CODEWITHSUNDEEP
c++eigenstdvector
Poia
Poia

Reputation: 87

C++: how to convert std::vector to Eigen::MatrixXd?

I have std::vector<double> param with (n + n * n) length and I need to move last (n * n) elements to MatrixXd. I did it in this way:

MatrixXd cov(n, n);
for(int i = 0 ; i < n ; ++i) {
    for(int c = 0 ; c < n ; ++c) {
        cov(i, c) = param(n + i * n + c);
    }
}

Is there better way to do this?

Edit: better means faster ;)

Upvotes: 4

Views: 9795

Answers (2)

Avi Ginsburg
Avi Ginsburg

Reputation: 10596

Roger's answer is good, with the exception that if you want to utilize vectorization, the Eigen::Map doesn't know if it's aligned or not, thus no vectorization. If you want that, you need to make a copy of the data and not just map to it.

Eigen::MatrixXd copiedMatrix = Eigen::Map<Eigen::MatrixXd>(&param[n], n, n);

Upvotes: 7

Roger Figueroa Quintero
Roger Figueroa Quintero

Reputation: 368

If you guarantee that the vector stores its elements contiguously, the simplest and fastest way is the following:

std::vector<double> param;
//code//
.
.
.
//code//
double *v = &param[n];
Eigen::Map<Eigen::MatrixXd> matrix(v,n + n * n,1);

In this way the memory of param is reused.

Upvotes: 3

Related Questions