fix circuit plot error for gates with classical controls when reverse_states=False

[kpobrien]

For circuits with a classical control and when reverse_state=False, the error reported in #220 occurs. This pull request changes the sign of the argument of \ctrl when reverse_state=False. With this change, the test circuit is plotted without errors:

from qutip_qip.circuit import QubitCircuit
qc = QubitCircuit(1, num_cbits=1, reverse_states=False)
qc.add_gate("X", targets=0, classical_controls=[0])
print(qc.latex_code())
qc.png

\documentclass[border=3pt]{standalone}
\usepackage[braket]{qcircuit}
\begin{document}
\Qcircuit @C=1cm @R=1cm {
 &  &  \gate{X}  & \qw \\ 
 &  &  \ctrl{-1}  & \qw \\ 
}
\end{document}

A more complex test case:

from qutip_qip.circuit import QubitCircuit, CircuitSimulator
from qutip_qip.operations import hadamard_transform

qc = QubitCircuit(3, num_cbits=2, reverse_states=False)
qc.user_gates = {"H": hadamard_transform(1)}
qc.add_gate("CNOT", targets=1, controls=2)
qc.add_gate("H", targets=2)
qc.add_measurement("M1", targets=1, classical_store=0)
qc.add_measurement("M2", targets=2, classical_store=1)
qc.add_gate("X", targets=0, classical_controls=0)
qc.add_gate("Z", targets=0, classical_controls=1)
qc.png

from qutip_qip.circuit import QubitCircuit, CircuitSimulator
from qutip_qip.operations import hadamard_transform

qc = QubitCircuit(3, num_cbits=2, reverse_states=True)
qc.user_gates = {"H": hadamard_transform(1)}
qc.add_gate("CNOT", targets=1, controls=2)
qc.add_gate("H", targets=2)
qc.add_measurement("M1", targets=1, classical_store=0)
qc.add_measurement("M2", targets=2, classical_store=1)
qc.add_gate("X", targets=0, classical_controls=0)
qc.add_gate("Z", targets=0, classical_controls=1)
qc.png

[BoxiLi]

Thank you! Could you add the simple example to the tests/test_circuit.py? Similar to

qutip-qip/tests/test_circuit.py

Lines 678 to 693 in 6a2977e

	def test_latex_code_teleportation_circuit(self):
	qc = _teleportation_circuit()
	latex = qc.latex_code()
	assert latex == self._latex_template % "\n".join([
	r" & \lstick{c1} & \qw & \qw & \qw & \qw"
	r" & \qw \cwx[4] & \qw & \qw & \ctrl{2} & \qw \\ ",
	r" & \lstick{c0} & \qw & \qw & \qw & \qw"
	r" & \qw & \qw \cwx[2] & \ctrl{1} & \qw & \qw \\ ",
	r" & \lstick{\ket{0}} & \qw & \targ & \qw & \qw"
	r" & \qw & \qw & \gate{X} & \gate{Z} & \qw \\ ",
	r" & \lstick{\ket{0}} & \gate{{\rm H}} & \ctrl{-1} &"
	r" \targ & \qw & \qw & \meter & \qw & \qw & \qw \\ ",
	r" & \lstick{\ket{q0}} & \qw & \qw & \ctrl{-1} &"
	r" \gate{{\rm H}} & \meter & \qw & \qw & \qw & \qw \\ ",
	"",
	])

[kpobrien]

Certainly and thanks for the quick response. Is adding the following code below the function test_latex_code_teleportation_circuit what you had in mind? If yes, I'll revise the PR to reflect this.

    def test_latex_code_classical_controls(self):
        qc = QubitCircuit(1, num_cbits=1, reverse_states=True)
        qc.add_gate("X", targets=0, classical_controls=[0])
        latex = qc.latex_code()
        assert latex == self._latex_template % "\n".join([
            r" &  &  \ctrl{1}  & \qw \\ ",
            r" &  &  \gate{X}  & \qw \\ ",
            "",
        ])

    def test_latex_code_classical_controls_non_reversed(self):
        qc = QubitCircuit(1, num_cbits=1, reverse_states=False)
        qc.add_gate("X", targets=0, classical_controls=[0])
        latex = qc.latex_code()
        assert latex == self._latex_template % "\n".join([
            r" &  &  \gate{X}  & \qw \\ ",
            r" &  &  \ctrl{-1}  & \qw \\ ",
            "",
        ])

[BoxiLi]

Yes, precisely. Maybe in one test function instead of two.

After being added, these tests will be automatically executed on GitHub. You can also test it locally with pytest tests to verify if everything is correct.

[kpobrien]

Yes, precisely. Maybe in one test function instead of two.

Done, I updated the PR and it passes the tests locally. I see a message "3 workflows awaiting approval", so you may need to approve running the tests on Github.

[BoxiLi]

All looks good, thanks for your contribution!