def _optimize_polyfit(server_address, periodic_callback):
settings_0 = np.zeros(len(template.CURRENT_FIELDS))
sets = np.identity(len(settings_0))
M = np.zeros(sets.shape)
sets_a = np.linspace(-5, 5, 5) * 0.01
readings = np.zeros([len(settings_0), 5])
for i in range(0, len(settings_0)):
for j in range(0, len(sets_a)):
is_steering, steering = _check_beam_steering(True)
if not is_steering:
raise AbortOptimizationException()
setting_test = sets[i,:] * sets_a[j]
template.write_epics_values(server_address, template.CURRENT_FIELDS, setting_test)
periodic_callback(server_address)
readings[:,j] = template.read_epics_values(server_address, template.BPM_FIELDS)
for k in range(0, len(settings_0)):
M[i,k] = np.polyfit(sets_a, readings[k,:], 1)[0]
# inverse response matrix
MI = np.linalg.pinv(M.T)
# reset the beam-line
template.write_epics_values(server_address, template.CURRENT_FIELDS, settings_0)
periodic_callback(server_address)
readings_1 = np.asarray(template.read_epics_values(server_address, template.BPM_FIELDS))
# create settings to cancel out offsets
new_sets = np.dot(MI, -readings_1)
return {
'message': '',
'success': True,
'result': new_sets,
}
def _cost_function(values, server_address, periodic_callback):
is_steering, steering = _check_beam_steering(True)
if not is_steering:
raise AbortOptimizationException()
template.write_epics_values(server_address, template.CURRENT_FIELDS, values)
# periodic_callback() either waits for the remote EPICS or runs a local sim which populates local EPICS
periodic_callback(server_address)
readings = template.read_epics_values(server_address, template.BPM_FIELDS)
#cost = np.sum((np.array(readings) * 1000.) ** 2)
cost = np.sum((np.array(
[readings[4] - readings[6], readings[5] - readings[7], readings[6], readings[7]]
) * 1000.) ** 2)
return cost
def _ping_epics(server_address):
return template.read_epics_values(server_address, [template.BPM_FIELDS[0]]) is not None