Array operation; Fortran equivalent of (:) - c++

Question

I am converting from Fortran to C++ . In Fortran on a 2D array ie A(1000,3) i can do the following:
A(1,:)=(x0,y0,z0)
I have searched online and haven't found anything close to that for C++ so I can vectorize some operations. Any help would be appreciated.

Answer 1

The C++ language does not include a built-in matrix type, but includes all the tools you need to build one. Fortunately, that does not mean you need to go and build one all by yourself. Indeed, that would be a bit like reinventing the wheel.
I suggest you rely on one of the existing C++ open source matrix/vector packages, rather than defining ad hoc types manually. One possibility consists of using the Armadillo library.
Armadillo Wikipedia article
Armadillo online documentation
What you want to do can be done like this:
A.row(0) = rowvec{x0, y0, z0};
Example of self-contained source code below:
#include <armadillo>
#include <iostream>
using arma::Mat;
using arma::vec;
using arma::rowvec;
using arma::colvec;
int main(int argc, const char* argv[])
{
Mat<double> A = Mat<double>(1000, 3, arma::fill::zeros);
std::cout << "Shape of matrix A: " <<
A.n_rows << "x" << A.n_cols << std::endl;
double x0 = 1.5;
double y0 = 2.5;
double z0 = 3.5;
A.row(0) = rowvec{x0, y0, z0}; // HERE
std::cout << " A(0,0)=" << A(0,0) << " A(0,1)=" << A(0,1)
<< " A(0,2)=" << A(0,2) << std::endl;
std::cout << " A(1,0)=" << A(1,0) << " A(1,1)=" << A(1,1)
<< " A(1,2)=" << A(1,2) << std::endl;
return 0;
}
The specialized libraries go much further than this. For example, you can write a matrix product just like P=A*B, and the library will gracefully call the appropriate highly optimized BLAS/LAPACK backend routines for you. You can remain blissfully ignorant of the sad facts that the backend is Fortran-based and thus insists on numbering array elements from 1 not 0, and also insists on storing 2-dimensional arrays in contiguous columns, unlike C/C++ which prefers contiguous rows.
Of course it will take some time to learn about one the existing libraries, but it is unfortunately possible that writing/debugging your own type system will take even more time.
Alternatively, if you insist on designing/using your own C++ linear algebra package, you could do worse than getting a copy of this book:
"The C++ Programming Language, 4th edition"
by Bjarne Stroustrup, ISBN 978-0321-563842.
As you probably know, the C++ initial design effort, and a good chunk of the subsequent language evolution was led by this author. The 4th edition of his book includes a chapter 29 (taking only 30 pages) whose title is: "A matrix design". Chapter 29 makes extensive use of "advanced" features of C++11 such as template types and operator overloading.
Regarding elementary C++ types:
Using an array of std::arrays assume that you know the size of your matrix at compile time. This is a restriction that was fortunately removed from modern versions of Fortran.
Using a vector of vector would allocate the memory space of the matrix as N separate areas, and that would be both less time-efficient and incompatible with BLAS/LAPACK backend libraries.

Answer 2

If using std::array containers and using std::array::operator= you could do this:
std::array<std::array<int, 3>, 1000> A;
int x0, y0, z0;
A[0] = std::array<int, 3>{x0, y0, z0};