Aberrations simulation support : from twisted to PyTango

[utkarshp1161]

Discussed in this issue: #66

What is done:

1. Client Notebook: 2_b_Aberrations_twin : Generate aberrations using PyTemlib and register in twin and get images
2. changes in CORRECTOR.py
   • added command : set_aberrations_coeff_sim and get_aberrations_coeff_sim
     

     asyncroscopy/asyncroscopy/hardware/CORRECTOR.py

     Line 185 in 11b422a

     def set_aberrations_coeff_sim(self, json_aberrations_string: str):

     
     

     asyncroscopy/asyncroscopy/hardware/CORRECTOR.py

     Line 190 in 11b422a

     def get_aberrations_coeff_sim(self):

3. changes in ThermoDigitalTwin.py
   • added command : get_scanned_image_with_aberrations
     

     asyncroscopy/asyncroscopy/ThermoDigitalTwin.py

     Line 146 in 11b422a

     def get_scanned_image_with_aberrations(self) -> tuple[str, bytes]:

   • added internal helper : _acquire_stem_image_aberrations

     asyncroscopy/asyncroscopy/ThermoDigitalTwin.py

     Line 194 in 11b422a

     def _acquire_stem_image_aberrations(self, imsize: int, dwell_time: float, detector_list: list) -> np.ndarray:

   • How is aberrations coeff's are read using CORRECTOR? - see

     asyncroscopy/asyncroscopy/ThermoDigitalTwin.py

     Lines 173 to 174 in 11b422a

     	ab = corrector.get_aberrations_coeff_sim()# is a json string
     	self.ab = json.loads(ab)

   • added "corrector": self.corrector_device_address in _connect_detector_proxies

     asyncroscopy/asyncroscopy/ThermoDigitalTwin.py

     Line 100 in 11b422a

     "corrector": self.corrector_device_address

4. The simulation code is written in asyncroscopy/simulation/StemSim.py which borrows logic from https://github.com/pycroscopy/asyncroscopy/blob/twisted-legacy/asyncroscopy/cloned_repos/pystemsim/data_generator.py
5. 2 cif files added to data/cif_files for simulation
6. Test : tests/test_stem_sim.py added to test asyncroscopy/simulation/StemSim.py

Maybe future:
A) The image simulation returns a fixed (905,905) image:

asyncroscopy/asyncroscopy/ThermoDigitalTwin.py

Lines 194 to 231 in 11b422a

	def _acquire_stem_image_aberrations(self, imsize: int, dwell_time: float, detector_list: list) -> np.ndarray:
	"""
	"""
	size = imsize
	beam_current = 100 # pA
	ab = self.ab
	ab['acceleration_voltage'] = 60e3 # eV
	fov = 96 # angstroms
	ab['FOV'] = fov /12 # Angstroms
	ab['convergence_angle'] = 30 # mrad
	ab['wavelength'] = it.get_wavelength(ab['acceleration_voltage'])
	cif_path = (
	PROJECT_ROOT
	/ "data"
	/ "cif_files"
	/ "WS2_ortho.cif"
	)
	print("Reading CIF from:", cif_path)
	xtal = read(cif_path)
	xtal = xtal * (30, 20, 1)
	positions = xtal.get_positions()[:, :2]
	pixel_size = 0.106 # angstrom/pixel
	frame = (0,fov,0,fov) # limits of the image in angstroms
	potential = dg.create_pseudo_potential(xtal, pixel_size, sigma=1, bounds=frame, atom_frame=11)
	probe = dg.get_probe(ab, potential)
	image = dg.convolve_kernel(potential, probe)
	noisy_image = dg.lowfreq_noise(image, noise_level=0.5, freq_scale=.04)
	scan_time = dwell_time * size * size
	counts = scan_time * (beam_current * 1e-12) / (1.602e-19)
	sim_im = dg.poisson_noise(noisy_image, counts=counts)
	# convert args dict
	# time.sleep(1)
	image = np.array(image, dtype=np.float32)
	sim_im = np.array(sim_im, dtype=np.float32)
	print(sim_im.shape) # TODO: the simulation is independent of size and always returns - (905, 905) image -> need to check
	return sim_im

B) The simulation related to nanoparticles

asyncroscopy/asyncroscopy/ThermoDigitalTwin.py

Line 233 in 11b422a

def _acquire_stem_image(self, imsize: int, dwell_time: float, detector_list: list) -> np.ndarray:

can go to https://github.com/pycroscopy/asyncroscopy/blob/dev-up/asyncroscopy/simulation/StemSim.py ?
C) Call CORRECTOR's all other commands through microscope as illustrated here for this case

asyncroscopy/asyncroscopy/ThermoDigitalTwin.py

Lines 173 to 174 in 11b422a

	ab = corrector.get_aberrations_coeff_sim()# is a json string
	self.ab = json.loads(ab)

• Why makes sense ? as EELS detector should be also called from microscope.get_spectrum(detector = "eels")

[AustinHouston]

A few things I would overhaul before merging

there should be no _sim functions (such as set_aberrations_coeff_sim and get_scanned_image_with_aberrations)
that's the whole point of asyncroscopy - the notebook can run on a thermo, a jeol, a nion, or a digital twin - doesn't matter.

to fix this, just the _get_scanned_image function in ThermoMicroscope does not change
the _get_scanned_image in TwinMicroscope just checks the corrector, makes the probe, etc.
(the abberations should be attributes of the corrector, with init values and limits)

[AustinHouston]

Now that you have implemented the ceos communication this should all be easier to do.

[utkarshp1161]

(the abberations should be attributes of the corrector, with init values and limits)

I see two caveat's to it.
a) The actual corrector values are currently handled using json dump and not as attributes,

asyncroscopy/asyncroscopy/hardware/CORRECTOR.py

Lines 136 to 150 in 4ceef8c

	def acquire_tableau(self, args: str) -> str:
	"""
	Run a correction tableau on the CEOS corrector.
	Parameters are packed into one space-separated string because
	Tango scalar commands accept only one input argument::
	proxy.run_tableau("Fast 18")
	proxy.run_tableau("Full 0")
	"""
	parts = args.strip().split()
	tab_type, angle_str = parts
	angle = float(angle_str)
	return self._call("acquireTableau", {"tabType": tab_type, "angle": angle})

b) The tableau acquired from the corrector has different enteries(even across different type: enhanced, fast) than from one we get from the simulation:

• The one we get on actual corrector: - Case 1: ab_msg = corrector_proxy.acquire_tableau('Enhanced 40')

{'A1': [-1.722984470953829e-09, -1.3730557569257383e-09], 'A2': [1.3574035073166904e-07, -3.743269769849526e-10], 'C3': [-1.9124382493333032e-07, 0.0], 'C1': [-8.459171809676403e-10, 0.0], 'A4': [2.5050956356475984e-07, -5.679158622685287e-07], 'A3': [8.007267595329108e-08, 1.8635595817710287e-07], 'A5': [2.8761931131477272e-05, -2.9930062430712584e-05], 'B2': [5.072942708488215e-09, -1.0153237290799634e-08], 'B4': [5.410183467091773e-07, 6.431281147405137e-07], 'S3': [6.957003967213865e-07, 2.806877916263183e-07], 'C5': [-0.00015297021485109578, 0.0], 'D4': [8.243145596145829e-07, 1.0160618230844962e-05], 'WD': [-4.329105939535138e-06, -9.531246135832501e-06], 'acceleration_voltage': 60000.0, 'FOV': 10.0, 'convergence_angle': 3, 'wavelength': np.float64(0.004866060499839593), 'C01a': 0.0, 'C01b': 0.0, 'C10': 0.0, 'C12a': 0.0, 'C12b': 0.0, 'C21a': 0.0, 'C21b': 0.0, 'C23a': 0.0, 'C23b': 0.0, 'C30': 0.0, 'C32a': 0.0, 'C32b': 0.0, 'C34a': 0.0, 'C34b': 0.0, 'C41a': 0.0, 'C41b': 0.0, 'C43a': 0.0, 'C43b': 0.0, 'C45a': 0.0, 'C45b': 0.0, 'C50': 0.0, 'C52a': 0.0, 'C52b': 0.0, 'C54a': 0.0, 'C54b': 0.0, 'C56a': 0.0, 'C56b': 0.0}

• The one we get on actual corrector: - Case 2: ab_msg = corrector_proxy.acquire_tableau('Fast 18')

{'A1': [9.74690226274642e-11, -1.3434929311990205e-09], 'A2': [7.590644738179404e-08, -3.566485459511854e-08], 'B2': [1.1154583587841222e-08, 3.025011301977759e-09], 'C1': [-1.6599977642222055e-09, 0.0], 'WD': [-8.611027075525367e-05, -0.00024514653261427925]}

• The one we get from simulation:
  {'C10': 0, 'C12a': 0, 'C12b': 0.38448128113770325, 'C21a': -68.45251255685642, 'C21b': 64.85359774641199, 'C23a': 11.667578600494137, 'C23b': -29.775627778458194, 'C30': 123, 'C32a': 95.3047364258614, 'C32b': -189.72105710231244, 'C34a': -47.45099594807912, 'C34b': -94.67424667529909, 'C41a': -905.31842572806, 'C41b': 981.316128853203, 'C43a': 4021.8433526960034, 'C43b': 131.72716642732158, 'C45a': -4702.390968272048, 'C45b': -208.25028574642903, 'C50': 552000.0, 'C52a': -0.0, 'C52b': 0.0, 'C54a': -0.0, 'C54b': -0.0, 'C56a': -36663.643489934424, 'C56b': 21356.079837905396, 'acceleration_voltage': 200000, 'FOV': 34.241659495148205, 'Cc': 1000000.0, 'convergence_angle': 30, 'wavelength': 0.0025079340450548005}

[utkarshp1161]

Since, this needs more thought about how to go about attributes. Merging this for now as it doesn't break other features and tests.