BNLearner_tpl.h

Go to the documentation of this file.

/****************************************************************************
 *   This file is part of the aGrUM/pyAgrum library.                        *
 *                                                                          *
 *   Copyright (c) 2005-2026 by                                             *
 *       - Pierre-Henri WUILLEMIN(_at_LIP6)                                 *
 *       - Christophe GONZALES(_at_AMU)                                     *
 *                                                                          *
 *   The aGrUM/pyAgrum library is free software; you can redistribute it    *
 *   and/or modify it under the terms of either :                           *
 *                                                                          *
 *    - the GNU Lesser General Public License as published by               *
 *      the Free Software Foundation, either version 3 of the License,      *
 *      or (at your option) any later version,                              *
 *    - the MIT license (MIT),                                              *
 *    - or both in dual license, as here.                                   *
 *                                                                          *
 *   (see https://agrum.gitlab.io/articles/dual-licenses-lgplv3mit.html)    *
 *                                                                          *
 *   This aGrUM/pyAgrum library is distributed in the hope that it will be  *
 *   useful, but WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED,          *
 *   INCLUDING BUT NOT LIMITED TO THE WARRANTIES MERCHANTABILITY or FITNESS *
 *   FOR A PARTICULAR PURPOSE  AND NONINFRINGEMENT. IN NO EVENT SHALL THE   *
 *   AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER *
 *   LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,        *
 *   ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR  *
 *   OTHER DEALINGS IN THE SOFTWARE.                                        *
 *                                                                          *
 *   See LICENCES for more details.                                         *
 *                                                                          *
 *   SPDX-FileCopyrightText: Copyright 2005-2026                            *
 *       - Pierre-Henri WUILLEMIN(_at_LIP6)                                 *
 *       - Christophe GONZALES(_at_AMU)                                     *
 *   SPDX-License-Identifier: LGPL-3.0-or-later OR MIT                      *
 *                                                                          *
 *   Contact  : info_at_agrum_dot_org                                       *
 *   homepage : http://agrum.gitlab.io                                      *
 *   gitlab   : https://gitlab.com/agrumery/agrum                           *
 *                                                                          *
 ****************************************************************************/
 
#pragma once
 

#include <fstream>
 
#include <agrum/BN/learning/BNLearner.h>   // to ease IDE parser
 
#ifndef DOXYGEN_SHOULD_SKIP_THIS
 
// to help IDE parser
#  include <agrum/BN/learning/BNLearner.h>
#  include <agrum/BN/learning/BNLearnUtils/BNLearnerListener.h>
 
namespace gum {
 
  namespace learning {
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >::BNLearner(std::string_view                  filename,
                                       const std::vector< std::string >& missingSymbols,
                                       const bool                        induceTypes) :
        IBNLearner(filename, missingSymbols, induceTypes) {
      GUM_CONSTRUCTOR(BNLearner);
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >::BNLearner(const DatabaseTable& db) : IBNLearner(db) {
      GUM_CONSTRUCTOR(BNLearner);
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >::BNLearner(std::string_view                   filename,
                                       const gum::BayesNet< GUM_SCALAR >& bn,
                                       const std::vector< std::string >&  missing_symbols) :
        IBNLearner(filename, bn, missing_symbols) {
      GUM_CONSTRUCTOR(BNLearner);
    }

    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >::BNLearner(const BNLearner< GUM_SCALAR >& src) : IBNLearner(src) {
      GUM_CONSTRUCTOR(BNLearner);
    }

    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >::BNLearner(BNLearner< GUM_SCALAR >&& src) : IBNLearner(src) {
      GUM_CONSTRUCTOR(BNLearner);
    }

    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >::~BNLearner() {
      GUM_DESTRUCTOR(BNLearner);
    }

 
    // ##########################################################################
    // ##########################################################################

    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >&
        BNLearner< GUM_SCALAR >::operator=(const BNLearner< GUM_SCALAR >& src) {
      IBNLearner::operator=(src);
      return *this;
    }

    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >&
        BNLearner< GUM_SCALAR >::operator=(BNLearner< GUM_SCALAR >&& src) noexcept {
      IBNLearner::operator=(std::move(src));
      return *this;
    }

    template < GUM_Numeric GUM_SCALAR >
    BayesNet< GUM_SCALAR > BNLearner< GUM_SCALAR >::learnBN() {
      // create the score, the prior and the estimator
      auto notification = checkScorePriorCompatibility();
      if (notification != "") { std::cout << "[aGrUM notification] " << notification << std::endl; }
      createPrior_();
      createScore_();
 
      std::unique_ptr< ParamEstimator > param_estimator(
          createParamEstimator_(scoreDatabase_.parser(), true));
 
      return dag2BN_.createBN< GUM_SCALAR >(*(param_estimator.get()), learnDag_());
    }
 
    // check that the database contains the nodes of the dag, else raise an exception
    template < GUM_Numeric GUM_SCALAR >
    void BNLearner< GUM_SCALAR >::_checkDAGCompatibility_(const DAG& dag) {
      // if the dag contains no node, this is compatible with the database
      if (dag.size() == 0) return;
 
      // check that the dag corresponds to the database
      std::vector< NodeId > ids;
      ids.reserve(dag.sizeNodes());
      for (const auto node: dag)
        ids.push_back(node);
      std::sort(ids.begin(), ids.end());
 
      if (ids.back() >= scoreDatabase_.names().size()) {
        std::string           str = "Learning parameters corresponding to the dag is impossible "
                                    "because the database does not contain the following nodeID";
        std::vector< NodeId > bad_ids;
        for (const auto node: ids) {
          if (node >= scoreDatabase_.names().size()) bad_ids.push_back(node);
        }
        if (bad_ids.size() > 1) str += 's';
        str += ": ";
        bool deja = false;
        for (const auto node: bad_ids) {
          if (deja) str += ", ";
          else deja = true;
          str += std::to_string(node);
        }
        GUM_ERROR(MissingVariableInDatabase, str)
      }
    }
 
    // learns a BN (its parameters) using a basic learning when its structure is known
    template < GUM_Numeric GUM_SCALAR >
    BayesNet< GUM_SCALAR > BNLearner< GUM_SCALAR >::_learnParameters_(const DAG& dag,
                                                                      bool takeIntoAccountScore) {
      // if the dag contains no node, return an empty BN
      if (dag.size() == 0) return BayesNet< GUM_SCALAR >();
 
      // be sure that the database contains dag's node ids
      _checkDAGCompatibility_(dag);
 
      // create the prior
      createPrior_();
 
      // check that the database does not contain any missing value
      if (scoreDatabase_.databaseTable().hasMissingValues()
          || ((priorDatabase_ != nullptr)
              && (priorType_ == BNLearnerPriorType::DIRICHLET_FROM_DATABASE)
              && priorDatabase_->databaseTable().hasMissingValues())) {
        GUM_ERROR(MissingValueInDatabase,
                  "In general, the BNLearner is unable to cope with "
                      << "missing values in databases. To learn parameters in "
                      << "such situations, you should first use method " << "useEM()");
      }
 
      // create the usual estimator
      DBRowGeneratorParser parser(scoreDatabase_.databaseTable().handler(), DBRowGeneratorSet());
      std::unique_ptr< ParamEstimator > param_estimator(
          createParamEstimator_(parser, takeIntoAccountScore));
 
      return dag2BN_.createBN< GUM_SCALAR >(*(param_estimator.get()), dag);
    }
 
    // initialize the parameter learning by EM
    // GCC 16 false positive: VRP inlines shared_ptr<ParamEstimator> control block and
    // incorrectly reports -Warray-bounds on the mutex; #pragma GCC diagnostic cannot
    // suppress warnings whose primary location is in system headers (stl_construct.h),
    // so we disable VRP for this function via optimize pragma instead
#  if defined(__GNUC__) && !defined(__clang__)
#    pragma GCC push_options
#    pragma GCC optimize("no-tree-vrp")
#  endif
 
    template < GUM_Numeric GUM_SCALAR >
    std::pair< std::shared_ptr< ParamEstimator >, std::shared_ptr< ParamEstimator > >
        BNLearner< GUM_SCALAR >::_initializeEMParameterLearning_(const DAG& dag,
                                                                 bool       takeIntoAccountScore) {
      // be sure that the database contains dag's node ids
      _checkDAGCompatibility_(dag);
 
      // create the prior
      createPrior_();
 
      // propagate the messages of dag2BN_ to the BNLearner so that the objects that listen
      // to the BNLearner can be informed of the progress of the EM's execution by dag2BN_
      // BNLearnerListener listener(this, dag2BN_);
 
      // get the column types
      const auto&       database = scoreDatabase_.databaseTable();
      const std::size_t nb_vars  = database.nbVariables();
      const std::vector< gum::learning::DBTranslatedValueType > col_types(
          nb_vars,
          gum::learning::DBTranslatedValueType::DISCRETE);
 
      // create the bootstrap estimator
      DBRowGenerator4CompleteRows generator_bootstrap(col_types);
      DBRowGeneratorSet           genset_bootstrap;
      genset_bootstrap.insertGenerator(generator_bootstrap);
      DBRowGeneratorParser              parser_bootstrap(database.handler(), genset_bootstrap);
      std::shared_ptr< ParamEstimator > param_estimator_bootstrap(
          createParamEstimator_(parser_bootstrap, takeIntoAccountScore));
 
      // create the EM estimator
      BayesNet< GUM_SCALAR >         dummy_bn;
      DBRowGeneratorEM< GUM_SCALAR > generator_EM(col_types, dummy_bn);
      DBRowGenerator&                gen_EM = generator_EM;   // fix for g++-4.8
      DBRowGeneratorSet              genset_EM;
      genset_EM.insertGenerator(gen_EM);
      DBRowGeneratorParser              parser_EM(database.handler(), genset_EM);
      std::shared_ptr< ParamEstimator > param_estimator_EM(
          createParamEstimator_(parser_EM, takeIntoAccountScore));
 
      return {param_estimator_bootstrap, param_estimator_EM};
    }
 
#  if defined(__GNUC__) && !defined(__clang__)
#    pragma GCC pop_options
#  endif
 
    // learns a BN (its parameters) with EM when its structure is known
    template < GUM_Numeric GUM_SCALAR >
    BayesNet< GUM_SCALAR >
        BNLearner< GUM_SCALAR >::_learnParametersWithEM_(const DAG& dag,
                                                         bool       takeIntoAccountScore) {
      // if the dag contains no node, return an empty BN
      if (dag.size() == 0) return BayesNet< GUM_SCALAR >();
 
      // get a pair containing the bootstrap and the EM estimators
      auto estimators = _initializeEMParameterLearning_(dag, takeIntoAccountScore);
 
      // perform the EM algorithm
      return dag2BN_.createBNwithEM< GUM_SCALAR >(*(estimators.first.get()),
                                                  *(estimators.second.get()),
                                                  dag);
    }

    template < GUM_Numeric GUM_SCALAR >
    BayesNet< GUM_SCALAR >
        BNLearner< GUM_SCALAR >::_learnParametersWithEM_(const BayesNet< GUM_SCALAR >& bn,
                                                         bool takeIntoAccountScore) {
      // if the dag contains no node, return an empty BN
      if (bn.internalDag().size() == 0) return BayesNet< GUM_SCALAR >();
 
      // get a pair containing the bootstrap and the EM estimators
      auto estimators = _initializeEMParameterLearning_(bn.internalDag(), takeIntoAccountScore);
 
      return dag2BN_.createBNwithEM< GUM_SCALAR >(*(estimators.first.get()),
                                                  *(estimators.second.get()),
                                                  bn);
    }

    template < GUM_Numeric GUM_SCALAR >
    BayesNet< GUM_SCALAR > BNLearner< GUM_SCALAR >::learnParameters(const DAG& dag,
                                                                    bool takeIntoAccountScore) {
      if (!scoreDatabase_.databaseTable().hasMissingValues() || !useEM_) {
        // here, we learn without EM
        return _learnParameters_(dag, takeIntoAccountScore);
      } else {
        // here we learn with EM
        return _learnParametersWithEM_(dag, takeIntoAccountScore);
      }
    }

    template < GUM_Numeric GUM_SCALAR >
    BayesNet< GUM_SCALAR >
        BNLearner< GUM_SCALAR >::learnParameters(const BayesNet< GUM_SCALAR >& bn,
                                                 bool takeIntoAccountScore) {
      if (!scoreDatabase_.databaseTable().hasMissingValues() || !useEM_) {
        DAG         dag;
        const auto& db = scoreDatabase_.databaseTable();
        for (const auto n: bn.nodes()) {
          dag.addNodeWithId(db.columnFromVariableName(bn.variable(n).name()));
        }
        for (const auto& arc: bn.arcs()) {
          dag.addArc(db.columnFromVariableName(bn.variable(arc.tail()).name()),
                     db.columnFromVariableName(bn.variable(arc.head()).name()));
        }
 
        // create le DAG en fonction des
        return _learnParameters_(dag, takeIntoAccountScore);
      } else {
        return _learnParametersWithEM_(bn, takeIntoAccountScore);
      }
    }

    template < GUM_Numeric GUM_SCALAR >
    BayesNet< GUM_SCALAR > BNLearner< GUM_SCALAR >::learnParameters(bool take_into_account_score) {
      return learnParameters(initialDag_, take_into_account_score);
    }
 
    template < GUM_Numeric GUM_SCALAR >
    NodeProperty< Sequence< std::string > >
        BNLearner< GUM_SCALAR >::_labelsFromBN_(std::string_view              filename,
                                                const BayesNet< GUM_SCALAR >& src) {
      std::ifstream in(std::string(filename), std::ifstream::in);
 
      if ((in.rdstate() & std::ifstream::failbit) != 0) {
        GUM_ERROR(gum::IOError, "File " << filename << " not found")
      }
 
      CSVParser parser(in, std::string(filename));
      parser.next();
      auto names = parser.current();
 
      NodeProperty< Sequence< std::string > > modals;
 
      for (gum::Idx col = 0; col < names.size(); col++) {
        if (src.exists(names[col])) {
          gum::NodeId graphId = src.idFromName(names[col]);
          modals.insert(col, gum::Sequence< std::string >());
 
          for (gum::Size i = 0; i < src.variable(graphId).domainSize(); ++i)
            modals[col].insert(src.variable(graphId).label(i));
        }
        // else: no problem, a column not in the BN...
      }
 
      return modals;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    std::string BNLearner< GUM_SCALAR >::toString() const {
      const auto st = state();
 
      Size maxkey = 0;
      for (const auto& tuple: st)
        if (std::get< 0 >(tuple).length() > maxkey) maxkey = std::get< 0 >(tuple).length();
 
      std::string s;
      for (const auto& tuple: st) {
        s += std::format("{:<{}} : {}", std::get< 0 >(tuple), maxkey, std::get< 1 >(tuple));
        if (std::get< 2 >(tuple) != "") s += std::format("  ({})", std::get< 2 >(tuple));
        s += '\n';
      }
      return s;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    std::vector< std::tuple< std::string, std::string, std::string > >
        BNLearner< GUM_SCALAR >::state() const {
      std::vector< std::tuple< std::string, std::string, std::string > > vals;
 
      std::string key;
      std::string comment;
      const auto& db = database();
 
      vals.emplace_back("Filename", filename_, "");
      vals.emplace_back("Size",
                        "(" + std::to_string(nbRows()) + "," + std::to_string(nbCols()) + ")",
                        "");
 
      std::string vars = "";
      for (NodeId i = 0; i < db.nbVariables(); i++) {
        if (i > 0) vars += ", ";
        vars += nameFromId(i) + "[" + std::to_string(db.domainSize(i)) + "]";
      }
      vals.emplace_back("Variables", vars, "");
      vals.emplace_back("Induced types", inducedTypes_ ? "True" : "False", "");
      vals.emplace_back("Missing values", hasMissingValues() ? "True" : "False", "");
 
      key = "Algorithm";
      switch (selectedAlgo_) {
        case AlgoType::GREEDY_HILL_CLIMBING :
          vals.emplace_back(key, "Greedy Hill Climbing", "");
          break;
        case AlgoType::EXTENDED_GREEDY_HILL_CLIMBING :
          vals.emplace_back(key, "Extended Greedy Hill Climbing", "");
          break;
        case AlgoType::K2 : {
          vals.emplace_back(key, "K2", "");
          const auto& k2order = algoK2_.order();
          vars                = "";
          for (NodeId i = 0; i < k2order.size(); i++) {
            if (i > 0) vars += ", ";
            vars += nameFromId(k2order.atPos(i));
          }
          vals.emplace_back("K2 order", vars, "");
        } break;
        case AlgoType::LOCAL_SEARCH_WITH_TABU_LIST :
          vals.emplace_back(key, "Local Search with Tabu List", "");
          vals.emplace_back("Tabu list size", std::to_string(nbDecreasingChanges_), "");
          break;
        case AlgoType::MIIC : vals.emplace_back(key, "MIIC", ""); break;
        case AlgoType::PC : vals.emplace_back(key, "PC", ""); break;
        case AlgoType::FCI : vals.emplace_back(key, "FCI", ""); break;
        case AlgoType::GREEDY_THICK_THINNING :
          vals.emplace_back(key, "Greedy Thick Thinning", "");
          break;
        default : vals.emplace_back(key, "(unknown)", "?"); break;
      }
 
      key = "Score";
 
      if (isScoreBased()) {
        switch (scoreType_) {
          case ScoreType::AIC : vals.emplace_back(key, "AIC", ""); break;
          case ScoreType::BIC : vals.emplace_back(key, "BIC", ""); break;
          case ScoreType::BD : vals.emplace_back(key, "BD", ""); break;
          case ScoreType::BDeu : vals.emplace_back(key, "BDeu", ""); break;
          case ScoreType::fNML : vals.emplace_back(key, "fNML", ""); break;
          case ScoreType::K2 : vals.emplace_back(key, "K2", ""); break;
          case ScoreType::LOG2LIKELIHOOD : vals.emplace_back(key, "Log2Likelihood", ""); break;
          case ScoreType::MDL : vals.emplace_back(key, "MDL", ""); break;
          default : vals.emplace_back(key, "(unknown)", "?"); break;
        }
      }
 
      if (isConstraintBased()) {
        key = "Correction";
        switch (kmodeMiic_) {
          case CorrectedMutualInformation::KModeTypes::MDL :
            vals.emplace_back(key, "MDL", "");
            break;
          case CorrectedMutualInformation::KModeTypes::NML :
            vals.emplace_back(key, "NML", "");
            break;
          case CorrectedMutualInformation::KModeTypes::NoCorr :
            vals.emplace_back(key, "No correction", "");
            break;
          default : vals.emplace_back(key, "(unknown)", "?"); break;
        }
      }
 
      key     = "Prior";
      comment = checkScorePriorCompatibility();
      switch (priorType_) {
        case BNLearnerPriorType::NO_prior : vals.emplace_back(key, "-", comment); break;
        case BNLearnerPriorType::DIRICHLET_FROM_DATABASE :
          vals.emplace_back(key, "Dirichlet", comment);
          vals.emplace_back("Dirichlet from database", priorDbname_, "");
          break;
        case BNLearnerPriorType::DIRICHLET_FROM_BAYESNET :
          vals.emplace_back(key, "Dirichlet", comment);
          vals.emplace_back("Dirichlet from Bayesian network : ", _prior_bn_.toString(), "");
          break;
        case BNLearnerPriorType::BDEU : vals.emplace_back(key, "BDEU", comment); break;
        case BNLearnerPriorType::SMOOTHING : vals.emplace_back(key, "Smoothing", comment); break;
        default : vals.emplace_back(key, "(unknown)", "?"); break;
      }
 
      if (priorType_ != BNLearnerPriorType::NO_prior)
        vals.emplace_back("Prior weight", std::to_string(priorWeight_), "");
 
      if (databaseWeight() != double(nbRows())) {
        vals.emplace_back("Database weight", std::to_string(databaseWeight()), "");
      }
 
      if (useEM_) {
        comment = "";
        if (!hasMissingValues()) comment = "But no missing values in this database";
        vals.emplace_back("use EM", "True", "");
        std::string s     = "[";
        bool        first = true;
        if (dag2BN_.isEnabledMinEpsilonRate()) {
          s += std::format("MinRate: {}", dag2BN_.minEpsilonRate());
          first = false;
        }
        if (dag2BN_.isEnabledEpsilon()) {
          if (!first) s += ", ";
          first = false;
          s += std::format("MinDiff: {}", dag2BN_.epsilon());
        }
        if (dag2BN_.isEnabledMaxIter()) {
          if (!first) s += ", ";
          first = false;
          s += std::format("MaxIter: {}", dag2BN_.maxIter());
        }
        if (dag2BN_.isEnabledMaxTime()) {
          if (!first) s += ", ";
          first = false;
          s += std::format("MaxTime: {}", dag2BN_.maxTime());
        }
        s += "]";
        vals.emplace_back("EM stopping criteria", s, comment);
      }
 
      std::string res;
      bool        nofirst;
      if (constraintIndegree_.maxIndegree() < std::numeric_limits< Size >::max()) {
        vals.emplace_back("Constraint Max InDegree",
                          std::to_string(constraintIndegree_.maxIndegree()),
                          "");
      }
      if (!constraintForbiddenArcs_.arcs().empty()) {
        res     = "{";
        nofirst = false;
        for (const auto& arc: constraintForbiddenArcs_.arcs()) {
          if (nofirst) res += ", ";
          else nofirst = true;
          res += nameFromId(arc.tail()) + "->" + nameFromId(arc.head());
        }
        res += "}";
        vals.emplace_back("Constraint Forbidden Arcs", res, "");
      }
      if (!constraintMandatoryArcs_.arcs().empty()) {
        res     = "{";
        nofirst = false;
        for (const auto& arc: constraintMandatoryArcs_.arcs()) {
          if (nofirst) res += ", ";
          else nofirst = true;
          res += nameFromId(arc.tail()) + "->" + nameFromId(arc.head());
        }
        res += "}";
        vals.emplace_back("Constraint Mandatory Arcs", res, "");
      }
      if (!constraintPossibleEdges_.edges().empty()) {
        res     = "{";
        nofirst = false;
        for (const auto& edge: constraintPossibleEdges_.edges()) {
          if (nofirst) res += ", ";
          else nofirst = true;
          res += nameFromId(edge.first()) + "--" + nameFromId(edge.second());
        }
        res += "}";
        vals.emplace_back("Constraint Possible Edges", res, "");
      }
      if (!constraintSliceOrder_.sliceOrder().empty()) {
        res               = "{";
        nofirst           = false;
        const auto& order = constraintSliceOrder_.sliceOrder();
        for (const auto& p: order) {
          if (nofirst) res += ", ";
          else nofirst = true;
          res += nameFromId(p.first) + ":" + std::to_string(p.second);
        }
        res += "}";
        vals.emplace_back("Constraint Slice Order", res, "");
      }
      if (!constraintNoParentNodes_.nodes().empty()) {
        res     = "{";
        nofirst = false;
        for (const auto& node: constraintNoParentNodes_.nodes()) {
          if (nofirst) res += ", ";
          else nofirst = true;
          res += nameFromId(node);
        }
        res += "}";
        vals.emplace_back("Constraint No Parent Nodes", res, "");
      }
      if (!constraintNoChildrenNodes_.nodes().empty()) {
        res     = "{";
        nofirst = false;
        for (const auto& node: constraintNoChildrenNodes_.nodes()) {
          if (nofirst) res += ", ";
          else nofirst = true;
          res += nameFromId(node);
        }
        res += "}";
        vals.emplace_back("Constraint No Children Nodes", res, "");
      }
      if (initialDag_.size() != 0) {
        vals.emplace_back("Initial DAG", "True", initialDag_.toDot());
      }
 
      return vals;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    void BNLearner< GUM_SCALAR >::copyState(const BNLearner< GUM_SCALAR >& learner) {
      switch (learner.selectedAlgo_) {
        case AlgoType::EXTENDED_GREEDY_HILL_CLIMBING : useExtendedGreedyHillClimbing(); break;
        case AlgoType::GREEDY_HILL_CLIMBING : useGreedyHillClimbing(); break;
        case AlgoType::GREEDY_THICK_THINNING :
          useGreedyThickThinning();
          setGreedyThickThinningReversals(learner.greedyThickThinningReversals());
          break;
        case AlgoType::K2 : useK2(learner.algoK2_.order()); break;
        case AlgoType::LOCAL_SEARCH_WITH_TABU_LIST :
          useLocalSearchWithTabuList(learner.nbDecreasingChanges_);
          break;
        case AlgoType::MIIC : useMIIC(); break;
        case AlgoType::PC : usePC(); break;
        case AlgoType::FCI : useFCI(); break;
      }
 
      switch (learner.scoreType_) {
        case ScoreType::K2 : useScoreK2(); break;
        case ScoreType::AIC : useScoreAIC(); break;
        case ScoreType::BIC : useScoreBIC(); break;
        case ScoreType::BD : useScoreBD(); break;
        case ScoreType::BDeu : useScoreBDeu(); break;
        case ScoreType::fNML : useScorefNML(); break;
        case ScoreType::MDL : useScoreMDL(); break;
        case ScoreType::LOG2LIKELIHOOD : useScoreLog2Likelihood(); break;
      }
 
      switch (learner.kmodeMiic_) {
        case CorrectedMutualInformation::KModeTypes::MDL : useMDLCorrection(); break;
        case CorrectedMutualInformation::KModeTypes::NML : useNMLCorrection(); break;
        case CorrectedMutualInformation::KModeTypes::NoCorr : useNoCorrection(); break;
      }
 
      switch (learner.priorType_) {
        case BNLearnerPriorType::NO_prior : useNoPrior(); break;
        case BNLearnerPriorType::DIRICHLET_FROM_DATABASE :
          useDirichletPrior(learner.priorDbname_, learner.priorWeight_);
          break;
        case BNLearnerPriorType::DIRICHLET_FROM_BAYESNET :
          useDirichletPrior(learner._prior_bn_);
          break;
        case BNLearnerPriorType::BDEU : useBDeuPrior(learner.priorWeight_); break;
        case BNLearnerPriorType::SMOOTHING : useSmoothingPrior(learner.priorWeight_); break;
      }
 
      useEM_   = learner.useEM_;
      noiseEM_ = learner.noiseEM_;
      dag2BN_  = learner.dag2BN_;
 
      setMaxIndegree(learner.constraintIndegree_.maxIndegree());
      for (const auto src: learner.constraintNoParentNodes_.nodes()) {
        try {
          const auto dst = idFromName(learner.nameFromId(src));
          addNoParentNode(dst);
        } catch (const MissingVariableInDatabase&) {
          // nothing to do
        }
      }
      for (const auto src: learner.constraintNoChildrenNodes_.nodes()) {
        try {
          const auto dst = idFromName(learner.nameFromId(src));
          addNoChildrenNode(dst);
        } catch (const MissingVariableInDatabase&) {
          // nothing to do
        }
      }
      for (const auto& arc: learner.constraintForbiddenArcs_.arcs()) {
        try {
          const auto src = idFromName(learner.nameFromId(arc.tail()));
          const auto dst = idFromName(learner.nameFromId(arc.head()));
          addForbiddenArc(src, dst);
        } catch (const MissingVariableInDatabase&) {
          // nothing to do
        }
      }
      for (const auto& arc: learner.constraintMandatoryArcs_.arcs()) {
        try {
          const auto src = idFromName(learner.nameFromId(arc.tail()));
          const auto dst = idFromName(learner.nameFromId(arc.head()));
          addMandatoryArc(src, dst);
        } catch (const MissingVariableInDatabase&) {
          // nothing to do
        }
      }
      for (const auto& edge: learner.constraintPossibleEdges_.edges()) {
        try {
          const auto src = idFromName(learner.nameFromId(edge.first()));
          const auto dst = idFromName(learner.nameFromId(edge.second()));
          addPossibleEdge(src, dst);
        } catch (const MissingVariableInDatabase&) {
          // nothing to do
        }
      }
      if (!learner.constraintSliceOrder_.sliceOrder().empty()) {
        NodeProperty< NodeId > slice_order;
        for (const auto& p: learner.constraintSliceOrder_.sliceOrder()) {
          try {
            slice_order.insert(idFromName(learner.nameFromId(p.first)), p.second);
          } catch (const MissingVariableInDatabase&) {
            // nothing to do
          }
        }
        setSliceOrder(slice_order);
      }
      if (!learner.constraintTotalOrder_.totalOrder().empty()) {
        setTotalOrder(learner.constraintTotalOrder_.totalOrder());
      }
    }
 
    template < GUM_Numeric GUM_SCALAR >
    void BNLearner< GUM_SCALAR >::createPrior_() {
      // first, save the old prior, to be delete if everything is ok
      Prior* old_prior = prior_;
 
      // create the new prior
      switch (priorType_) {
        case BNLearnerPriorType::NO_prior :
          prior_ = new NoPrior(scoreDatabase_.databaseTable(), scoreDatabase_.nodeId2Columns());
          break;
 
        case BNLearnerPriorType::SMOOTHING :
          prior_
              = new SmoothingPrior(scoreDatabase_.databaseTable(), scoreDatabase_.nodeId2Columns());
          break;
 
        case BNLearnerPriorType::DIRICHLET_FROM_DATABASE :
          if (priorDatabase_ != nullptr) {
            delete priorDatabase_;
            priorDatabase_ = nullptr;
          }
 
          priorDatabase_
              = new Database(priorDbname_, scoreDatabase_, scoreDatabase_.missingSymbols());
 
          prior_ = new DirichletPriorFromDatabase(scoreDatabase_.databaseTable(),
                                                  priorDatabase_->parser(),
                                                  priorDatabase_->nodeId2Columns());
          break;
 
        case BNLearnerPriorType::DIRICHLET_FROM_BAYESNET :
          prior_
              = new DirichletPriorFromBN< GUM_SCALAR >(scoreDatabase_.databaseTable(), &_prior_bn_);
          break;
 
        case BNLearnerPriorType::BDEU :
          prior_ = new BDeuPrior(scoreDatabase_.databaseTable(), scoreDatabase_.nodeId2Columns());
          break;
 
        default : GUM_ERROR(OperationNotAllowed, "The BNLearner does not support yet this prior")
      }
 
      // do not forget to assign a weight to the prior
      prior_->setWeight(priorWeight_);
 
      // remove the old prior, if any
      if (old_prior != nullptr) delete old_prior;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    std::ostream& operator<<(std::ostream& output, const BNLearner< GUM_SCALAR >& learner) {
      output << learner.toString();
      return output;
    }
 
    // =========================================================================
    // Delegation methods — each calls IBNLearner and returns *this
    // =========================================================================
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setInitialDAG(const DAG& dag) {
      IBNLearner::setInitialDAG(dag);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useEM(const double epsilon,
                                                            const double noise) {
      IBNLearner::useEM(epsilon, noise);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useEMWithRateCriterion(const double epsilon,
                                                                             const double noise) {
      IBNLearner::useEMWithRateCriterion(epsilon, noise);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useEMWithDiffCriterion(const double epsilon,
                                                                             const double noise) {
      IBNLearner::useEMWithDiffCriterion(epsilon, noise);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::forbidEM() {
      IBNLearner::forbidEM();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMsetEpsilon(const double eps) {
      IBNLearner::EMsetEpsilon(eps);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMdisableEpsilon() {
      IBNLearner::EMdisableEpsilon();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMenableEpsilon() {
      IBNLearner::EMenableEpsilon();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMsetMinEpsilonRate(const double rate) {
      IBNLearner::EMsetMinEpsilonRate(rate);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMdisableMinEpsilonRate() {
      IBNLearner::EMdisableMinEpsilonRate();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMenableMinEpsilonRate() {
      IBNLearner::EMenableMinEpsilonRate();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMsetMaxIter(const Size max) {
      IBNLearner::EMsetMaxIter(max);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMdisableMaxIter() {
      IBNLearner::EMdisableMaxIter();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMenableMaxIter() {
      IBNLearner::EMenableMaxIter();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMsetMaxTime(const double timeout) {
      IBNLearner::EMsetMaxTime(timeout);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMdisableMaxTime() {
      IBNLearner::EMdisableMaxTime();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMenableMaxTime() {
      IBNLearner::EMenableMaxTime();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMsetPeriodSize(const Size p) {
      IBNLearner::EMsetPeriodSize(p);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::EMsetVerbosity(const bool v) {
      IBNLearner::EMsetVerbosity(v);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useScoreAIC() {
      IBNLearner::useScoreAIC();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useScoreBD() {
      IBNLearner::useScoreBD();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useScoreBDeu() {
      IBNLearner::useScoreBDeu();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useScoreBIC() {
      IBNLearner::useScoreBIC();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useScoreK2() {
      IBNLearner::useScoreK2();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useScoreLog2Likelihood() {
      IBNLearner::useScoreLog2Likelihood();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useNoPrior() {
      IBNLearner::useNoPrior();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useBDeuPrior(double weight) {
      IBNLearner::useBDeuPrior(weight);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useSmoothingPrior(double weight) {
      IBNLearner::useSmoothingPrior(weight);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useDirichletPrior(std::string_view filename,
                                                                        double           weight) {
      IBNLearner::useDirichletPrior(filename, weight);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >&
        BNLearner< GUM_SCALAR >::useDirichletPrior(const gum::BayesNet< GUM_SCALAR >& bn,
                                                   double                             weight) {
      _prior_bn_ = bn;
      priorType_ = BNLearnerPriorType::DIRICHLET_FROM_BAYESNET;
      _setPriorWeight_(weight);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useGreedyHillClimbing() {
      IBNLearner::useGreedyHillClimbing();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useExtendedGreedyHillClimbing() {
      IBNLearner::useExtendedGreedyHillClimbing();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useGreedyThickThinning() {
      IBNLearner::useGreedyThickThinning();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setGreedyThickThinningReversals(bool allow) {
      IBNLearner::setGreedyThickThinningReversals(allow);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    bool BNLearner< GUM_SCALAR >::greedyThickThinningReversals() const {
      return IBNLearner::greedyThickThinningReversals();
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useLocalSearchWithTabuList(Size tabu_size,
                                                                                 Size nb_decrease) {
      IBNLearner::useLocalSearchWithTabuList(tabu_size, nb_decrease);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useK2(const Sequence< NodeId >& order) {
      IBNLearner::useK2(order);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useK2(const std::vector< NodeId >& order) {
      IBNLearner::useK2(order);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useMIIC() {
      IBNLearner::useMIIC();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::usePC() {
      IBNLearner::usePC();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useFCI() {
      IBNLearner::useFCI();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useFCIChi2Test() {
      IBNLearner::useFCIChi2Test();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useFCIG2Test() {
      IBNLearner::useFCIG2Test();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setFCIAlpha(double alpha) {
      IBNLearner::setFCIAlpha(alpha);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setFCIMaxPathLength(Size max_len) {
      IBNLearner::setFCIMaxPathLength(max_len);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setFCIExhaustiveSepSet(bool exhaustive) {
      IBNLearner::setFCIExhaustiveSepSet(exhaustive);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    bool BNLearner< GUM_SCALAR >::fciExhaustiveSepSet() const {
      return IBNLearner::fciExhaustiveSepSet();
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useChi2Test() {
      IBNLearner::useChi2Test();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useG2Test() {
      IBNLearner::useG2Test();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setPCAlpha(double alpha) {
      IBNLearner::setPCAlpha(alpha);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setPCStable(bool stable) {
      IBNLearner::setPCStable(stable);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setPCMaxCondSetSize(Size max_k) {
      IBNLearner::setPCMaxCondSetSize(max_k);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setPCUnshieldedColliderSorted(bool sorted) {
      IBNLearner::setPCUnshieldedColliderSorted(sorted);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useNMLCorrection() {
      IBNLearner::useNMLCorrection();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useMDLCorrection() {
      IBNLearner::useMDLCorrection();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::useNoCorrection() {
      IBNLearner::useNoCorrection();
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setMaxIndegree(Size max_indegree) {
      IBNLearner::setMaxIndegree(max_indegree);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >&
        BNLearner< GUM_SCALAR >::setSliceOrder(const NodeProperty< NodeId >& slice_order) {
      IBNLearner::setSliceOrder(slice_order);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setSliceOrder(
        const std::vector< std::vector< std::string > >& slices) {
      IBNLearner::setSliceOrder(slices);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >&
        BNLearner< GUM_SCALAR >::setTotalOrder(const std::vector< std::string >& order) {
      IBNLearner::setTotalOrder(order);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >&
        BNLearner< GUM_SCALAR >::setTotalOrder(const Sequence< NodeId >& order) {
      IBNLearner::setTotalOrder(order);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setForbiddenArcs(const ArcSet& set) {
      IBNLearner::setForbiddenArcs(set);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addForbiddenArc(const Arc& arc) {
      IBNLearner::addForbiddenArc(arc);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addForbiddenArc(NodeId tail, NodeId head) {
      IBNLearner::addForbiddenArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addForbiddenArc(std::string_view tail,
                                                                      std::string_view head) {
      IBNLearner::addForbiddenArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseForbiddenArc(const Arc& arc) {
      IBNLearner::eraseForbiddenArc(arc);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseForbiddenArc(NodeId tail, NodeId head) {
      IBNLearner::eraseForbiddenArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseForbiddenArc(std::string_view tail,
                                                                        std::string_view head) {
      IBNLearner::eraseForbiddenArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addMandatoryArc(const Arc& arc) {
      IBNLearner::addMandatoryArc(arc);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addMandatoryArc(NodeId tail, NodeId head) {
      IBNLearner::addMandatoryArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addMandatoryArc(std::string_view tail,
                                                                      std::string_view head) {
      IBNLearner::addMandatoryArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseMandatoryArc(const Arc& arc) {
      IBNLearner::eraseMandatoryArc(arc);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseMandatoryArc(NodeId tail, NodeId head) {
      IBNLearner::eraseMandatoryArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseMandatoryArc(std::string_view tail,
                                                                        std::string_view head) {
      IBNLearner::eraseMandatoryArc(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addPossibleEdge(const Edge& edge) {
      IBNLearner::addPossibleEdge(edge);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addPossibleEdge(NodeId tail, NodeId head) {
      IBNLearner::addPossibleEdge(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addPossibleEdge(std::string_view tail,
                                                                      std::string_view head) {
      IBNLearner::addPossibleEdge(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::erasePossibleEdge(const Edge& edge) {
      IBNLearner::erasePossibleEdge(edge);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::erasePossibleEdge(NodeId tail, NodeId head) {
      IBNLearner::erasePossibleEdge(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::erasePossibleEdge(std::string_view tail,
                                                                        std::string_view head) {
      IBNLearner::erasePossibleEdge(tail, head);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setMandatoryArcs(const ArcSet& set) {
      IBNLearner::setMandatoryArcs(set);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::setPossibleEdges(const EdgeSet& set) {
      IBNLearner::setPossibleEdges(set);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >&
        BNLearner< GUM_SCALAR >::setPossibleSkeleton(const UndiGraph& skeleton) {
      IBNLearner::setPossibleSkeleton(skeleton);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addNoParentNode(NodeId node) {
      IBNLearner::addNoParentNode(node);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addNoParentNode(std::string_view name) {
      IBNLearner::addNoParentNode(name);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseNoParentNode(NodeId node) {
      IBNLearner::eraseNoParentNode(node);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseNoParentNode(std::string_view name) {
      IBNLearner::eraseNoParentNode(name);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addNoChildrenNode(NodeId node) {
      IBNLearner::addNoChildrenNode(node);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::addNoChildrenNode(std::string_view name) {
      IBNLearner::addNoChildrenNode(name);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseNoChildrenNode(NodeId node) {
      IBNLearner::eraseNoChildrenNode(node);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::eraseNoChildrenNode(std::string_view name) {
      IBNLearner::eraseNoChildrenNode(name);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::allowArcAdditions(bool allow) {
      IBNLearner::allowArcAdditions(allow);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::allowArcDeletions(bool allow) {
      IBNLearner::allowArcDeletions(allow);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::allowArcReversals(bool allow) {
      IBNLearner::allowArcReversals(allow);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    BNLearner< GUM_SCALAR >& BNLearner< GUM_SCALAR >::allowArcTriangleDeletions(bool allow) {
      IBNLearner::allowArcTriangleDeletions(allow);
      return *this;
    }
 
    template < GUM_Numeric GUM_SCALAR >
    bool BNLearner< GUM_SCALAR >::isConstraintBased() const {
      return IBNLearner::isConstraintBased();
    }
 
    template < GUM_Numeric GUM_SCALAR >
    bool BNLearner< GUM_SCALAR >::isScoreBased() const {
      return IBNLearner::isScoreBased();
    }
 
  } /* namespace learning */
 
} /* namespace gum */
 
#endif /* DOXYGEN_SHOULD_SKIP_THIS */