TeNPy Forum

Currently, the swapping for TEBD is not supported, although it should not be too difficult to implement, following the way you have outlined. There are a few things to be careful about, though: You need to adjust the existing model code to get the exponentials of the NNN terms. avoid onsite terms ap...

What exactly do you mean by computing the entanglement eigenvectors? The eigenvectors of which matrix? For a finite system, I could imagine to look at individual Schmidt vectors (=singular vectors) on the left or right of a given cut in the form of an MPS. You can obtain them by just the part of the...

Good question, thanks for pointing that out. The dominant eigenvector with eigenvalue 1 should always be in 0 charge sector: in canonical form the dominant eigenvector is just the identity, which has no charge. But I agree that it is not clear a priori (at least not to me) in which sector the next t...

Short answer : Try measuring correlation functions, or expectation values of |M| or M^2 instead of M=\sum_i X_i ;) Reason : You try to observe spontaneous symmetry breaking. By definition, spontaneous symmetry meaking means that you have a degenerate ground state (in the thermodynamic limit): the s...

In iDMRG, the iMPS unit cell is the central part of a growing environment on the left and right. The idea of the Engine.init_env() is to re-use the environment (and state) from a previous run. If the state has a sufficiently large correlation length, the environmnet "pins" DMRG to find the...

Another serious bug, this time in the models.
The problem is/was that adding the hermitian conjugate with was not done correctly.
Further details at issue #23.

If you have implemented your own model, check that you do it correctly in your case

I found a serious bug in npc.inner . This bug could result in wrong expecation values and Lanczos/DMRG giving completely wrong results. Luckily, the bug was only on the github master branch for two weeks (since March 02), so I hope it didn't cause too many problems yet. In case you did a "git p...

Sorry, that's indeed true. The bug only appeared if you use a mixer and have a model without onsite terms.
Should be fine now

PS: You can also report problems like that as github issues.

I recently changed the order of axes in DMRG from (vR.p1) to (p1.vR) to avoid unnecessary transposes. However, the newest version should consistently use "(p1.vR)". Did you actually store some MPS (e.g. with pickle) and load it again? Or do you restart the whole simulation/initialization o...

I just noticed an error in my last post above, see the edit. I hope that was clear anyways...

Thanks for pointing that out.
Luckily, both problems were just a problem of the Lattice.plot_coupling(ax) function, and not the actual couplings of the model.
I just fixed the bug in the most recent version on github.

The np_conserved.expm() expects a np_conserved Array, not a numpy array or string (it doesn't know anything about what you might mean with "Sz"). Therefore, you should get the Sz-operator as an np_conserved.Array, preferably out of your "Site" classes. Assuming that you have a un...

Oh, sorry, I had the Pauli Sz in mind, and the factor 1/2 of course matters. My bad

Yes, as Leon pointed out, you need to modify the prefactor for the hopping constants, such that an electron picks up a phase when going around the cylinder. This can be done by providing a (numpy) array instead of just a scalar as parameter. Another very important thing here is to "reuse" ...

Since exponentiating a matrix is often needed for tensor networks, the scipy.linalg.expm function is also implemented for np_conserved Arrays . When you talk about string order parameters, do you really want to calculate that? What you wrote suggests that you try to calculate something like \langle ...

The "charges" you get are the "quantum numbers" you conserved. Of course, it depends on the model which "quantum numbers" you can actually preserve. For example, for a spin-model, you can preserve the Magnetization (e.g. Heisenberg XXZ chain), or just the parity of the ...

The general setup looks fine. Regarding the bond dimension, you need to carefully check if larger \chi leads to different results. I would be very surprised if 30 is enough for convergence. To check that it works you need to compare results you have with existing results in the literature ;) Finite ...

The correct permutation depends on the lattice you choose, and possibly on the "order" argument of that lattice. Yes, you should simply call it like psi.compute_K(perm=M.lat) # where M is your model to automatically figure out the "correct" permuation. This permutation will corre...

Sorry for not coming back to this earlier. The real question is: what do you mean with "random" when talking about an MPS? When you draw a "random" state uniformly from the Hilbert space, it has a volume law entanglement; in fact it is almost maximally entangled! Therefore, it wo...

What kind of "gate" do you have in mind? The usual TEBD where the "gates" are U=exp(i H_bond dt) is implemented in TeNPy.

On general grounds: DMRG is a variational method which looks for the best approximation of the ground state in the space of MPS with a given maximal bond dimension (assuming you put a "chi_max" in the DMRG_parameters). This class of MPS can only represent states with a entanglement entropy...

The problem was just that pickle needs the files to be opened in "binary" mode since Python 3, and the example stems from a time where TeNPy was still running with Python 2
Anyways, I fixed it (and some other issue it had), it should now work again in the newest version.

psi.expectation_value("Sx") tries to evaluate the expectation value \langle S^x_i\rangle on each site i of the MPS. If you called psi.group_sites(k) beforehand, the local sites i refer to GroupedSite , each containing k of the previous sites, and the "Sx" is no longer defined (o...

That's the point: if your result is not unique (due to the degeneracy), any (small) change of the setup can change the result - it's basically random, what you get. Your results should usually still be reproducable, when you run the same code on the same machine with the same python version, but cha...

The problem you have met comes from the degeneracy of the states - you have a bunch of correct, possible solutions, and which one is returned depends on numerical round off errors. I don't get the comment on truncation. Let me clarify: The truncation respects all the parameters chi_max, svd_min and ...