Evaluation backends

[mateuszkupper]

Creates an interface with backends for diagram evaluation:

• removes the need to double, or convert the channel diagrams manually in order to evaluate them
• diagrams are evaluated like in the example below:

from optyx.core.backends import PercevalBackend

circuit = BS @ BS >> qmode @ MZI(0.4, 0.34) @ qmode
result = circuit.eval(PercevalBackend()) #returns an `EvalResult` object

• we can obtain for example a probability distribution from the object: result.prob_dist() which returns a dictionary with photon number configurations and corresponding probabilities
• the available backends are: QuimbBackend (exact and approximate using hyperoptimisers), DiscopyBackend, PercevalBackend (only for pure states)

More examples:

from optyx.core.backends import QuimbBackend
from cotengra import HyperCompressedOptimizer

d.eval(QuimbBackend(hyperoptimiser=HyperCompressedOptimizer())).prob_dist(round_digits=5)

from optyx.core.backends import PercevalBackend

circuit = BS @ BS >> qmode @ MZI(0.4, 0.34) @ qmode

result = circuit.eval(PercevalBackend())