Merge pull request #339 from qiboteam/fix_delta_sign

Simplify Ramsey code

src/qibocal/protocols/characterization/ramsey.py

@@ -45,7 +45,7 @@ class RamseyResults(Results):
 class RamseyData(DataUnits):
  """Ramsey acquisition outputs."""
 
- def __init__(self, n_osc, t_max):
+ def __init__(self, n_osc, t_max, detuning_sign):
  super().__init__(
  name="data",
  quantities={"wait": "ns", "qubit_freqs": "Hz"},
@@ -56,6 +56,7 @@ def __init__(self, n_osc, t_max):
 
  self._n_osc = n_osc
  self._t_max = t_max
+ self._detuning_sign = detuning_sign
 
  @property
  def n_osc(self):
@@ -67,6 +68,11 @@ def t_max(self):
  """Final delay between RX(pi/2) pulses in ns."""
  return self._t_max
 
+ @property
+ def detuning_sign(self):
+ """Sign for induced detuning."""
+ return self._detuning_sign
+
 
 def _acquisition(
  params: RamseyParameters,
@@ -104,17 +110,20 @@ def _acquisition(
  # create a DataUnits object to store the results,
  # DataUnits stores by default MSR, phase, i, q
  # additionally include wait time and t_max
- data = RamseyData(params.n_osc, params.delay_between_pulses_end)
+ data = RamseyData(params.n_osc, params.delay_between_pulses_end, detuning_sign=+1)
 
  # sweep the parameter
  for wait in waits:
  for qubit in qubits:
  RX90_pulses2[qubit].start = RX90_pulses1[qubit].finish + wait
  ro_pulses[qubit].start = RX90_pulses2[qubit].finish
  if params.n_osc != 0:
+ # FIXME: qblox will induce a positive detuning with minus sign
  RX90_pulses2[qubit].relative_phase = (
  RX90_pulses2[qubit].start
- * (-2 * np.pi)
+ * data.detuning_sign
+ * 2
+ * np.pi
  * (params.n_osc)
  / params.delay_between_pulses_end
  )
@@ -203,9 +212,12 @@ def _fit(data: RamseyData) -> RamseyResults:
  popt[4] / (x_max - x_min),
  ]
  delta_fitting = popt[2] / (2 * np.pi)
- # FIXME: check this formula
- delta_phys = +int((delta_fitting - data.n_osc / data.t_max) * 1e9)
- corrected_qubit_frequency = int(qubit_freq + delta_phys)
+ delta_phys = data.detuning_sign * int(
+ (delta_fitting - data.n_osc / data.t_max) * 1e9
+ )
+ # FIXME: for qblox the correct formula is the following (there is a bug related to the phase)
+ # corrected_qubit_frequency = int(qubit_freq + delta_phys)
+ corrected_qubit_frequency = int(qubit_freq - delta_phys)
  t2 = 1.0 / popt[4]
 
  except Exception as e: