TeNPy Forum

When you just call the MPS.__init__() function, it will just assume that it is in right-canonical form. In most cases, people will in some way intialize the MPS from a product state with bond dimension 1 (using MPS.from_product_state or MPS.from_lat_product_state), which is trivially in both left- a...

Hi Leandro, apologies for not replying earlier! Just to check: Did this work out, did you manage it in the end? Or do you still need help? I hope you didn't give up! The t-J model is definitely sufficiently important/generic that it would be a good idea to include it into TeNPy itself. Would you lik...

By default, the option `P_tol_to_trunc` causes DMRG to set the P_tol parameter in the lanczos_params, which causes this warning in case DMRG actually does not use lanczos to diagonalize the effective Hamiltonian (it does exact diagonalizaiton for small enough chi where this is actually faster). I've...

1.) The simplest/naive thing is to just use the usual add_coupling with increasingly longer range interaction |i-j| up to some finite cutoff. Alternatively, the more clever/state of the art way would be to fit the powerlaw by a sum of exponentials, and then use `add_exponentially_decaying_interactio...

Yes, that's the general way to go about this: just define your lattice with the correct number of sites per unit sell, and add the corresponding couplings. Some inspiration might be given by the Hofstadter models in TeNPy. I guess it would also be useful to re-implement the FermionicHubbardModel wit...

Hi Nick, not sure if you already solved this. If you want, you can create a diagonal np_conserved tensor from that with the following snippet, but it is numerically inenfficient to actually do that (because tensordot is more expensive than just scaling an axis, which is what we usually do. # assumin...

What error do you get? Can you please include the error traceback? Otherwise I can only guess what the problem might be.... It seems strange that it works for smaller J_z, but not for larger one. By the way, you can actually do this without implementing a custom lattice/model, just based on the tenp...

Yes, I think you need to recreate the environments from scratch.
Not completely sure, but I think you could try using the old ones as an initial guess for the new environments.

The disentanglers implemented in TeNPy act on two or more consecutive sites in the MPS language, but of course you can try to apply a bunch of them one after the other. It also has code that tries to bring a few sites next to each other and then disentangle them, but in my experience that didn't wor...

You just need to import the file in python, which contains the model definition; then you can just use the "name" of the model in the simulation parameters. To import that file with the definition, you have two options: Either you use the command line option -1 of tenpy-run or python -m te...

To be clear, the actual convergence of DMRG is much faster than just 1/(#sweeps performed), as you see on the right - there's a log-scale on the yaxis

is not the energy you want to study for convergence' it is the total energy of the system that contains age sites, including all energy terms in the environments with whatever state you had at that point. Pictorially, the "total" system that infinite DMRG grew after a couple of sweeps look...

A bit late, but better late than never ;-) apply_local_op accepts an operator that acts on n neighboring sites, for example for two sites with legs 'p0', 'p1', 'p0*', 'p1*' . You can use the brand-new function tenpy.networks.site.kron to create the corresponding bond operator from the single-site op...

You shouldn't trust expectation values/correlation functions that were calculated from an MPS that is not in canonical form, because these functions implicitly assume that psi is in canonical form. When your DMRG is really converged and doesn't change the state during the sweep any more, you would e...

The answer comes probably a bit too late, but anyways: How do you want to "fix" the states outside? If you imagine to project them to the local basis states with exactly 0/2 sites, you would have a product state with the rest. The hopping term wouldn't contribute, and you can just calculat...

They will only depend on the charge conservation you're using.

I guess you're aware of this, but just to be clear: when you do this, you write down all 2^L states, so you could just as well use exact diagonalization....

Hi Pablo, there is a way to do it more effiently, although I only did that numerically: I created the AKLT model by just defining bond terms, and then initializing the MPO from these bond terms - there's a function in TeNPy to automatically do the SVD of the bond term to get the MPO. Take a look at ...

It's a problem of the inital state and charge conservation, not the jordan wigner strings. TwoSiteDMRG updates two sites in the subspace created by projecting H to the current psi (i.e. your initial state) on other sites; this projection will render H zero because each term in H has at least two C/C...

Yes, that's right.
I tried to explain this in the section "spinful fermions" of the Jordan-Wigner user guide you referenced.
If you have a suggestion how to formulate this a bit clearer, feel free to submit a corresponding pull request on github!

rubenv Good catch, the fix is of course correct. Let me explain: The variation compression needs a copy of the old psi to compress against, and that copy is made during the class intialization, more precisely in [tenpyclass=tenpy.algorithms.mps_common.VariationalApplyMPO]init_env[/inline]. So it mo...

Whoa, I wasn't aware that you want to run it for non-hermitian H. Lancos implicitly assumes that H is hermitian! For that reason at least TeNPy's implementation of TDVP (which uses Lancos to do the local time evolution) does not work for that case! Dropping the hermiticity of H is in general very da...

Hi Ruben, you're right, the DMRG code supports segment boundary conditions already. Right now, you need to extract the left and right environment from a previous DMRG run, and then specify it in resume_data['init_env_data'] to the subsequent DMRG run by hand. There's still some subtleties, e.g. how ...

Update:
when trying to install TeNPy on a cluster with the Intel Python distribution, we encountered the same runtime warnings.
Looking into this now...

Sorry, I didn't realize you were looking at infinite MPS. For an infinite MPS, the leg between unit cells is non-trivial, and in general changes when you do a time evolution. This is where the error is coming from: when you initialize the MPSEnvironment, it needs to know how to connect the left-/rig...

For single-site DMRG to work you need to enable the "mixer" by setting dmrg_params['mixer']=True , otherwise it cannot grow the bond dimension. Be warned that the default mixer_params might still disable the mixer too early for larger/more correlated systems. I've just released a bugfix ve...