TeNPy Forum

MPOs are extensive sums of local terms; thus they take a Jordan block form W = \begin{pmatrix} 1 & C & D \\ 0 & A & B \\ 0 & 0 & 1 \end{pmatrix} = \text{e.g. for transverse field ising} \begin{pmatrix} 1 & X & - Z \\ 0 & 0 & X \\ 0 & 0 & 1 \end{pmatrix...

Why? What do you want to do with the state afterwards? Yes, of course, you can just insert a projector onto the index with maximal singualar value at the boundary, i.e. bond dimension 1, and formally declare it a finite MPS. But this truncation is likely not the "optimal" state of maximal ...

I'm sorry that this wasn't clear, but TeNPy's TDVP isn't intended to be used with segment boundary conditions; as of now, TeNPy only supports segment MPS for DMRG. There, the idea is to just keep the state outside of the segment (or "window" in the literature) fixed during the optimization...

Since you apply 6 steps of randomizing, it might still have some "memory" of the initial state. This function takes an initlal MPS, I guess you take an initial product state. Did you try using another one? It's not completely clear to me what initial states the authors of 1208.2623 took, j...

For two dimensional systems, there can be separate boundary conditions in x and y, hence the bc_x and bc_y, which really are just the two values for the bc=[bc_x, bc_y] of the lattice.
So yes, for a 1D lattice, just putting bc='periodic' is the correct thing to get periodic boundary conditions.

See the reply in the separate thread

Welcome to TenPy, happy new year, and sorry for the long delay in the answer, I had a few things I needed to finish before Christmas..... I think the issue here is that the mixer is disabled, just try setting 'mixer': True . The reason why you need the mixer is that you always just optimize two site...

Hi Pablo, I hope you also had some nice holidays and wish you a happy new year! The log files should report the "norm_error" - is it close to zero in your case? You can also explicitly check the output of norm_test to check whether the state is in in canonical form. (To be clear, psi.test_...

Question 1: It's a dense matrix representation of the N (=particle number) operator. If you truncate to nmax=10, you have 11 local basis states |n>, where n=0,...,10 is the boson number. Hence, N is 11x11 matrix, with n on the diagonal. Question 2/3/4: The cutoff nmax for the Site is the maximal bos...

It's a bit unconvenient: you need to override the ordering method of the lattice to support that.
I've quickly pushed an update 087beba that does that.
You can do similar things for other lattice classes as well.

To be honest, I'm quite surprised by 2102.10188 that it actually seems to work with MPS. They simulate eq. 2, with all-to-all couplings H=\sum_{i<j} S_i \cdot S_j . Naively, I would have expected much more entanglement than only ~log(N) to build up with the time evolution. On the other hand, it kind...

Welcome! It looks like conda is only looking in the win-32 channel, i.e. for a 32-bit package. However, we've only packaged/compiled a 64-bit version, since virtually all hardware (on which you'd like to run TeNPy) these days has 64-bit CPUs. Is your Windows 64 bit? Then you should also download the...

I'm not too surprised that W_II fails for this. Intuitively, the W_I/W_II methods from the ExpMPOEvolution approximates e^{-iH dt} by taking into account non-overlapping terms of H; it is not designed for a long-range coupling going accross the whole system (which you get by using periodic BC for a ...

Indeed, the tenpy.simulations.Simulation classes are built to support exactly this in a generic way for the various algorithms. The idea is that you start your simulation with `run_simulation`, it saves snapshot at certain checkpoints (e.g. during DMRG between sweeps), and you can resume from these ...

1) You refer to H = ZZ - g_x X, where the spin flip symmetry is a product(X). TeNPy implements the TF Ising model as H = XX - g Z, such that the symmetry becomes product(Z), i.e. the parity of sum(Z) is conserved. This is implemented in tenpy.models.tf_ising.TFIChain , it will by default use conserv...

Just initializing the two TDVP environments independently at the same time will not yield the correct result, since each engine has an MPOEnvironment for the partial contractions of <psi|H|psi>, and the one of engine1 will not get updated during the evolution of engine2. You could call engine1.envir...

You chose an artificial test case of completely decoupled spins; the spin pairs 2-4 and 3-5 will just each propagate together. This is basically a product state, except that 2-4 and 3-5 are entangled. After a single time step with the ExpMPOEvolution, the MPS bond dimension grew to include the Schmi...

Take a look at the irregular lattices; exactly this example can be found in the user guide.
Since the commit b49f0e0 that I just pushed, you can do that by adding the model parameter "irregular_remove".

I'm not sure if I understand what you're doing with which boundary conditions. Indeed, "segment" boundary conditions yield a finite DMRG, because the "segment" is a finite region extracted from an infinite MPS, and you only optimize that region keeping the environments fixed. The...

Whatch out! The model parameters bc_MPS, bc_x, bc_y are only used if you don't specify the lattice instance directly, since the lattice instance already has the boundary conditions! model_params = { "S": 0.5, "conserve": "Sz", "lattice": 'Honeycomb', "Lx&...

What you see here is a symmetry breaking transition; I think it's physic and not a bug in TeNPy. For large Jz, the ground state is an anti-ferromagnet forming an Neel state |up down up down ...> , but by symmetry (product of X on each site), the |down up down up ...> has the same energy. Which one i...

Since I don't own any Apple products, I can hardly test this and have no idea what's required to get it to work. However, the kind folks from conda-forge apparently have a working CI bot system that should help us: see the conda-forge docs If you want support for the M1, you'll have to get active: s...

I think you should just look at the Transfermatrix with the smallest possible number of sites. This would be 2 sites for a Neel state, 1 for a completely translation invariant state. Sadly, we don't have an algorithm for the last case at the moment -even single site DMRG requires a two-site MPS unit...

Hi Pablo, which TeNPy function are you referring to? TransferMatrix. eigenvectors () returns the eigenvalues \lambda of the full transfer matrix, there's no ambiguity there. correlation_length returns the xi in your definition, to make sure it's independent of L; it's basically just changing the uni...

You can use algorithm_params: diag_method: lanczos lanczos_params: E_shift: -10 # choose small enough such that E + E_shift < 0 (also for excitations!) It basically does the same trick as you suggested - it just replaces H -> H + E_shift[inline], but the [inline]E_shift is automatically subtracted a...