def split_gradient_at(grad: np.ndarray, time_point: float, system: Opts = Opts()):
"""
Split gradient waveform `grad` into two at time point `time_point`.
Parameters
----------
grad : numpy.ndarray
Gradient waveform to be split into two gradient waveforms.
time_point : float, optional
Time point at which `grad` will be split into two gradient waveforms.
system : Opts, optional
System limits. Default is a system limits object initialised to default values.
Returns
-------
grad1, grad2 : numpy.ndarray
Gradient waveforms after splitting.
"""
grad_raster_time = system.grad_raster_time
time_index = round(time_point / grad_raster_time)
time_point = time_index * grad_raster_time
time_index += 1
if grad.type == 'trap':
ch = grad.channel
grad.delay = round(grad.delay / grad_raster_time) * grad_raster_time
grad.rise_time = round(grad.rise_time / grad_raster_time) * grad_raster_time
grad.flat_time = round(grad.flat_time / grad_raster_time) * grad_raster_time
grad.fall_time = round(grad.fall_time / grad_raster_time) * grad_raster_time
if grad.flat_time == 0:
times = [0, grad.rise_time, grad.rise_time + grad.fall_time]
amplitudes = [0, grad.amplitude, 0]
else:
times = [0, grad.rise_time, grad.rise_time + grad.flat_time,
grad.rise_time + grad.flat_time + grad.fall_time]
amplitudes = [0, grad.amplitude, grad.amplitude, 0]
if time_point < grad.delay:
times = np.insert(grad.delay + times, 0, 0)
amplitudes = [0, amplitudes]
grad.delay = 0
amplitudes = np.array(amplitudes)
times = np.array(times)
amp_tp = np.interp(x=time_point, xp=times, fp=amplitudes)
times1 = np.append(times[np.where(times < time_point)], time_point)
amplitudes1 = np.append(amplitudes[np.where(times < time_point)], amp_tp)
times2 = np.insert(times[times > time_point], 0, time_point) - time_point
amplitudes2 = np.insert(amplitudes[times > time_point], 0, amp_tp)
grad1 = make_extended_trapezoid(channel=ch, system=system, times=times1, amplitudes=amplitudes1,
skip_check=True)
grad1.delay = grad.delay
grad2 = make_extended_trapezoid(channel=ch, system=system, times=times2, amplitudes=amplitudes2,
skip_check=True)
grad2.delay = time_point
return grad1, grad2
elif grad.type == 'grad':
if time_index == 1 or time_index >= len(grad.t):
return grad
else:
grad1 = grad
grad2 = grad
grad1.last = grad.waveform[time_index]
grad2.first = grad.waveform[time_index]
grad2.delay = grad.delay + grad.t[time_index]
grad1.t = grad.t[:time_index]
grad1.waveform = grad.waveform[:time_index]
grad2.t = grad.t[time_index:]
grad2.waveform = grad.waveform[time_index:]
return grad1, grad2
else:
raise ValueError('Splitting of unsupported event.')
def split_gradient(grad: np.ndarray, system: Opts = Opts()):
"""
Split gradient waveform `grad` into two gradient waveforms at the center.
Parameters
----------
grad : numpy.ndarray
Gradient waveform to be split into two gradient waveforms.
system : Opts, optional
System limits. Default is a system limits object initialised to default values.
Returns
-------
grad1, grad2 : numpy.ndarray
Split gradient waveforms.
"""
grad_raster_time = system.grad_raster_time
total_length = calc_duration(grad)
if grad.type == 'trap':
ch = grad.channel
grad.delay = round(grad.delay / grad_raster_time) * grad_raster_time
grad.rise_time = round(
grad.rise_time / grad_raster_time) * grad_raster_time
grad.flat_time = round(
grad.flat_time / grad_raster_time) * grad_raster_time
grad.fall_time = round(
grad.fall_time / grad_raster_time) * grad_raster_time
times = [0, grad.rise_time]
amplitudes = [0, grad.amplitude]
ramp_up = make_extended_trapezoid(channel=ch,
system=system,
times=times,
amplitudes=amplitudes,
skip_check=True)
ramp_up.delay = grad.delay
times = [0, grad.fall_time]
amplitudes = [grad.amplitude, 0]
ramp_down = make_extended_trapezoid(channel=ch,
system=system,
times=times,
amplitudes=amplitudes,
skip_check=True)
ramp_down.delay = total_length - grad.fall_time
ramp_down.t = ramp_down.t * grad_raster_time
flat_top = SimpleNamespace()
flat_top.type = 'grad'
flat_top.channel = ch
flat_top.delay = grad.delay + grad.rise_time
flat_top.t = np.arange(step=grad_raster_time,
stop=ramp_down.delay - grad_raster_time -
grad.delay - grad.rise_time)
flat_top.waveform = grad.amplitude * np.ones(len(flat_top.t))
flat_top.first = grad.amplitude
flat_top.last = grad.amplitude
return ramp_up, flat_top, ramp_down
elif grad.type == 'grad':
raise ValueError(
'Splitting of arbitrary gradients is not implemented yet.')
else:
raise ValueError('Splitting of unsupported event.')
