Hello,

Basically I am trying to develop some iterative algorithms that involve matrix with varying size, for instance, something like

Code: Select all

j=1; A(100,100)= known;
P= randn(100, 1);
Q= randn(100, 1);
for iter = 1:100
     Q(:, j) = A* P(:,j);
   
    // update j, P, and therefore Q

end

Here, P and Q are initially n*1 matrix, and they grow during the iteration, so the column size is dynamic, I would like to know the best practice to use Eigen for this case, it seems to me I have to use "resize" every time the column size has been increased, is there any better solution ?

if you know a reasonable bound m for j, better allocate n*m matrices P and Q once.
Otherwise you can use conservativeResize(NoChange,alloc_size) and double alloc_size whenever alloc_size<=j. Increasing it one by one will be rather slow.