with 1 if it is scalar)
minReference to a number which will be populated by the maximum allowed value of the hyperparameter
maxReference to a number which will be populated by the minimum allowed value of the hyperparameter

Definition at line 168 of file CovarianceFunctionBase.C.

Referenced by StochasticTools::GaussianProcess::generateTuningMap().

172 {
173  // First, check the dependent covariances
174  for (const auto dependent_covar : _covariance_functions)
175  if (dependent_covar->getTuningData(name, size, min, max))
176  return true;
177 
178  min = 1e-9;
179  max = 1e9;
180 
181  const auto tensor_value = _hyperparameters.find(name);
182  if (tensor_value == _hyperparameters.end())
183  {
184  size = 0;
185  return false;
186  }
187 
188  if (isScalarHyperParameter(tensor_value->second))
189  {
190  size = 1;
191  return true;
192  }
193 
194  if (isVectorHyperParameter(tensor_value->second))
195  {
196  size = tensor_value->second.numel();
197  return true;
198  }
199 
200  mooseError("Unsupported hyperparameter rank ", tensor_value->second.dim(), " for ", name, ".");
201 }
CovarianceFunctionBase::_hyperparameters
HyperParameterMap _hyperparameters
Map of hyperparameters stored as rank-0 or rank-1 tensors.
Definition: CovarianceFunctionBase.h:109
CovarianceFunctionBase::isVectorHyperParameter
static bool isVectorHyperParameter(const torch::Tensor &tensor)
Return true if a hyperparameter tensor stores a vector of values.
Definition: CovarianceFunctionBase.C:50
CovarianceFunctionBase::isScalarHyperParameter
static bool isScalarHyperParameter(const torch::Tensor &tensor)
Return true if a hyperparameter tensor stores one scalar value.
Definition: CovarianceFunctionBase.C:44
CovarianceFunctionBase::_covariance_functions
std::vector< CovarianceFunctionBase * > _covariance_functions
Vector of pointers to the dependent covariance functions.
Definition: CovarianceFunctionBase.h:124
max
auto max(const L &left, const R &right)
MooseObject::name
const std::string & name() const
MooseObject::mooseError
void mooseError(Args &&... args) const
min
auto min(const L &left, const R &right)

◆ isScalarHyperParameter()

bool CovarianceFunctionBase::isScalarHyperParameter ( const torch::Tensor &  tensor)
staticinherited

Return true if a hyperparameter tensor stores one scalar value.

Definition at line 44 of file CovarianceFunctionBase.C.

Referenced by CovarianceFunctionBase::buildHyperParamMap(), CovarianceFunctionBase::getTuningData(), GaussianProcessData::initialize(), and CovarianceFunctionBase::loadHyperParamMap().

45 {
46  return tensor.dim() == 0;
47 }

◆ isTunable()

bool CovarianceFunctionBase::isTunable ( const std::string &  name) const
virtualinherited

Check if a given parameter is tunable.

Parameters
Thename of the hyperparameter

Definition at line 115 of file CovarianceFunctionBase.C.

Referenced by StochasticTools::GaussianProcess::generateTuningMap().

116 {
117  // First, we check if the dependent covariances have the parameter
118  for (const auto dependent_covar : _covariance_functions)
119  if (dependent_covar->isTunable(name))
120  return true;
121 
122  if (_tunable_hp.find(name) != _tunable_hp.end())
123  return true;
124  else if (_hyperparameters.find(name) != _hyperparameters.end())
125  mooseError("We found hyperparameter ", name, " but it was not declared tunable!");
126 
127  return false;
128 }
CovarianceFunctionBase::_hyperparameters
HyperParameterMap _hyperparameters
Map of hyperparameters stored as rank-0 or rank-1 tensors.
Definition: CovarianceFunctionBase.h:109
CovarianceFunctionBase::_tunable_hp
std::unordered_set< std::string > _tunable_hp
list of tunable hyper-parameters
Definition: CovarianceFunctionBase.h:112
CovarianceFunctionBase::_covariance_functions
std::vector< CovarianceFunctionBase * > _covariance_functions
Vector of pointers to the dependent covariance functions.
Definition: CovarianceFunctionBase.h:124
MooseObject::name
const std::string & name() const
MooseObject::mooseError
void mooseError(Args &&... args) const

◆ isVectorHyperParameter()

bool CovarianceFunctionBase::isVectorHyperParameter ( const torch::Tensor &  tensor)
staticinherited

Return true if a hyperparameter tensor stores a vector of values.

Definition at line 50 of file CovarianceFunctionBase.C.

Referenced by CovarianceFunctionBase::buildHyperParamMap(), CovarianceFunctionBase::getTuningData(), GaussianProcessData::initialize(), and CovarianceFunctionBase::loadHyperParamMap().

51 {
52  return tensor.dim() == 1;
53 }

◆ loadHyperParamMap()

void CovarianceFunctionBase::loadHyperParamMap ( const HyperParameterMap &  map)
inherited

Load some hyperparameters into the local map contained in this object.

Parameters
mapInput map of hyperparameters

Definition at line 131 of file CovarianceFunctionBase.C.

Referenced by LoadCovarianceDataAction::load(), StochasticTools::GaussianProcess::setupCovarianceMatrix(), and StochasticTools::GaussianProcess::tuneHyperParamsAdam().

132 {
133  // First, load the hyperparameters of the dependent covariance functions
134  for (const auto dependent_covar : _covariance_functions)
135  dependent_covar->loadHyperParamMap(map);
136 
137  // Then we load the hyperparameters of this object
138  for (auto & iter : _hyperparameters)
139  {
140  const auto map_iter = map.find(iter.first);
141  if (map_iter == map.end())
142  continue;
143 
144  if (!isScalarHyperParameter(map_iter->second) && !isVectorHyperParameter(map_iter->second))
145  mooseError(
146  "Unsupported hyperparameter rank ", map_iter->second.dim(), " for ", iter.first, ".");
147 
148  iter.second = map_iter->second.clone();
149  }
150 }
CovarianceFunctionBase::_hyperparameters
HyperParameterMap _hyperparameters
Map of hyperparameters stored as rank-0 or rank-1 tensors.
Definition: CovarianceFunctionBase.h:109
CovarianceFunctionBase::isVectorHyperParameter
static bool isVectorHyperParameter(const torch::Tensor &tensor)
Return true if a hyperparameter tensor stores a vector of values.
Definition: CovarianceFunctionBase.C:50
CovarianceFunctionBase::isScalarHyperParameter
static bool isScalarHyperParameter(const torch::Tensor &tensor)
Return true if a hyperparameter tensor stores one scalar value.
Definition: CovarianceFunctionBase.C:44
CovarianceFunctionBase::_covariance_functions
std::vector< CovarianceFunctionBase * > _covariance_functions
Vector of pointers to the dependent covariance functions.
Definition: CovarianceFunctionBase.h:124
MooseObject::mooseError
void mooseError(Args &&... args) const

◆ numOutputs()

unsigned int CovarianceFunctionBase::numOutputs ( ) const
inlineinherited

Return the number of outputs assumed for this covariance function.

Definition at line 90 of file CovarianceFunctionBase.h.

Referenced by GaussianProcessSurrogate::evaluate(), GaussianProcessTrainer::GaussianProcessTrainer(), and StochasticTools::GaussianProcess::linkCovarianceFunction().

90 { return _num_outputs; }
CovarianceFunctionBase::_num_outputs
const unsigned int _num_outputs
The number of outputs this covariance function is used to describe.
Definition: CovarianceFunctionBase.h:115

◆ validParams()

InputParameters LMC::validParams ( )
static

Definition at line 18 of file LMC.C.

19 {
20  InputParameters params = CovarianceFunctionBase::validParams();
21  params.addClassDescription("Covariance function for multioutput Gaussian Processes based on the "
22  "Linear Model of Coregionalization (LMC).");
23  params.addParam<unsigned int>(
24  "num_latent_funcs", 1., "The number of latent functions for the expansion of the outputs.");
25  params.makeParamRequired<unsigned int>("num_outputs");
26  params.makeParamRequired<std::vector<UserObjectName>>("covariance_functions");
27  return params;
28 }
InputParameters::addParam
void addParam(const std::string &name, const std::initializer_list< typename T::value_type > &value, const std::string &doc_string)
CovarianceFunctionBase::validParams
static InputParameters validParams()
Definition: CovarianceFunctionBase.C:56
InputParameters::makeParamRequired
void makeParamRequired(const std::string &name)
InputParameters::addClassDescription
void addClassDescription(const std::string &doc_string)

Member Data Documentation

◆ _a_coeffs

std::vector<const torch::Tensor *> LMC::_a_coeffs
private

The vectors in the $B = a_i a_i^T + diag(lambda_i)$ expansion.

Definition at line 72 of file LMC.h.

Referenced by computeAGradient(), computeBMatrix(), and LMC().

◆ _covariance_functions

std::vector<CovarianceFunctionBase *> CovarianceFunctionBase::_covariance_functions
protectedinherited

Vector of pointers to the dependent covariance functions.

Definition at line 124 of file CovarianceFunctionBase.h.

Referenced by CovarianceFunctionBase::buildHyperParamMap(), computeCovarianceMatrix(), computedKdhyper(), CovarianceFunctionBase::CovarianceFunctionBase(), CovarianceFunctionBase::dependentCovarianceTypes(), CovarianceFunctionBase::getTuningData(), CovarianceFunctionBase::isTunable(), and CovarianceFunctionBase::loadHyperParamMap().

◆ _dependent_covariance_names

const std::vector<UserObjectName> CovarianceFunctionBase::_dependent_covariance_names
protectedinherited

The names of the dependent covariance functions.

Definition at line 118 of file CovarianceFunctionBase.h.

Referenced by CovarianceFunctionBase::CovarianceFunctionBase(), and CovarianceFunctionBase::dependentCovarianceNames().

◆ _dependent_covariance_types

std::vector<std::string> CovarianceFunctionBase::_dependent_covariance_types
protectedinherited

The types of the dependent covariance functions.

Definition at line 121 of file CovarianceFunctionBase.h.

Referenced by CovarianceFunctionBase::CovarianceFunctionBase().

◆ _hyperparameters

HyperParameterMap CovarianceFunctionBase::_hyperparameters
protectedinherited

Map of hyperparameters stored as rank-0 or rank-1 tensors.

Definition at line 109 of file CovarianceFunctionBase.h.

Referenced by CovarianceFunctionBase::addRealHyperParameter(), CovarianceFunctionBase::addVectorRealHyperParameter(), CovarianceFunctionBase::buildHyperParamMap(), CovarianceFunctionBase::getTuningData(), CovarianceFunctionBase::isTunable(), and CovarianceFunctionBase::loadHyperParamMap().

◆ _lambdas

std::vector<const torch::Tensor *> LMC::_lambdas
private

Definition at line 73 of file LMC.h.

Referenced by computeBMatrix(), computeLambdaGradient(), and LMC().

◆ _num_expansion_terms

const unsigned int LMC::_num_expansion_terms
protected

The number of expansion terms in the output ovariance matrix.

Definition at line 67 of file LMC.h.

Referenced by computeCovarianceMatrix(), computedKdhyper(), and LMC().

◆ _num_outputs

const unsigned int CovarianceFunctionBase::_num_outputs
protectedinherited

The number of outputs this covariance function is used to describe.

Definition at line 115 of file CovarianceFunctionBase.h.

Referenced by computeCovarianceMatrix(), computedKdhyper(), computeLambdaGradient(), LMC(), and CovarianceFunctionBase::numOutputs().

◆ _tunable_hp

std::unordered_set<std::string> CovarianceFunctionBase::_tunable_hp
protectedinherited

list of tunable hyper-parameters

Definition at line 112 of file CovarianceFunctionBase.h.

Referenced by CovarianceFunctionBase::addRealHyperParameter(), CovarianceFunctionBase::addVectorRealHyperParameter(), computedKdhyper(), and CovarianceFunctionBase::isTunable().

The documentation for this class was generated from the following files:
Generated on Mon Aug 3 2026 21:36:24 for https://mooseframework.inl.gov by   doxygen 1.8.14