Changelog

Main updates

• The module is now called qat.fermion (replaces qat.dqs).

• The qchem submodule has been renamed chemistry (qat.fermion.chemistry).

• qat.fermion is fully compatible with PySCF 2.0 version.

• The SpinHamiltonian and FermionHamiltonian classes have been rewritten:

  • FermionHamiltonian now has a to_spin() method to transform it into spin representation.

  • Fermionic algebraic operations are now possible between the fermionic Hamiltonians FermionHamiltonian and ElectronicStructureHamiltonian.

  • Wick ordering and subsequent simplifications are now automatic for fermionic Hamiltonians. Upon defining a FermionHamiltonian, its terms will be automatically normally ordered and simplified if possible.

  • It is now possible to cast a FermionHamiltonian to an ElectronicStructureHamiltonian, using the to_electronic() method.

  • It is now possible to cast a ElectronicStructureHamiltonian to an FermionHamiltonian, using the to_fermion() method.

• Many improvements have been made to the chemistry module:

  • the class MolecularHamiltonian has been added. It allows for easier basis changes

  and active space selection. It also allows an easier transformation into ElectronicStructureHamiltonian using get_electronic_hamiltonian().

  • the class MoleculeInfo has been added to help with atomic and molecular

  computations. It allows for an easier restriction of the active space of the studied system using restrict_active_space() method, which changes dynamically the relevant MoleculeInfo attributes.

A set of new plugins is now available:

• AdaptVQEPlugin: This plugin implements the ADAPT-VQE algorithm, to efficiently build an ansatz from a pool of operators,

• GradientDescentOptimizer: Allows for natural gradient descent-based optimizations,

• SeqOptim: An implementation of the quantum-classical hybrid sequential minimal optimization, method introduced by Nakanashi et. al. has been added,

• ZeroNoiseExtrapolator: A plugin helping with multiqubit gate noise mitigation,

• MultipleLaunchesAnalyzer: A plugin allowing multiple optimizations for various random initial parameters at once.

Miscellanous

• qat.dqs.fermionic_util functions have now been included in qat.fermion.util.

  • the function exact_eigen_energies has been deleted. Use any eigenvalue solver on the Hamiltonian matrix directly. You can

  generate the matrix using get_matrix() method.

  • the function fermionic_hamiltonian_exponential has been deleted. Compute the exponential of the matrix obtained via

  get_matrix() directly.

• qat.dqs.impurity models have been relocated in qat.fermion.hamiltonians.

• qat.dqs.ansatz_circuits has been renamed qat.fermion.circuits.

• qat.dqs.qchem.ucc has been renamed qat.fermion.chemistry.ucc:

  • build_ucc_ansatz has been deprecated. It has been relocated to qat.fermion.chemistry.ucc_deprecated, but the method

  construct_ucc_ansatz() in qat.fermion.chemistry.ucc should be used instead. The underlying lower level methods have been clarified and can be used as well.

  • compute_active_space_integrals has been split into _compute_active_space_constant and compute_active_space_integrals.

  • build_cluster_ops is now a private method.

  • select_active_orbitals default threshold values have been updated.

  • init_uccsd is now a private method, and returns only the trial Mollet-Plesset 2nd order guess initial parameter.

  • The get_initial_params_and_cluster_ops function has been split into 3 separate functions for better clarity and maintenability:

    • guess_init_params to get the initial parameter guess computed via 2nd order Mollet-Plesset perturbation theory,

    • get_hf_ket to get the Hartree-Fock state,

    • get_cluster_ops to get the cluster operator list.

• select_excitation_operators and get_cluster_ops have been updated. The deprecated versions are located in qat.fermion.chemistry.ucc_deprecated.

• get_active_space_hamiltonian has been deprecated. It can still be found in qat.fermion.chemistry.ucc_deprecated.