Randomized MPO-MPS contraction

Papers
1

@miles @rittermarc

This preprint looks very interesting. See Fig. 1 for a comparison with zip-up and fit algorithms.

2 Likes
2

Thanks, Hiroshi. I think this method will have a big impact and the comparisons to other methods are very nice to see. Iā€™m doing some work right now using similar techniques, partly inspired by this paper.

3

Thank you for the information. For my QTT applications, D = chi = tilde chi leads to chi^4 scaling (3.3a), which is the same as the fit and zip-up algorithms. Nevertheless, the SCR algorithm could be advantageous over the zip-up algorithm in terms of speed and accuracy!