[BUG] Gaussian and DragCorrection envelopes contain `nan` values when `amplitude` is set to zero.

[fedonman]

Expected behavior

When running:

drag_pair = DragPair(amplitude=0.0, duration=40, num_sigmas=4, drag_coefficient=1.2)
print(drag_pair.I.envelope())
print(drag_pair.Q.envelope())

The envelopes should contain a list of zeros.

Actual behavior

Instead, the following is printed:

[nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
 nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
 nan nan nan nan]
[nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
 nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
 nan nan nan nan]

Additional information

No response

Source code

No response

Tracebacks

No response

Existing GitHub issues

• I have searched existing GitHub issues to make sure the issue does not already exist.

[visagim]

The formula for the envelope of the gaussian is:

def envelope(self, resolution: float = 1):
    """Returns the pulse matrix

    Args:
        resolution (int, optional): Pulse resolution. Defaults to 1.

    Returns:
        np.ndarray: pulse matrix
        resolution (int, optional): Pulse resolution. Defaults to 1.
    """
    x = np.arange(self.duration / resolution) * resolution

    gaussian = self.amplitude * np.exp(-0.5 * (x - self.mu) ** 2 / self.sigma**2)
    norm = np.amax(np.real(gaussian))

    gaussian = gaussian - gaussian[0]  # Shift to avoid introducing noise at time 0
    corr_norm = np.amax(np.real(gaussian))

    return gaussian * norm / corr_norm

If the whole envelope is 0, then division by corr_norm will return nan. This is because Gaussian does not expect the amplitude to be 0, same way as it does not expect sigma to be 0 (this would also result in an error).
Are there any scenarios in which one might send a pulse with 0 amplitude? If that is the case, there are probably more scenarios (in the old workflow with pulse_shape), where we operate over envelopes to add distorsions and so on which might give similar errors.

[GuillermoAbadLopez]

I mean a 0 amplitude pulse doesn't make much sense from base. But surely we should add a Raise with an if statement as Victor did also in the pulseshapes!