FeatureSpace.hpp 4.95 KB
Newer Older
Thomas Purcell's avatar
Thomas Purcell committed
1
2
3
#ifndef FEATURE_SPACE
#define FEATURE_SPACE

Thomas Purcell's avatar
Thomas Purcell committed
4
#include <mpi_interface/MPI_Interface.hpp>
Thomas Purcell's avatar
Thomas Purcell committed
5
6
#include <feature_creation/node/FeatureNode.hpp>
#include <feature_creation/node/operator_nodes/allowed_ops.hpp>
Thomas Purcell's avatar
Thomas Purcell committed
7
#include <feature_creation/node/value_storage/nodes_value_containers.hpp>
Thomas Purcell's avatar
Thomas Purcell committed
8

Thomas Purcell's avatar
Thomas Purcell committed
9
10
#include <boost/serialization/shared_ptr.hpp>

Thomas Purcell's avatar
Thomas Purcell committed
11
#include <iostream>
Thomas Purcell's avatar
Thomas Purcell committed
12
#include <iomanip>
Thomas Purcell's avatar
Thomas Purcell committed
13

Thomas Purcell's avatar
Thomas Purcell committed
14
// namespace mpi = boost::mpi;
15
16
17
18
19
/**
 * @brief Feature Space for SISSO calculations
 * @details Stores and performs all feature calculations for SIS
 *
 */
Thomas Purcell's avatar
Thomas Purcell committed
20
21
class FeatureSpace
{
22
    std::vector<std::shared_ptr<FeatureNode>> _phi_selected; //!< selected features
23
24
25
26
27
28
29
30
31
32
33
34
35
    std::vector<node_ptr> _phi; //!< all features
    std::vector<node_ptr> _phi_0; //!< initial feature space

    std::vector<std::string> _allowed_ops; //!< list of all allowed operators strings
    std::vector<un_op_node_gen> _un_operators; //!< list of all unary operators
    std::vector<bin_op_node_gen> _com_bin_operators; //!< list of all commutable binary operators
    std::vector<bin_op_node_gen> _bin_operators; //!< list of all binary operators

    std::vector<double> _scores; //!< projection scores for each feature
    std::vector<double> _prop; //!< property to learn

    std::vector<int> _start_gen; //!< list of starting index for each generation

36
    std::shared_ptr<MPI_Interface> _mpi_comm; //!< MPi communicator
37
38
39
40

    double _l_bound; //!< lower bound for absolute value of the features
    double _u_bound; //!< upper bound for absolute value of the features

41
42
43
44
    int _max_phi; //!< Maximum rung for the feature creation
    int _n_sis_select; //!< Number of features to select for each dimensions
    int _n_samp; //!< Number of samples
    int _n_feat; //!< Total number of features
Thomas Purcell's avatar
Thomas Purcell committed
45
    int _n_rung_store; //!< Total rungs stored
46
    int _n_rung_generate; //!< Total number of rungs to generate on the fly
Thomas Purcell's avatar
Thomas Purcell committed
47
48

public:
49
50
51
52
53
54
55
56
57
58
    /**
     * @brief Constructor for the feature space
     * @details constructs the feature space from an initial set of features and a list of allowed operatiors
     *
     * @param mpi_comm MPI communicator for the calculations
     * @param allowed_ops list of allowed operators
     * @param max_phi highest rung value for the calculation
     * @param n_sis_select number of features to select during each SIS step
     * @param max_abs_feat_val maximum absolute feature value
     */
Thomas Purcell's avatar
Thomas Purcell committed
59
    FeatureSpace(
Thomas Purcell's avatar
Thomas Purcell committed
60
        std::shared_ptr<MPI_Interface> mpi_comm,
Thomas Purcell's avatar
Thomas Purcell committed
61
62
63
64
        std::vector<node_ptr> phi_0,
        std::vector<std::string> allowed_ops,
        int max_phi=1,
        int n_sis_select=1,
Thomas Purcell's avatar
Thomas Purcell committed
65
        int max_store_rung=2,
66
        int n_rung_generate=0,
Thomas Purcell's avatar
Thomas Purcell committed
67
68
        double min_abs_feat_val=1e-50,
        double max_abs_feat_val=1e50
Thomas Purcell's avatar
Thomas Purcell committed
69
70
    );

71
72
73
74
    /**
     * @brief Generate the full feature set from the allowed operators and initial feature set
     * @details populates phi with all features from an initial set and the allowed operators
     */
Thomas Purcell's avatar
Thomas Purcell committed
75
76
    void generate_feature_space();

77
78
79
    /**
     * @brief Accessor function for _phi_selected
     */
80
    inline std::vector<std::shared_ptr<FeatureNode>> phi_selected(){return _phi_selected;};
81
82
83
84

    /**
     * @brief Accessor function for _phi
     */
Thomas Purcell's avatar
Thomas Purcell committed
85
    inline std::vector<node_ptr> phi(){return _phi;};
86
87
88
89

    /**
     * @brief Accessor function for _phi_0
     */
Thomas Purcell's avatar
Thomas Purcell committed
90
    inline std::vector<node_ptr> phi0(){return _phi_0;};
91
92
93
94

    /**
     * @brief Accessor function for _scores
     */
Thomas Purcell's avatar
Thomas Purcell committed
95
    inline std::vector<double>& scores(){return _scores;};
96
97
98
99

    /**
     * @brief Accessor function for _mpi_comm
     */
Thomas Purcell's avatar
Thomas Purcell committed
100
    inline std::shared_ptr<MPI_Interface> mpi_comm(){return _mpi_comm;}
101
102
103
104
105
106
107

    /**
     * @brief calculate the projection scores for all features for a given property
     * @details Calculate the projection score based on the Pearson correlation
     *
     * @param prop [description]
     */
108
    void project_r(double* prop, int size);
Thomas Purcell's avatar
Thomas Purcell committed
109

110
111
112
113
114
115
116
117
118
    std::vector<double> project_r(double* prop, int size, std::vector<node_ptr>& phi);

    void generate_new_feats(std::vector<node_ptr>::iterator& feat, std::vector<node_ptr>& feat_set, int& feat_ind, double l_bound=1e-50, double u_bound=1e50);

    void project_generated(double* prop, int size, std::vector<std::shared_ptr<FeatureNode>>& phi_selected, std::vector<double>& scores_selected, std::vector<double>& scores_comp);

    bool valid_score_against_past(double* val_ptr, std::vector<double>& scores_comp);

    bool valid_score_against_current(int end_check, double* val_ptr, double cur_score, std::vector<double>& scores_selected, std::vector<double>& scores_comp);
119
120
121
122
123
124
    /**
     * @brief Perform SIS on a feature set with a specified property
     * @details Perform sure-independence screening with either the correct property
     *
     * @param prop The property to calculate SIS from
     */
Thomas Purcell's avatar
Thomas Purcell committed
125
    void sis(std::vector<double>& prop);
126
127
128
129
130
131
132
133
134

    /**
     * @brief Is a feature in this process' _phi?
     *
     * @param ind index
     * @return True if feature is in this _phi
     */
    inline bool feat_in_phi(int ind){return (ind >= _phi[0]->feat_ind()) && (ind <= _phi.back()->feat_ind());}

Thomas Purcell's avatar
Thomas Purcell committed
135
136
137
};

#endif