CompressedSparseMatrix.hpp

Go to the documentation of this file.

#ifndef TATAMI_COMPRESSED_SPARSE_MATRIX_H
#define TATAMI_COMPRESSED_SPARSE_MATRIX_H
 
#include <vector>
#include <algorithm>
#include <memory>
#include <utility>
#include <stdexcept>
#include <cstddef>
 
#include "sanisizer/sanisizer.hpp"
 
#include "../base/Matrix.hpp"
#include "../utils/ElementType.hpp"
#include "../utils/copy.hpp"
#include "../utils/PseudoOracularExtractor.hpp"
 
#include "primary_extraction.hpp"
#include "secondary_extraction.hpp"
 
namespace tatami {
 
namespace CompressedSparseMatrix_internal {
 
/********************
 *** Primary full ***
 ********************/
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class PrimaryMyopicFullDense final : public MyopicDenseExtractor<Value_, Index_> {
public:
    PrimaryMyopicFullDense(const ValueStorage_& values, const IndexStorage_& indices, const PointerStorage_& pointers, const Index_ secondary) :
        my_values(values),
        my_indices(indices), 
        my_pointers(pointers),
        my_secondary(secondary) 
    {} 
 
    const Value_* fetch(const Index_ i, Value_* const buffer) {
        const auto start_pos = my_pointers[i], end_pos = my_pointers[i+1];
        std::fill_n(buffer, my_secondary, static_cast<Value_>(0));
        for (auto x = start_pos; x < end_pos; ++x) {
            buffer[my_indices[x]] = my_values[x];
        }
        return buffer;
    }
 
private:
    const ValueStorage_& my_values;
    const IndexStorage_& my_indices;
    const PointerStorage_& my_pointers;
    Index_ my_secondary;
};
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class PrimaryMyopicFullSparse final : public MyopicSparseExtractor<Value_, Index_> {
public:
    PrimaryMyopicFullSparse(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const Options& opt
    ) :
        my_values(values),
        my_indices(indices),
        my_pointers(pointers),
        my_secondary(secondary),
        my_needs_value(opt.sparse_extract_value),
        my_needs_index(opt.sparse_extract_index) 
    {} 
 
    SparseRange<Value_, Index_> fetch(const Index_ i, Value_* const value_buffer, Index_* const index_buffer) {
        const auto offset = my_pointers[i];
        I<decltype(offset)> delta = my_pointers[i+1] - offset;
 
        SparseRange<Value_, Index_> output(delta, NULL, NULL);
        if (my_needs_value) {
            output.value = sparse_utils::extract_primary_vector(my_values, offset, delta, value_buffer);
        }
        if (my_needs_index) {
            output.index = sparse_utils::extract_primary_vector(my_indices, offset, delta, index_buffer);
        }
        return output;
    }
 
private:
    const ValueStorage_& my_values;
    const IndexStorage_& my_indices;
    const PointerStorage_& my_pointers;
    Index_ my_secondary;
    bool my_needs_value, my_needs_index;
};
 
/*********************
 *** Primary block ***
 *********************/
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class PrimaryMyopicBlockDense final : public MyopicDenseExtractor<Value_, Index_> {
public:
    PrimaryMyopicBlockDense(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const Index_ block_start,
        const Index_ block_length
    ) :
        my_values(values), 
        my_indices(indices), 
        my_pointers(pointers), 
        my_secondary(secondary), 
        my_block_start(block_start), 
        my_block_length(block_length) 
    {} 
 
    const Value_* fetch(const Index_ i, Value_* const buffer) {
        auto iStart = my_indices.begin() + my_pointers[i];
        auto iEnd = my_indices.begin() + my_pointers[i + 1];
        sparse_utils::refine_primary_block_limits(iStart, iEnd, my_secondary, my_block_start, my_block_length);
        const auto start_pos = (iStart - my_indices.begin());
        const auto end_pos = (iEnd - my_indices.begin());
 
        std::fill_n(buffer, my_block_length, static_cast<Value_>(0));
        for (auto x = start_pos; x < end_pos; ++x) {
            buffer[my_indices[x] - my_block_start] = my_values[x];
        }
        return buffer;
    }
 
private:
    const ValueStorage_& my_values;
    const IndexStorage_& my_indices;
    const PointerStorage_& my_pointers;
    Index_ my_secondary;
    Index_ my_block_start, my_block_length;
};
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class PrimaryMyopicBlockSparse final : public MyopicSparseExtractor<Value_, Index_> {
public:
    PrimaryMyopicBlockSparse(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const Index_ block_start,
        const Index_ block_length,
        const Options& opt
    ) :
        my_values(values),
        my_indices(indices),
        my_pointers(pointers),
        my_secondary(secondary),
        my_block_start(block_start),
        my_block_length(block_length),
        my_needs_value(opt.sparse_extract_value),
        my_needs_index(opt.sparse_extract_index) 
    {} 
 
    SparseRange<Value_, Index_> fetch(const Index_ i, Value_* const value_buffer, Index_* const index_buffer) {
        auto iStart = my_indices.begin() + my_pointers[i];
        auto iEnd = my_indices.begin() + my_pointers[i + 1];
        sparse_utils::refine_primary_block_limits(iStart, iEnd, my_secondary, my_block_start, my_block_length);
        const auto offset = iStart - my_indices.begin();
        const auto delta = iEnd - iStart;
 
        SparseRange<Value_, Index_> output(delta, NULL, NULL);
        if (my_needs_value) {
            output.value = sparse_utils::extract_primary_vector(my_values, offset, delta, value_buffer);
        }
        if (my_needs_index) {
            output.index = sparse_utils::extract_primary_vector(my_indices, offset, delta, index_buffer);
        }
        return output;
    }
 
private:
    const ValueStorage_& my_values;
    const IndexStorage_& my_indices;
    const PointerStorage_& my_pointers;
    Index_ my_secondary;
    Index_ my_block_start, my_block_length;
    bool my_needs_value, my_needs_index;
};
 
/***********************
 *** Primary indexed ***
 ***********************/
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class PrimaryMyopicIndexDense final : public MyopicDenseExtractor<Value_, Index_> {
public:
    PrimaryMyopicIndexDense(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const VectorPtr<Index_>& indices_ptr
    ) :
        my_values(values),
        my_indices(indices),
        my_pointers(pointers),
        my_retriever(*indices_ptr, secondary),
        my_num_indices(indices_ptr->size()) 
    {}
 
    const Value_* fetch(const Index_ i, Value_* const buffer) {
        std::fill_n(buffer, my_num_indices, static_cast<Value_>(0));
        const auto vIt = my_values.begin() + my_pointers[i];
        my_retriever.populate(
            my_indices.begin() + my_pointers[i], 
            my_indices.begin() + my_pointers[i+1],
            [&](const auto s, const auto offset) -> void {
                buffer[s] = *(vIt + offset);
            }
        );
        return buffer;
    }
 
private:
    const ValueStorage_& my_values;
    const IndexStorage_& my_indices;
    const PointerStorage_& my_pointers;
    sparse_utils::RetrievePrimarySubsetDense<Index_> my_retriever;
    std::size_t my_num_indices;
};
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class PrimaryMyopicIndexSparse final : public MyopicSparseExtractor<Value_, Index_> {
public:
    PrimaryMyopicIndexSparse(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const VectorPtr<Index_>& indices_ptr,
        const Options& opt
    ) :
        my_values(values),
        my_indices(indices), 
        my_pointers(pointers),
        my_retriever(*indices_ptr, secondary),
        my_needs_value(opt.sparse_extract_value),
        my_needs_index(opt.sparse_extract_index) {} 
 
    SparseRange<Value_, Index_> fetch(const Index_ i, Value_* const value_buffer, Index_* const index_buffer) {
        Index_ count = 0;
        auto vcopy = value_buffer;
        auto icopy = index_buffer;
 
        auto vIt = my_values.begin() + my_pointers[i];
        my_retriever.populate(
            my_indices.begin() + my_pointers[i], 
            my_indices.begin() + my_pointers[i+1],
            [&](const auto offset, const auto ix) -> void {
                ++count;
                if (my_needs_value) {
                    *vcopy = *(vIt + offset);
                    ++vcopy;
                }
                if (my_needs_index) {
                    *icopy = ix;
                    ++icopy;
                }
            }
        );
 
        return SparseRange<Value_, Index_>(count, my_needs_value ? value_buffer : NULL, my_needs_index ? index_buffer : NULL);
    }
 
private:
    const ValueStorage_& my_values;
    const IndexStorage_& my_indices;
    const PointerStorage_& my_pointers;
    sparse_utils::RetrievePrimarySubsetSparse<Index_> my_retriever;
    bool my_needs_value, my_needs_index;
};
 
/**********************
 *** Secondary full ***
 **********************/
 
template<typename Index_, class IndexStorage_, class PointerStorage_>
class ServeIndices {
public:
    ServeIndices(const IndexStorage_& i, const PointerStorage_& p) : my_indices(i), my_pointers(p) {}
 
private:
    const IndexStorage_& my_indices;
    const PointerStorage_& my_pointers;
 
public:
    typedef ElementType<PointerStorage_> Pointer;
 
    Pointer start_offset(const Index_ primary) const {
        return my_pointers[primary];
    }
 
    Pointer end_offset(const Index_ primary) const {
        return my_pointers[primary + 1];
    }
 
    auto raw(const Index_) const {
        return my_indices.begin();
    }
};
 
template<typename Index_, class IndexStorage_, class PointerStorage_>
auto make_ServeIndices(const IndexStorage_& i, const PointerStorage_& p) {
    return ServeIndices<Index_, IndexStorage_, PointerStorage_>(i, p);
}
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class SecondaryMyopicFullDense final : public MyopicDenseExtractor<Value_, Index_> {
public:
    SecondaryMyopicFullDense(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary
    ) :
        my_values(values),
        my_cache(make_ServeIndices<Index_>(indices, pointers), secondary, pointers.size() - 1) 
    {} 
 
    const Value_* fetch(const Index_ i, Value_* const buffer) {
        std::fill_n(buffer, my_cache.size(), static_cast<Value_>(0));
        my_cache.search(
            i,
            [&](const Index_, const Index_ index_primary, const auto ptr) -> void {
                buffer[index_primary] = my_values[ptr];
            }
        );
        return buffer;
    }
 
private:
    const ValueStorage_& my_values;
    sparse_utils::FullSecondaryExtractionCache<Index_, ServeIndices<Index_, IndexStorage_, PointerStorage_> > my_cache;
};
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class SecondaryMyopicFullSparse final : public MyopicSparseExtractor<Value_, Index_> {
public:
    SecondaryMyopicFullSparse(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const Options& opt
    ) :
        my_values(values),
        my_cache(make_ServeIndices<Index_>(indices, pointers), secondary, pointers.size() - 1),
        my_needs_value(opt.sparse_extract_value),
        my_needs_index(opt.sparse_extract_index) 
    {} 
 
    SparseRange<Value_, Index_> fetch(const Index_ i, Value_* const value_buffer, Index_* const index_buffer) {
        Index_ count = 0;
        my_cache.search(i, [&](const Index_ primary, Index_, const ElementType<PointerStorage_> ptr) -> void {
            if (my_needs_value) {
                value_buffer[count] = my_values[ptr];
            }
            if (my_needs_index) {
                index_buffer[count] = primary;
            }
            ++count;
        });
        return SparseRange<Value_, Index_>(count, my_needs_value ? value_buffer : NULL, my_needs_index ? index_buffer : NULL);
    }
 
private:
    const ValueStorage_& my_values;
    sparse_utils::FullSecondaryExtractionCache<Index_, ServeIndices<Index_, IndexStorage_, PointerStorage_> > my_cache;
    bool my_needs_value, my_needs_index;
};
 
/***********************
 *** Secondary block ***
 ***********************/
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class SecondaryMyopicBlockDense final : public MyopicDenseExtractor<Value_, Index_> {
public:
    SecondaryMyopicBlockDense(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const Index_ block_start,
        const Index_ block_length
    ) :
        my_values(values),
        my_cache(make_ServeIndices<Index_>(indices, pointers), secondary, block_start, block_length) 
    {}
 
    const Value_* fetch(const Index_ i, Value_* const buffer) {
        std::fill_n(buffer, my_cache.size(), static_cast<Value_>(0));
        my_cache.search(
            i,
            [&](const Index_, const Index_ index_primary, const auto ptr) -> void {
                buffer[index_primary] = my_values[ptr];
            }
        );
        return buffer;
    }
 
private:
    const ValueStorage_& my_values;
    sparse_utils::BlockSecondaryExtractionCache<Index_, ServeIndices<Index_, IndexStorage_, PointerStorage_> > my_cache;
};
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class SecondaryMyopicBlockSparse final : public MyopicSparseExtractor<Value_, Index_> {
public:
    SecondaryMyopicBlockSparse(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        const Index_ block_start,
        const Index_ block_length,
        const Options& opt
    ) :
        my_values(values),
        my_cache(make_ServeIndices<Index_>(indices, pointers), secondary, block_start, block_length), 
        my_needs_value(opt.sparse_extract_value),
        my_needs_index(opt.sparse_extract_index) 
    {} 
 
    SparseRange<Value_, Index_> fetch(const Index_ i, Value_* const value_buffer, Index_* const index_buffer) {
        Index_ count = 0;
        my_cache.search(
            i, 
            [&](Index_ primary, Index_, auto ptr) -> void {
                if (my_needs_value) {
                    value_buffer[count] = my_values[ptr];
                }
                if (my_needs_index) {
                    index_buffer[count] = primary;
                }
                ++count;
            }
        );
        return SparseRange<Value_, Index_>(count, my_needs_value ? value_buffer : NULL, my_needs_index ? index_buffer : NULL);
    }
 
private:
    const ValueStorage_& my_values;
    sparse_utils::BlockSecondaryExtractionCache<Index_, ServeIndices<Index_, IndexStorage_, PointerStorage_> > my_cache;
    bool my_needs_value, my_needs_index;
};
 
/***********************
 *** Secondary index ***
 ***********************/
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class SecondaryMyopicIndexDense final : public MyopicDenseExtractor<Value_, Index_> {
public:
    SecondaryMyopicIndexDense(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        VectorPtr<Index_> sub_ptr
    ) :
        my_values(values),
        my_cache(make_ServeIndices<Index_>(indices, pointers), secondary, std::move(sub_ptr))
    {}
 
    const Value_* fetch(const Index_ i, Value_* const buffer) {
        std::fill_n(buffer, my_cache.size(), static_cast<Value_>(0));
        my_cache.search(
            i, 
            [&](const Index_, const Index_ index_primary, const ElementType<PointerStorage_> ptr) -> void {
                buffer[index_primary] = my_values[ptr];
            }
        );
        return buffer;
    }
 
private:
    const ValueStorage_& my_values;
    sparse_utils::IndexSecondaryExtractionCache<Index_, ServeIndices<Index_, IndexStorage_, PointerStorage_> > my_cache;
};
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class SecondaryMyopicIndexSparse final : public MyopicSparseExtractor<Value_, Index_> {
public:
    SecondaryMyopicIndexSparse(
        const ValueStorage_& values,
        const IndexStorage_& indices,
        const PointerStorage_& pointers,
        const Index_ secondary,
        VectorPtr<Index_> sub_ptr,
        const Options& opt
    ) :
        my_values(values),
        my_cache(make_ServeIndices<Index_>(indices, pointers), secondary, std::move(sub_ptr)),
        my_needs_value(opt.sparse_extract_value),
        my_needs_index(opt.sparse_extract_index) 
    {} 
 
    SparseRange<Value_, Index_> fetch(const Index_ i, Value_* const value_buffer, Index_* const index_buffer) {
        Index_ count = 0;
        my_cache.search(
            i,
            [&](Index_ primary, Index_, ElementType<PointerStorage_> ptr) -> void {
                if (my_needs_value) {
                    value_buffer[count] = my_values[ptr];
                }
                if (my_needs_index) {
                    index_buffer[count] = primary;
                }
                ++count;
            }
        );
        return SparseRange<Value_, Index_>(count, my_needs_value ? value_buffer : NULL, my_needs_index ? index_buffer : NULL);
    }
 
private:
    const ValueStorage_& my_values;
    sparse_utils::IndexSecondaryExtractionCache<Index_, ServeIndices<Index_, IndexStorage_, PointerStorage_> > my_cache;
    bool my_needs_value, my_needs_index;
};
 
}
struct CompressedSparseMatrixOptions {
    bool check = true;
};
 
template<typename Value_, typename Index_, class ValueStorage_, class IndexStorage_, class PointerStorage_>
class CompressedSparseMatrix : public Matrix<Value_, Index_> {
public:
    CompressedSparseMatrix(
        const Index_ nrow,
        const Index_ ncol,
        ValueStorage_ values,
        IndexStorage_ indices,
        PointerStorage_ pointers,
        const bool csr,
        const CompressedSparseMatrixOptions& options
    ) : 
        my_nrow(nrow),
        my_ncol(ncol),
        my_values(std::move(values)),
        my_indices(std::move(indices)),
        my_pointers(std::move(pointers)),
        my_csr(csr)
    {
        if (options.check) {
            const auto nnzero = my_values.size();
            if (!safe_non_negative_equal(nnzero, my_indices.size())) {
                throw std::runtime_error("'my_values' and 'my_indices' should be of the same length");
            }
 
            const auto npointers = my_pointers.size(); 
            const auto check_pointers = [&](const auto dim) {
                // subtracting 1 from npointers (once we know it's >= 1) instead of adding 1 to dim, as the latter might overflow.
                return npointers >= 1 && safe_non_negative_equal(npointers - 1, dim);
            };
            if (my_csr) {
                if (!check_pointers(my_nrow)) {
                    throw std::runtime_error("length of 'pointers' should be equal to 'nrow + 1'");
                }
            } else {
                if (!check_pointers(my_ncol)){
                    throw std::runtime_error("length of 'pointers' should be equal to 'ncols + 1'");
                }
            }
 
            if (my_pointers[0] != 0) {
                throw std::runtime_error("first element of 'pointers' should be zero");
            }
            const auto last = my_pointers[npointers - 1]; // don't use back() as this is not guaranteed to be available for arbitrary PointerStorage_.
            if (!safe_non_negative_equal(nnzero, last)) {
                throw std::runtime_error("last element of 'pointers' should be equal to length of 'indices'");
            }
 
            const ElementType<IndexStorage_> max_index = (my_csr ? my_ncol : my_nrow);
            for (I<decltype(npointers)> i = 1; i < npointers; ++i) {
                const auto start = my_pointers[i - 1], end = my_pointers[i];
                if (start == end) {
                    continue;
                } else if (end < start) {
                    throw std::runtime_error("'pointers' should be in non-decreasing order");
                }
 
                for (auto x = start; x < end; ++x) {
                    if (my_indices[x] < 0 || my_indices[x] >= max_index) {
                        throw std::runtime_error("'indices' should contain non-negative integers less than the number of " + (my_csr ? std::string("columns") : std::string("rows")));
                    }
                }
 
                for (I<decltype(start)> j = start + 1; j < end; ++j) { // +1 is safe as we know start < end.
                    if (my_indices[j] <= my_indices[j - 1]) {
                        throw std::runtime_error("'indices' should be strictly increasing within each " + (my_csr ? std::string("row") : std::string("column")));
                    }
                }
            }
        }
    }
 
    // For back-compatibility only
    CompressedSparseMatrix(Index_ nrow, Index_ ncol, ValueStorage_ values, IndexStorage_ indices, PointerStorage_ pointers, bool csr, bool check = true) :
        CompressedSparseMatrix(
            nrow,
            ncol,
            std::move(values),
            std::move(indices),
            std::move(pointers),
            csr,
            [&]{ 
                CompressedSparseMatrixOptions options;
                options.check = check;
                return options;
            }()
        ) 
    {}
private:
    Index_ my_nrow, my_ncol;
    ValueStorage_ my_values;
    IndexStorage_ my_indices;
    PointerStorage_ my_pointers;
    bool my_csr;
 
public:
    Index_ nrow() const { return my_nrow; }
 
    Index_ ncol() const { return my_ncol; }
 
    bool is_sparse() const { return true; }
 
    double is_sparse_proportion() const { return 1; }
 
    bool prefer_rows() const { return my_csr; }
 
    double prefer_rows_proportion() const { return static_cast<double>(my_csr); }
 
    bool uses_oracle(const bool) const { return false; }
 
    using Matrix<Value_, Index_>::dense_row;
 
    using Matrix<Value_, Index_>::dense_column;
 
    using Matrix<Value_, Index_>::sparse_row;
 
    using Matrix<Value_, Index_>::sparse_column;
 
private:
    Index_ secondary() const {
        if (my_csr) {
            return my_ncol;
        } else {
            return my_nrow;
        }
    }
 
    /*****************************
     ******* Dense myopic ********
     *****************************/
public:
    std::unique_ptr<MyopicDenseExtractor<Value_, Index_> > dense(
        const bool row,
        const Options&
    ) const {
        if (my_csr == row) {
            return std::make_unique<CompressedSparseMatrix_internal::PrimaryMyopicFullDense<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary()
            );
        } else {
            return std::make_unique<CompressedSparseMatrix_internal::SecondaryMyopicFullDense<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary()
            ); 
        }
    }
 
    std::unique_ptr<MyopicDenseExtractor<Value_, Index_> > dense(
        const bool row,
        const Index_ block_start,
        const Index_ block_length,
        const Options&
    ) const {
        if (my_csr == row) {
            return std::make_unique<CompressedSparseMatrix_internal::PrimaryMyopicBlockDense<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), block_start, block_length
            );
        } else {
            return std::make_unique<CompressedSparseMatrix_internal::SecondaryMyopicBlockDense<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), block_start, block_length
            );
        }
    }
 
    std::unique_ptr<MyopicDenseExtractor<Value_, Index_> > dense(
        const bool row,
        VectorPtr<Index_> indices_ptr,
        const Options&
    ) const {
        if (my_csr == row) {
            return std::make_unique<CompressedSparseMatrix_internal::PrimaryMyopicIndexDense<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), std::move(indices_ptr)
            );
        } else {
            return std::make_unique<CompressedSparseMatrix_internal::SecondaryMyopicIndexDense<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), std::move(indices_ptr)
            );
        }
    }
 
    /******************************
     ******* Sparse myopic ********
     ******************************/
public:
    std::unique_ptr<MyopicSparseExtractor<Value_, Index_> > sparse(
        const bool row,
        const Options& opt
    ) const {
        if (my_csr == row) {
            return std::make_unique<CompressedSparseMatrix_internal::PrimaryMyopicFullSparse<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), opt
            );
        } else {
            return std::make_unique<CompressedSparseMatrix_internal::SecondaryMyopicFullSparse<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), opt
            );
        }
    }
 
    std::unique_ptr<MyopicSparseExtractor<Value_, Index_> > sparse(
        const bool row,
        const Index_ block_start,
        const Index_ block_length,
        const Options& opt
    ) const {
        if (my_csr == row) {
            return std::make_unique<CompressedSparseMatrix_internal::PrimaryMyopicBlockSparse<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), block_start, block_length, opt
            );
        } else {
            return std::make_unique<CompressedSparseMatrix_internal::SecondaryMyopicBlockSparse<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), block_start, block_length, opt
            );
        }
    }
 
    std::unique_ptr<MyopicSparseExtractor<Value_, Index_> > sparse(
        const bool row,
        VectorPtr<Index_> indices_ptr,
        const Options& opt
    ) const {
        if (my_csr == row) {
            return std::make_unique<CompressedSparseMatrix_internal::PrimaryMyopicIndexSparse<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), std::move(indices_ptr), opt
            );
        } else {
            return std::make_unique<CompressedSparseMatrix_internal::SecondaryMyopicIndexSparse<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> >(
                my_values, my_indices, my_pointers, secondary(), std::move(indices_ptr), opt
            );
        }
    }
 
    /*******************************
     ******* Dense oracular ********
     *******************************/
public:
    std::unique_ptr<OracularDenseExtractor<Value_, Index_> > dense(
        const bool row,
        std::shared_ptr<const Oracle<Index_> > oracle,
        const Options& opt
    ) const {
        return std::make_unique<PseudoOracularDenseExtractor<Value_, Index_> >(std::move(oracle), dense(row, opt));
    }
 
    std::unique_ptr<OracularDenseExtractor<Value_, Index_> > dense(
        const bool row,
        std::shared_ptr<const Oracle<Index_> > oracle,
        const Index_ block_start,
        const Index_ block_length,
        const Options& opt
    ) const {
        return std::make_unique<PseudoOracularDenseExtractor<Value_, Index_> >(std::move(oracle), dense(row, block_start, block_length, opt));
    }
 
    std::unique_ptr<OracularDenseExtractor<Value_, Index_> > dense(
        const bool row,
        std::shared_ptr<const Oracle<Index_> > oracle,
        VectorPtr<Index_> my_indices_ptr,
        const Options& opt
    ) const {
        return std::make_unique<PseudoOracularDenseExtractor<Value_, Index_> >(std::move(oracle), dense(row, std::move(my_indices_ptr), opt));
    }
 
    /********************************
     ******* Sparse oracular ********
     ********************************/
public:
    std::unique_ptr<OracularSparseExtractor<Value_, Index_> > sparse(
        const bool row,
        std::shared_ptr<const Oracle<Index_> > oracle,
        const Options& opt
    ) const {
        return std::make_unique<PseudoOracularSparseExtractor<Value_, Index_> >(std::move(oracle), sparse(row, opt));
    }
 
    std::unique_ptr<OracularSparseExtractor<Value_, Index_> > sparse(
        const bool row,
        std::shared_ptr<const Oracle<Index_> > oracle,
        const Index_ block_start,
        const Index_ block_length,
        const Options& opt
    ) const {
        return std::make_unique<PseudoOracularSparseExtractor<Value_, Index_> >(std::move(oracle), sparse(row, block_start, block_length, opt));
    }
 
    std::unique_ptr<OracularSparseExtractor<Value_, Index_> > sparse(
        const bool row,
        std::shared_ptr<const Oracle<Index_> > oracle,
        VectorPtr<Index_> my_indices_ptr,
        const Options& opt
    ) const {
        return std::make_unique<PseudoOracularSparseExtractor<Value_, Index_> >(std::move(oracle), sparse(row, std::move(my_indices_ptr), opt));
    }
};
 
template<typename Value_, typename Index_, class ValueStorage_ = std::vector<Value_>, class IndexStorage_ = std::vector<Index_>, class PointerStorage_ = std::vector<std::size_t> >
class CompressedSparseColumnMatrix final : public CompressedSparseMatrix<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> {
public:
    CompressedSparseColumnMatrix(Index_ nrow, Index_ ncol, ValueStorage_ values, IndexStorage_ indices, PointerStorage_ pointers, bool check = true) :
        CompressedSparseMatrix<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_>(nrow, ncol, std::move(values), std::move(indices), std::move(pointers), false, check) {}
};
 
template<typename Value_, typename Index_, class ValueStorage_ = std::vector<Value_>, class IndexStorage_ = std::vector<Index_>, class PointerStorage_ = std::vector<std::size_t> >
class CompressedSparseRowMatrix final : public CompressedSparseMatrix<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_> {
public:
    CompressedSparseRowMatrix(Index_ nrow, Index_ ncol, ValueStorage_ values, IndexStorage_ indices, PointerStorage_ pointers, bool check = true) :
        CompressedSparseMatrix<Value_, Index_, ValueStorage_, IndexStorage_, PointerStorage_>(nrow, ncol, std::move(values), std::move(indices), std::move(pointers), true, check) {}
};
 
}
 
#endif