Johnnie Gray

Hi @sandner-art, thanks for the thorough issue. Yes `inplace` is simply not a valid kwarg for `contract_ind` at the moment (not all methods support it). It could do with a...

Thanks for the issue @garrison, yes this would definitely be nice to handle. Does creating a size 1 bond make more sense that leaving it disconnected? I think probably yes...

I've added a `create_bond` kwarg in https://github.com/jcmgray/quimb/commit/463b825b1d76675910a133d039951cffadbdb81c to all the canonize/compress sweeping methods. By default it is off, for explicitness.

Hi @talentkeychen, yes I *think* this should work just as-is, supplying the non-hermitian terms. The local evolution in the `TEBD` class is generated now as `np.linalg.expm(real_or_imag_dt * hij)`, where hij...

I did some brief numerical tests of the scaling over in https://github.com/kahypar/kahypar/issues/141 in case they are of interest.

Hi @lin-cp, thanks for the interesting example. For what its worth I get a much more minor performance difference (4.5 vs 6.0 seconds), but worse nonetheless. If you check the...

The intermediate step is as listed above: - "jmik,ijkl->jmil" > Is there any empirical rule to judge if using opt_einsum will help to save time? 1. Generally the speedup listed...

Yes in my experience, and from their results, optimizing the combine function (i.e. sample `sum`, `max`, `min` for combining size_a and size_b) is subsumed by optimizing a `costmod` coefficient like:...

Thanks for the issue @projekter! I think simply passing on the `output_inds` would not quite be sufficient here, since the local output indices for a slice of sites will not...

> Can we add memory_limit support? I think one possible option here, that would make life easier, would be to simply write a separate function that that runs after the...