The QuantMark library can be used to analyze VQE-algorithms. The functionalities of the library buildon top of tequila.

Currently this library supports only python 3.7, 3.8 and 3.9

Before installing the library it is recommended to install psi4 if you want to use molecules. It is recommended to use conda for easy installation ( instructions).

git clone https://github.com/ohtu2021-kvantti/LibMark.git
cd LibMark
pip install .

You also need to install the quantum backends that you intend to use. In the example below qulacs is used.

pip install qulacs

conda create -n envname python=3.7 # Create environment. (optional python=3.8 or python=3.9)

conda activate envname # Activate to environment.

# install psi4
conda install -c psi4 psi4 # for python 3.7
conda install -c psi4/label/dev psi4 # for python 3.8 or 3.9

# Clone LibMark and move into the created folder.

pip install . # Install quantmark.

pip install qulacs # Install qulacs (and or any other quantum backend).

LibMark is relying stongly on tequila.

import quantmark as qm
import tequila as tq

# Define optimizer.
optimizer = qm.QMOptimizer(module="scipy", method="BFGS")

# Define backend.
backend = qm.QMBackend(backend='qulacs')

# Make tequila molecule object.
molecule = tq.chemistry.Molecule(
    geometry='H 0.0 0.0 0.0\nH 0.0 0.0 1.6',
    basis_set='sto-3g'
)

# Make tequila circuit.
circuit = molecule.make_uccsd_ansatz(trotter_steps=1)

# Run the benchmark.
algorithm = qm.VQEAlgorithm(
    molecule=molecule,
    circuit=circuit,
    optimizer=optimizer,
    backend=backend,
    repetitions=10
)

# Analyze the algorithm.
print(algorithm.analyze())

# Analyze only circuit (Does not run the algorithm).
print(algorithm.analyze_circuit())

# Changing parts between analyzing (can change one or multiple).
algorithm.molecule = new_molecule
algorithm.circuit = new_circuit
algorithm.optimizer = new_optimizer
algorithm.backend = new_backend

print(algorithm.analyze())

• average_history: The average values after minimizing iterations.
• accuracy_history: The average accuracy (compared to target_value) after minimizing iterations.
• qubit_count: The amount of qubits the circuit needs.
• gate_depth: The gate depth of the circuit.
• gate_count:The amount of gates the circuit uses.
• parameter_count: The amount of parameters on the circuit that have to be optimized.
• average_iterations: The average amount of iterations that the minimizing process takes.
• success_rate: The fraction of runs that got a result that is accurate to the FCI value with the accuracy '1 / 627.5094740631'.

You can add an issue or add code with the usual fork(or branch), do, pull request system.

The same as normal installation, but in editable mode.

pip install -e .

Also it is recommended to install the testing and linting libraries.

pip install pytest
pip install flake8

To run tests use the command pytest in the project root folder.

To run linter use the command flake8 in the project root folder. Flake8 does not give an error when using spaces for indentation, but please use tabs.