Hi Pavel, mostly it will be addition and multiplication of the entries, together with some operations involving the indices. This is still in the design stages, so I will know more in the incoming weeks.
Thank you for your idea. Perhaps I will have to go with a unified wrapper class, then try different underlying structures to see which one works best. Thank you very much for your suggestions. best regards Janek Kozicki Pavel Holoborodko said: (by the date of Thu, 28 Sep 2023 20:21:24 +0900) > What operations do you need to do with such tensors? > > You can implement the simplest way to store the sparse tensor - by storing > the coordinates of the non-zero values + values itself. > Then implement tensor operations for such a storage format. > > If density of non-zeros is very low (as you say), then this simple data > structure will be efficient and simple. > > > On Wed, Sep 27, 2023 at 9:21 PM Janek Kozicki (yade) < > [email protected]> wrote: > > > Hi, > > > > in my incoming work on implementing Loop Quantum Gravity I will need > > a 15-dimensional sparse matrix (or tensor) to store the quantum > > gravity data in a 15-dimensional basis. The data is very sparse, most > > of it are zeros. But if I tried to use a regular vector<vector<....>> 15 > > times > > I will quickly run out of memory. > > > > > > So, does libeigen offer such data type? > > > > best regards > > Janek Kozicki -- Janek Kozicki, PhD. DSc. Arch. Assoc. Prof. Gdansk University of Technology (Gdansk Tech) Faculty of Applied Physics and Mathematics Institute of Physics and Applied Computer Science Division of Theoretical Physics and Quantum Information -- http://yade-dem.org/ http://pg.edu.pl/p/jan-kozicki-19725 http://mostwiedzy.pl/en/jan-kozicki,19725-1
