eigen3.cpp
#include <iostream>
#include <Eigen/Dense> // must include Eigen before <h5cpp/core>
#include <h5cpp/all>
using namespace std;
// EIGEN3 templates are unusually complex, let's use our template definitions
template<class T> using Matrix = Eigen::Matrix<T,Eigen::Dynamic,Eigen::Dynamic>;
template<class T> using Colvec = Eigen::Matrix<T,Eigen::Dynamic,1>;
// only EIGEN::DYNAMIC [ARRAY|MATRIX|VECTOR] are supported
// in other words Eigen::Matrix<T,S,S> where S \in unsigned will not work, rather cast static allocation into Dynamic (heap memory) structure
int main(){
{ // CREATE - WRITE
Matrix<short> M(2,3); // create a matrix
h5::fd_t fd = h5::create("linalg.h5",H5F_ACC_TRUNC); // and a file
h5::ds_t ds = h5::create<short>(fd,"create then write"
,h5::current_dims{10,20}
,h5::max_dims{10,H5S_UNLIMITED}
,h5::chunk{2,3} | h5::fill_value<short>{3} | h5::gzip{9}
);
h5::write( ds, M, h5::offset{2,2}, h5::stride{1,3} );
}
{
Colvec<float> V(4); // create a vector
// simple one shot write that computes current dimensions and saves matrix
h5::write( "linalg.h5", "one shot create write", V);
// what if you want to position a matrix inside a higher dimension with some added complexity?
h5::write( "linalg.h5", "arma vec inside matrix", V // object contains 'count' and rank being written
,h5::current_dims{40,50} // control file_space directly where you want to place vector
,h5::offset{5,0} // when no explicit current dimension given current dimension := offset .+ object_dim .* stride (hadamard product)
,h5::stride{4,4}, h5::block{3,3}
,h5::max_dims{40,H5S_UNLIMITED} // wouldn't it be nice to have unlimited dimension? if no explicit chunk is set, then the object dimension
// is used as unit chunk
);
}
{ // CREATE - READ: we're reading back the dataset created in the very first step
// note that data is only data, can be reshaped, cast to any format and content be modified through filtering
auto fd = h5::open("linalg.h5", H5F_ACC_RDWR, // you can have multiple fd open with H5F_ACC_RDONLY, but single write
h5::fclose_degree_strong | h5::sec2); // and able to set various properties
}
{ // READ:
Matrix<short> M = h5::read<Matrix<short>>("linalg.h5","create then write"); // read entire dataset back with a single read
}
}