This PR patches the bug reported in #321. Previously the function pygsti.io.write_empty_protocol_data had the assumptions of qubits and that the readout was done in the standard basis hardcoded in. This lead to unexpected behavior when using models (really processor specs) with non-standard readout or even qudits with standard readout for dim>2. This PR adds in support for both of these things. For qudits this includes support for mismatching dimensions on the subsystems. This currently doesn't support multiple POVMs, and for non-standard POVMs if there were more than one we currently simply take the first in the list.
Let me know if you have any questions or concerns.
This PR patches the bug reported in #321. Previously the function
pygsti.io.write_empty_protocol_data
had the assumptions of qubits and that the readout was done in the standard basis hardcoded in. This lead to unexpected behavior when using models (really processor specs) with non-standard readout or even qudits with standard readout for dim>2. This PR adds in support for both of these things. For qudits this includes support for mismatching dimensions on the subsystems. This currently doesn't support multiple POVMs, and for non-standard POVMs if there were more than one we currently simply take the first in the list.Let me know if you have any questions or concerns.