Hi, I have a question regarding the bosonic sitetype, and the cutoff site dimension. Is there any way to create sites with site dependent cutoffs, i.e. for a chain of 5 sites, something like sites = siteinds("Boson",5; dim=[5,4,3,2,1]) as opposed to sites = siteinds("Boson",5; dim=5). I’m trying to reproduce some results from https://journals.aps.org/prl/pdf/10.1103/PhysRevLett.123.090402 which uses MPS techniques (T-TEDOPA) and they use a site dependent cutoff to handle the high temperature, see the definition of d'_{n} near the end of the first column of the 4th page.
Good question. Here’s the recommended way to do it:
dims=[5,4,3,2]
sites = [siteind("Boson"; dim=d) for d in dims]
That is, there is a lesser-known function siteind that just makes one site Index as opposed to the siteinds function which makes a whole array of them.
At an even more low level, you can just make an Index with the same properties by doing:
Index(5,"Boson,Site")
to make a 5-dimensional Boson site index. The key thing here, besides getting the dimension to be the one you want is that the "Site" and "Boson" tags can be quite important for allowing our op and OpSum systems to recognize this Index as certain kind of physical index (for making operators) and for treating the indices specially in certain algorithms.
But we do ultimately recommend the siteind and siteinds functions since they will ensure these tags are correct and also offer the choice to turn on quantum numbers if you want to conserve those.