v2.7.2

In this release we:

• Improved krylov energy minimization used by the DMRG solver in emu-mps. It should now converge more reliably with a lower max_krylov_dim

v2.7.1

In this version we:

• Fix an issue where the energy minimization for the DMRG solver in emu-mps failed to converge
• Fix an issue where Energy, EnergySecondMoment and EnergyVariance observables errored in emu-sv at t=0
• Fix an issue where Energy, EnergySecondMoment and EnergyVariance returned finite imaginary part in emu-mps at t=0 when Lindblad noise was present

v2.7.0

In this release we fixed for both emulators

• fix a bug where log-level behaved as if it was WARNING if INFO or DEBUG was chosen, suppressing a lot of output
• support for prefer_device_noise_model from EmulationConfig
• we support measurement of observables at t=0 (to help conform to QutipBackendV2)
• The dt parameter in the configs can now be any float rather than int
• support for pulser 1.7 including :
  • support for the n_trajectories config that was added to EmulationConfig

v2.6.1

• Documentation updated
• SparseOperator class in emu-sv
• bug fix: broken plot link in emu_mps: Errors in TDVO

v2.6.0

Documentation updated
SparseOperator class in emu-sv

v2.5.2

Fixed an issue in emu-base, where local channels were incorrectly processed when only a subset of qubits received a laser pulse.

Previously, the extraction of omega, delta, and phi assumed that all qubits were present in the Pulser samples, which caused an error when dealing with local channels experiments.

The fix ensures that missing qubits are cleanly ignored in the computation of omega, delta and phi.

v2.5.1

Fixed a bug in emu-mps and emu-sv where State objects did not serialize correctly, because a check that the state had not been changed since creation failed on the pulser side due to different tolerances for state normalization.

v2.5.0

Leakage noise is supported in emu-mps. This means that when leakage is turned on in pulser.NoiseModel emu-mps will support 3 level systems and 3 by 3 jump operators. The reference basis in this situation is the following ("g","r","x"), where "x" is the leakage state.

v2.4.4

Made optimize_qubit_ordering default to True in MPSConfig. Rather than having MPSConfig error when it is True in the presence of Fidelity, Expectation or StateResult observables, it will now issue a warning that these observables are not supported, and set the value to False

v2.4.3

In emu-mps, when emulating effective noise, the state norm is now restored after each truncation. When the maximum bond dimension for the simulation is hit, truncations can cause a significant norm decay. Since the state norm is used to determine the jump times when doing Monte Carlo, this can cause the simulation to see erroneously high noise rates.