[Bug] MultivariateStrategy.SERIES has different behaviors when passed to circuit or directly to QNN

[n-toscano]

Short description

When creating a FM that takes more than one variable as input, we need the two encodings not to commute with each other. If the config is passed directly to QNN.from_configs() then an entangling layer is placed in between, and all works. It's not the case if one passes it to the QuantumCircuit.

What is the expected result?

No response

What is the actual result?

No response

Steps/Code to reproduce

from qadence import *

n_qubits = 4
fm_config = FeatureMapConfig(
    num_features=2,
    inputs = ["x", "y"],
    multivariate_strategy= MultivariateStrategy.SERIES,
)
observable_config = ObservableConfig(detuning=Z)
qnn = QNN.from_configs(
    register=n_qubits,
    fm_config=fm_config,
    obs_config=observable_config
)

vs

fm = create_fm_blocks(register=n_qubits, config=fm_config)
circuit = QuantumCircuit(support=n_qubits, fm)

Tracebacks (optional)

No response

Environment details (optional)

No response

Would you like to work on this issue?

Yes

[mlahariya]

To review.