[0.7.0] - 2020-09-04

Release Notes

The big new feature is the implementation of the W_I and W_II method for approximating exponentials of an MPO with an MPO, and MPS compression / MPO application to an MPS, to allow time evolution with ExpMPOEvolution().

Changelog

Backwards incompatible changes

• Remove argument leg0 from build_MPO.

• Remove argument leg0 from from_grids, instead optionally give all legs as argument.

• Moved/renamed the module tenpy.algorithms.mps_sweeps to tenpy.algorithms.mps_common. The old mps_sweeps still exists for compatibility, but raises a warning upon import.

• Moved/renamed the module tenpy.algorithms.purification_tebd to tenpy.algorithms.purification (for the PurificaitonTEBD and PurificationTEBD2) and tenpy.algorithms.disentangler (for the disentanglers).

Added

• VariationalCompression and VariationalApplyMPO for variational compression

• PurificationApplyMPO and PurificationTwoSiteU for variational compression with purifications.

• Argument insert_all_id for tenpy.networks.mpo.MPOGraph.from_terms() and from_term_list()

• implemented the IrregularLattice.

• extended user guide on lattices, Details on the lattice geometry.

• Function to approximate a decaying function by a sum of exponentials.

• spatial_inversion() to perform an explicit spatial inversion of the MPS.

Changed

• By default, for an usual MPO define IdL and IdR on all bonds. This can generate “dead ends” in the MPO graph of finite systems, but it is useful for the make_WI/make_WII for MPO-exponentiation.

• tenpy.models.lattice.Lattice.plot_basis() now allows to shade the unit cell and shift the origin of the plotted basis.

• Don’t use bc_shift in tenpy.models.lattice.Lattice.plot_couplings() any more - it lead to confusing figures. Instead, the new keyword wrap=True allows to directly connect all sites. This is done to avoid confusing in combination with plot_bc_identified().

• Error handling of non-zero qtotal for TransferMatrix.

Fixed

• Removed double counting of chemical potential terms in the BosonicHaldaneModel and FermionicHaldaneModel.

• Wrong results of tenpy.networks.mps.MPS.get_total_charge() with only_physical_legs=True.

• tenpy.models.lattice.Lattice.plot_bc_identified() had a sign error for the bc_shift.

• calc_H_MPO_from_bond() didn’t work for charges with blocks > 1.

• TEBD: keep qtotal of the B tensors constant

• order model parameter was read out but not used in tenpy.models.model.CouplingMPOModel.init_lattice() for 1D lattices.