has_mixture always returning False

[Lucaman99]

When has_mixture is called on a circuit containing mixer gates, a combination of unitary/mixer gates, or exclusively unitary gates, the method seems to always return False as opposed to its has_unitary counterpart, which returns True if and only if a circuit contains exclusively unitary gates.

[dabacon]

This is certainly a bug.

@Strilanc I think design for _act_on_ should consider problems like this.

[artvandalay404]

Don't know if this bug still exists? The tests are passing in has_mixture_protocol_test, and it worked for a simple GHZ circuit I ran.

[viathor]

This appears to have been fixed:

In [1]: import cirq

In [2]: q = cirq.LineQubit(0)

In [3]: def f(g):
   ...:     return cirq.has_mixture(g.on(q)), cirq.has_mixture(cirq.Circuit(g.on(q)))
   ...: 

In [4]: f(cirq.S)
Out[4]: (True, True)

In [5]: f(cirq.phase_flip(0.1))
Out[5]: (True, True)

In [6]: f(cirq.phase_damp(0.1))
Out[6]: (False, False)

In [7]: f(cirq.depolarize(0.1))
Out[7]: (True, True)

In [8]: f(cirq.amplitude_damp(0.1))
Out[8]: (False, False)

In [9]: 

Output number six is wrong and I filed #4578 for it. All other outputs are correct (only amplitude damping channel fails to be unital and hence does not admit a mixed-unitary representation). Also, cirq.mixture does fail for the circuits even though has_mixture returns True. Did not file an issue for it since we already have #2382. Also filed #4579 to prevent re-occurrence of these and similar problems (for which we'll have a larger exposure as we increase the number of ways channels can be represented in cirq).

Closing this issue as its core problem is fixed and related problems are covered by the other issues I liked above.