Source code for experiments.show_signal_wobbler

"""
Providing the "Show Signal Wobbler" experiment. It provides the same
functionality as the "Show Signal" experiment with the added benefit of
using the wobbler for phase cycling to remove scattering of the IR pump
pulse on the detector.

For this experiment to run properly the chopping scheme needs to be
adjusted to the according wobbler frequency. The algorithm identifies
the chopper states and the wobbler states, phase cycles over the wobbler
states and then calculates the difference signal (absorption) from the
different chopper states.  

While this experiment has the same applications as "Show Signal" it can
also be used to setup the wobbler because it also displays the signal
for each of the different wobbler states. See the note for an in depth
explanantion.

Other typical applications of this experiment are:

    *   Optimizing spatial overlap of the pump and probe pulses with
        respect to the resulting signal
    *   Finding a scatter free and generally optimal position in the
        sample
    *   Scanning different delays using the interface on the GUI
    *   Adjusting the central wavelength of the spectrometer

Note:

    **General:**

        The wobbler electronics provides three different parameters to
        adjust the wobbler:

        1.  The duty cycle of the square pulse that drives the harmonic
            motion of the wobbler

        2.  The phase (delay of the square pulse) relative to the laser
            trigger

        3.  The amplitude (voltage of the square pulse) of the harmonic
            motion of the wobbler

        The duty cycle should in general be set to 50 %. The wobbler
        only needs 2 degrees of freedom to be adjusted correctly. And
        the duty cycle can change both amplitude and phase. 

        Adjust the amplitude first since it also changes the phase
        relationship between laser and wobbler. As the scatter
        supression for a given wavelength depends on the amplitude make
        sure to optimize it either to the central wavelength of the
        spectrometer or to the wavelength region that is of most
        interest. Afterwards adjust the phase. Note that it can take
        seconds for the wobbler to reach equilibrium after the phase was
        changed. Either wait until the signals for the different wobbler
        states stop changing or use the reference laser diode to observe
        when the dots stop shifting. 

        Sometimes when turning the chopper on and off, the choppers'
        phase moves relative to the laser. In the case where there is no
        wobbler this might cause the sign of the observed signal to be
        flipped. When using a wobbler in combination with a chopper the
        frequency of the chopper has to be reduced to accommodate the
        fact that each wobbler position must be overserved for each
        chopper state. If the phase of the chopper now changes, it
        might only flip the sign of the signal of one or two of the
        different wobbler states. This might cause the user to think
        that he has adjusted the wobbler correctly, while it is not
        adjusted correctly. This results from an incorrect setting of
        the TTL delay electronics and can be fixed by following the
        procedure below.

        Note that alternatively this could be remedied by a different
        method for referencing the chopper. E.g.: Using a photodiode
        plus additional electronics to reference the laser light
        directly or using a light barrier to reference the position on
        the chopper where the laser passes through.

    **Chopper adjustment for wobbler:**
    
        **General procedure:**

            First turn off the wobbler. Now move the TTL of the chopper
            such that the four lines representing the different wobbler
            states are identical (and the sign of the signal is
            correct). Turn on the wobbler and adjust it in a way that
            two pairs of line are opposite to each other (cancel out).
            This should reduce the scattering.

        **Detailed procedure:**

            For adjusting the wobbler the moveable path of the
            interferometer needs to be blocked. A pulse delay circuit
            which can also set the duty cycle is required for the
            choppers. The duty cycle of the outgoing chopper pulse
            (delayed pulse) should be tuned to 50 % with the delay
            electronics. The next step is to set the chopper phase
            correctly. For that it is necessary to tune the delay of the
            outgoing pulse. If only three lines are displayed (wobbler
            off) the chopper duty cycle is not correct. If everything
            was set correctly (especially the phase of the chopper via
            the delay electronics) and the wobbler is off, the signals
            for a given wobbler state all overlap and have the correct
            sign. 

#######################
Step by Step Algorithm:
#######################

**Acquisition:**

    1.  Preallocate dictionary (data container) which will contain data
        and information about scan index, delay index etc.
    2.  Read the data from the ADC
    3.  Place data into dictionary and hand it over to
        primary processing

**Primary Processing:**

    1.  Preallocate arrays for data, counts, weights, chopper states and
        wobbler states. Here there are 2 states. On (chopper high)
        and off (chopper low). Also there are (generally) 4 different
        wobbler states (left, center, right, center position of wobbler)
    2.  Subtract background from raw data (dark noise)
    3.  Linearize response of pixels
    4.  Calculate transmission, or more precisely, relative intensity
        (probe intensity / reference intensity) for each laser shot for
        each pixel pair
    5.  Identify the chopper states for all laser shots using the 
        corresponding channel(s) in the ADCs' data
    6.  Identify the different wobbler states for all laser shots
    7.  Sort the data (transmissions) for each state and calculate 
        statistics
    8.  Phase cycle by averaging the wobbler states in the transmission 
        space. Calculate the resulting phase cycled counts and weights
    9.  Put this information into data container and hand it over to
        secondary processing
    10. Save phase cycled data (and raw data) including phase cycled
        counts and weights if respective checkboxes on GUI were checked

**Secondary Processing:**

    1.  Calculate the absorption of the phase cycled data (-log10)
    2.  Calculate the pump probe signal from the phase cycled absorption
    3.  Calculate the absorption (-log10) of the sorted data (non phase 
        cycled)
    4.  Calculate the pump probe difference signal (chopper high - chopper low)
        of the non phase cycled data
    5.  Calculate the average intensities and standard deviation of 
        the intensities for each pixel
    6.  Put this information into data container
    7.  Hand over data to Pyqtplotting thread
    8.  Calculate the numbers which are displayed in the 
        "statistics box" on the GUI

**Pyqt Plotting:**

    1.  Remove old plots
    2.  Setup the plot that displays:
            
            * Signals (including the signals for each wobbler state
              and the resulting phase cycled signal)
            * Intensities and their standard deviation (multiplied by 5)

    3.  Plot the plots for the first time
    4.  Update plots

**Saving:**

Note that the phase cycled transmission is saved. Therefore there exists
no dimension for the Wobbler states.

    .. code-block::
        
        programming data dimension: 
        [(2 ([0] is current average scan, [1] is last single scan), n_probe_pixels, n_wobbler_states, n_chopper_states)]

        saving dimension: 
        [(n_probe_pixels, n_chopper_states)]

        raw data dimension: 
        [(n_channels, samples_to_acquire)]

################
Folder Structure
################

In **scans/** the first file of 3 contains the phase cycled data in the
saving dimensions. The counts file contains the number of times a given
state was observed. This should be used as weights when averaging
equally weighted. The weights file contains the inverse variances of a
given state for all pixels, etc. The dimensions of these files are as
saving dimension suggests. These files can be used to average in
different ways in order to obtain the complete resulting spectrum. Check
the actual code of the corresponding experiment to understand how to
calculate the desired end result (difference spectrum).

The data in **/averaged_data** is averaged equally weighted (using
counts). Likewise to the scan data the difference spectrum still needs
to be calculated from the transmissions for every state. Because the
array contains the transmissions averaged with counts it is only
intended to be used as a first indicator for the measurement.

The **/figure** folder is currently empty. It is possible to implement
plotting of figures which are saved here while the experiment is
running. This is however not implemented yet.

The **/hardware config** folder holds every file which contains hardware
configuration parameters. This folder is a compressed copy of the
hardware configuration folder in the software's directory. Here it is
possible to look up the ADC configuration to obtain what channel was
connected to which hardware element (e.g. chopper - ai78). It is also
possible to obtain the R-2R values, the FPAS configuration, the
linearization parameters, etc.

The **/raw data** folder is in the experiments directory only if the
"save raw data" checkbox on the GUI was checked. It contains the raw
(unedited... as raw as it can get) data. Basically, the voltages which
the ADC measured for each channel. The dimensions are as "raw data
dimensions" suggests.

The **background.npy** file is a copy of the most recently collected
dark noise background of the MCT detector array. It corrects for dark
noise. If no new background was collected before the experiment was
started the code will use the latest available background and display a
warning in the log.

**setupinfo.txt** is a ReadMe file that contains the most relevant
experimental parameters at first glance. Additionally, the user can
decide to write a comment in the readme editor of the GUI. The content
of this is written to the **notes.txt** file.

.. code-block::

    username/
    ├── date1_experimentname1_000/
    │   ├── averaged_data
    │   │   └──  date1_experimentname1_000.npy
    │   ├── figures
    │   ├── hardware config
    │   ├── raw data
    │   │   ├──  s000000_date1_experimentname1_000_raw.npy
    │   │   ├──  ...
    │   │   └──  s000099_date1_experimentname1_000_raw.npy
    │   ├── scans
    │   │   ├──  s000000_date1_experimentname1_000.npy
    │   │   ├──  s000000_counts_date1_experimentname1_000.npy
    │   │   ├──  s000000_weights_date1_experimentname1_000.npy
    │   │   ├──  ...
    │   │   └──  s000099_d000_date1_experimentname1_000.npy
    │   ├── setupinfo_date1_experimentname1_000.txt
    │   ├── notes_date1_experimentname1_000.txt
    │   └── background_date1_experimentname1_000.npy
    ├── date1_experimentname1_001/
    ├── date2_experimentname1_000/
    └── date2_experimentname2_000/
"""
if __name__ == "__main__":
    # Add directories to path for imports
    import os, sys, inspect

    currentdir = os.path.dirname(
        os.path.abspath(inspect.getfile(inspect.currentframe()))
    )
    parentdir = os.path.dirname(currentdir)

    sys.path.insert(0, parentdir)
    sys.path.insert(0, os.path.join(parentdir, "hardware_interfaces"))
    sys.path.insert(0, os.path.join(parentdir, "gui"))

import multiprocessing
from multiprocessing import Process, Queue
import sys

import threading

import numpy as np
from numpy import ndarray

# Matplotlib
import matplotlib
from matplotlib.backends.backend_qt5agg import (
    FigureCanvasQTAgg,
    NavigationToolbar2QT as NavigationToolbar,
)
from matplotlib.figure import Figure

# PyQTGraph
from pyqtgraph import PlotWidget, plot
import pyqtgraph as pg

from PyQt5 import QtWidgets, QtCore  # , QtGui
from PyQt5.QtCore import QRunnable, QThreadPool, pyqtSignal, QObject
from qt_multithreading_wrapper import Worker

# Data processing
import data_processing as dp
from data_processing import PixelResponseLinearization as PRL
from data_processing import ChopperStateFinder as CSF

from analog_digital_converter import AnalogDigitalConverter as ADC
from triax import Triax as Spectrometer

from save_data import SaveData, Background


[docs]class ShowSignalWobbler: """ Args: widget_pyqtgraph (QWidget): WidgetPyqtgraph object on which the plots are going to be displayed. Has methods for plot manipulation (i.e. removal of plots, autoscale). adc (ADC): Analog to digital converter hardware object which is used to communicate with and read data from the ADC. prl (PRL): Pixel response linearization object which grants the functionality to linearize raw data according to the linearization parameters specified in the corresponding *pixel_linearization_fit_parameters.json* file (for each lab). chopper_info (dict): Contains the information that is necessary to identify the different chopper states of the chopper that chops the pump pulse. It contains the keys "high voltage level" and "name". "high voltage level" is the voltage read by the ADC when the chopper reference output is high. It is needed as a reference for the digitization function that is used. The "name" key is required to determine to which channel of the adc the chopper is connected and its value needs to match the corresponding key in index_dict. wobbler_freq (float): Frequency in Hz with which the wobbler oscillates. background_handler (Background): Instance of Background class which can access the most recently collected background. This background is later subtracted from the raw data as dark noise correction. spectrometer (Spectrometer): Spectrometer/Triax hardware class which grants functionality to control the triax spectrometer. Needed to obtain e.g. wavenumber axis etc. info_queue (Queue): Multiprocessing queue object which is used to transfer/hand over information to lineEdits on GUI. Contains (if applicable) scan index, delay index, interleave index, values for statistics groupBox etc. saver (SaveData): Object that manages saving of data (including counts, weights, probe wavenumber axis etc.) into their respective directories. If None is passed, no data is going to be saved. If the raw data checkbox was checked on the GUI the savers raw data attribute is set to True and the raw data is saved automatically. Defaults to None. """ def __init__( self, widget_pyqtgraph, adc: ADC, prl: PRL, chopper_info: dict, wobbler_freq: float, background_handler: Background, spectrometer: Spectrometer, info_queue: Queue, saver: SaveData = None, ): # Initialise Queues self.acq_queue = Queue() self.processing_queue = Queue() self.plot_queue = Queue() if saver: saver.save_other(background_handler.load_background(), "background") self.acquisition = Acquisition(adc, spectrometer, self.acq_queue) self.primary_processing = PrimaryProcessing( self.acq_queue, self.processing_queue, adc.pixel_idx, adc.probe_pixel_idx, adc.reference_pixel_idx, adc.index_dict, prl, chopper_info, wobbler_freq, adc.laser_frequency, background_handler, saver, ) self.secondary_processing = SecondaryProcessing( self.processing_queue, self.plot_queue, info_queue ) # self.plotting = Plotting(mpl_widget, adc, self.plot_queue) self.plotting = PyqtPlotting(widget_pyqtgraph, adc, self.plot_queue)
[docs] def start(self): self.plotting.threadpool.start(self.plotting.work) self.secondary_processing.start() self.primary_processing.start() self.acquisition.start()
[docs]class Acquisition(threading.Thread): """ Acquisition class (python multithreaded). This is where the actual experiment is conducted. This class is used to control the hardware devices required for the experiment. The sole purpose of it is to collect the data according to the parameters specified for the experiment by moving the delay stages, opening and closing shutters etc. A dictionary which will contain data and other information (scan index etc.) is instantiated here. The acquisition class passes the collected information (raw data, scan index etc.) to the primary processing class. Note: **No data processing beyond what is required to conduct the experiment should be implemented in this class.** The rationale behind this is to minimize down time/ maximize laser time. Data processing costs computation time and will, generally speaking, slow down the measurement process because the computer is busy while the rest of the hardware is idle. If implemented correctly the data processing could be carried out while the data acquisition is waiting for all data to become available. But even in this scenario the problem that the data processing takes longer than the acquisition time can occur and is thus best avoided through parallelisation. The reason why this class is a child of the threading module instead of the multiprocessing module is that to use multiprocessing all objects passed to the function must be picklable. This is not the case for some of the objects interfacing with the hardware (e.g. ADC). In an ideal scenario the acquisition too, would run in its own process seperated from the GUI thread but this would only be possible with major restructuring of the software. Args: adc (ADC): Analog to digital converter hardware object which is used to communicate with and read data from the ADC. spectrometer (Spectrometer): Spectrometer/Triax hardware class which grants functionality to control the triax spectrometer. Needed to obtain e.g. wavenumber axis etc. acq_queue (Queue): Multiprocessing queue object that the acquisition thread uses to pass data to the primary processing process. """ def __init__(self, adc: ADC, spectrometer: Spectrometer, acq_queue): threading.Thread.__init__(self) self.adc = adc self.spectrometer = spectrometer # Multiprocessing Queue self.acq_queue = acq_queue # Initialize scan index self.scan_idx = 0 # Create dictionary that will hold data that is passed # to other queues self.data_container = {} # Create dictionary that holds information # in which scan we are (etc.) self.info = {} # Multiprocessing event to stop experiment self.exit = threading.Event()
[docs] def run(self): # * You might wonder why this acquisition # * is structured differently from the acquisition # * of the other (older/simpler) experiments # * [Or you might not wonder - # * Lets hope that someone fixed this discrepancy # * already] # * The way it is implemented here (the straight # * forward way) was chosen because it simplifies # * the indices counting a lot, and within multiprocessing # * should not lead to any loss of laser time. # * Before the 0th data set was acquired outside of # * the loop and was processed while the second # * acquisition was running while not self.exit.is_set(): # Read data self.adc.read() # Put relevant information into data container self.data_container["data"] = self.adc.data self.data_container["scan index"] = self.scan_idx self.data_container["probe axis"] = self.spectrometer.wn_axis # Give data of to queue self.acq_queue.put(self.data_container.copy()) # Update scan idx self.scan_idx += 1 # Tell other processes to stop self.acq_queue.put("stop")
[docs] def shutdown(self): # Setting exit will stop the loop # within run self.exit.set()
[docs]class PrimaryProcessing(multiprocessing.Process): """ Primary processing class (python multiprocess). This class' purpose is to process the raw data to a state where it can be saved onto the hard drive as npy (binary) files. This step generally includes linearization, normalisation, sorting and averaging. Besides the actual data additional information that is required for post processing purposes is calculated and saved. I.e. counts and weights. The last step of primary processing should always be to pass the data to secondary processing and save it to the hard disk. Note: The actual signal(s) are generally not intended to be calculated here. Signals and other information that is supposed to be displayed on the GUI should be calculated in secondary processing. The main reason for this is minimizing the risk of an error leading to a crash of the software which then in turn ruins the measurement. The more code that has to run the more likely a crash becomes. Others reasons mostly imply open questions regarding averaging. Generally, shot to shot normalized intensities that are sorted and averaged by their state are saved (*m2 method*). From this - for a simple experiment at least - the signal can be easily calculated while offering different choices of averaging in post processing. For more complex experiments e.g. VIPER or time domain experiments the argument of saving sorted transmissions instead of signals is even more compelling. In VIPER experiments more than one signal of interest is present in the different states. Saving each signal separately would actually increase the amount of data that has to be saved. For time domain experiments we want to save the data in the time domain for post processing reasons like zeropadding and apodization. The goal of primary processing is to make the data as compact as possible while keeping as much information and flexibility as possible. Even if the processes later on crash, the data is secured and can be analysed in post processing. Args: acq_queue (Queue): Multiprocessing queue object that the acquisition thread uses to pass data to the primary processing process. processing_queue (Queue): Multiprocessing queue object that the primary processing process uses to pass data to the secondary processing process. pixel_idx (ndarray): Array that contains the indices of the rows in the ADCs' data that correspond to pixel input channels. These are specified in the "analog input configuration.json" for each laboratory and can be easily accessed with the attribute "pixel_idx" of the ADC. * shape: 1D * E.g.: (64) or (128) probe_pixel_idx (ndarray): Array that contains the indices of the rows in the ADCs' data that correspond to *probe* pixel input channels. These are specified in the "analog input configuration.json" for each laboratory and can be easily accessed with the attribute "probe_pixel_idx" of the ADC. It is highly relevant that the order the pixels are listed in this array match the order of the array in the reference_pixel_idx argument. This means that if reference pixel 3 is listed first in the other array here probe pixel 3 needs to be listed first as well and so on. Otherwise the intensities on the probe array are not going to be normalized correctly. For the plotting to work correctly the pixels also need to be listed in the order of the wavenumber axis array of the spectrometer. * shape: 1D * E.g.: (32) or (64) reference_pixel_idx (ndarray): Array that contains the indices of the rows in the ADCs' data that correspond to *reference* pixel input channels. These are specified in the "analog input configuration.json" for each laboratory and can be easily accessed with the attribute "reference_pixel_idx" of the ADC. It is highly relevant that the order the pixels are listed in this array match the order of the array in the probe_pixel_idx argument. This means that if probe pixel 10 is listed first in the other array here reference pixel 10 needs to be listed first as well and so on. Otherwise the intensities on the probe array are not going to be normalized correctly. For the plotting to work correctly the pixels also need to be listed in the order of the wavenumber axis array of the spectrometer. * shape: 1D * E.g.: (32) or (64) index_dict (dict): Dictionary that maps the names of the input channels to their corresponding row in the ADCs' data as they are specified in the "analog input configuration.json" for each laboratory. I.e.: It contains the information which entries of the ADCs' data array belong choppers, wobblers etc. This dictionary can be easily accessed with the attribute "index_dict" of the ADC. adc (ADC): Analog to digital converter hardware object which is used to communicate with and read data from the ADC. prl (PRL): Pixel response linearization object which grants the functionality to linearize raw data according to the linearization parameters specified in the corresponding *pixel_linearization_fit_parameters.json* file (for each lab). chopper_info (dict): Contains the information that is necessary to identify the different chopper states of the chopper that chops the pump pulse. It contains the keys "high voltage level" and "name". "high voltage level" is the voltage read by the ADC when the chopper reference output is high. It is needed as a reference for the digitization function that is used. The "name" key is required to determine to which channel of the adc the chopper is connected and its value needs to match the corresponding key in index_dict. wobbler_freq (float): Frequency in Hz with which the wobbler oscillates. laser_freq (float): Frequency (repetition rate) of the laser in Hz. background_handler (Background): Instance of Background class which can access the most recently collected background. This background is later subtracted from the raw data as dark noise correction. saver (SaveData): Object that manages saving of data (including counts, weights, probe wavenumber axis etc.) into their respective directories. If None is passed, no data is going to be saved. If the raw data checkbox was checked on the GUI the savers raw data attribute is set to True and the raw data is saved automatically. Defaults to None. """ def __init__( self, acq_queue: Queue, processing_queue: Queue, pixel_idx: ndarray, probe_pixel_idx: ndarray, reference_pixel_idx: ndarray, index_dict: dict, prl: PRL, chopper_info: dict, wobbler_freq: float, laser_freq: float, background_handler: Background, saver: SaveData = None, ): super(multiprocessing.Process, self).__init__() # Multiprocessing Queue # Gets data from acquisition class self.acq_queue = acq_queue # Gives data to secondary processing class self.processing_queue = processing_queue # Pixel response linearisation self.prl = prl # Wobbler frequency in Hz self.wobbler_freq = wobbler_freq # Laser frequency in Hz self.laser_freq = laser_freq # Data saving instance self.saver = saver # Pixel index is an array which tells us which # entries in our adc data are pixels # Append probe and reference pixel indices self.pixel_idx = pixel_idx self.probe_pixel_idx = probe_pixel_idx self.ref_pixel_idx = reference_pixel_idx self.index_dict = index_dict # Load background data from file self.background = background_handler.load_background() # In this experiment we have two different chopper states self.n_chopper_states = np.array(2) # Save chopper high voltage level divided by 2 as list # We need this for the digitize function that identifies # the High and Low Chopper # We select half of the high voltage as the limit # at which the distinction between states is done self.chopper_voltage_level = [chopper_info["high voltage level"][0] / 2] self.chopper_name = chopper_info["name"][0] # Calculate the number of wobbler states we are going to observe self.wobbler_states = int(laser_freq // wobbler_freq) # Stack the different number of states into one array. self.number_of_possible_states = np.array( [self.n_chopper_states, self.wobbler_states] ) # Initialize array that # sorted data is written into. # This is different from other # experiments (without wobbler) # because the delay axis as well # as the average and single scan # dimension have been removed. # Here we need to phase cycle directly: # 1. The get_wobbler_states algorithm # is not "safe" in fringe cases. # This could lead to averaging # different wobbler states together # when averaging different scans. # If we directly average wobbler states # for each scan this problem does not # arise. # 2. It makes the most sense to directly # phase cycle the data of one scan: # this has to do with the # imprecision of the delay stage # when measuring two time the same delay # it will move to slightly different # locations yielding slightly different # phases. Averaging these my yield # artifacts. # 3. This reduces the size of the data # set that is going to be saved by # a factor of 4. self.data = np.zeros( ( self.probe_pixel_idx.size, *self.number_of_possible_states, ) # all chopper states, all wobbler states ) self.counts = np.zeros(self.data.shape) self.weights = np.zeros(self.data.shape) # Because we use a Wobbler in this experiment # we need a second set of arrays where # the phase cycled/ scatter free transmissions # are written into. # Initialize array that holds both # temporary and averaged data # In this case the data is average # relative intensity # (transmission (probe/ref)) # for each pixel pair. # By convention the averaged is the # 0 th entry of the 0th axis of the array # and the temp data is the 1st entry of the 0th # axis of the array. # dimensions: 2 = (current average scan, last single scan) # dimensions: (2, n_probe_pixels, n_chopper_states) # Phase cycled = scatter free self.phase_cycled_data = np.zeros( ( 2, # averaged and non averaged (scan) self.probe_pixel_idx.size, self.n_chopper_states, ) # all chopper states ) self.phase_cycled_counts = np.zeros(self.phase_cycled_data.shape) self.phase_cycled_weights = np.zeros(self.phase_cycled_data.shape)
[docs] def run(self): while True: # Get data / information from acquisition process data_container = self.acq_queue.get() if type(data_container) == str: if data_container == "stop": break # Write data in dictionary into variable raw_data = data_container["data"] scan_idx = data_container["scan index"] # Subtract background from raw data # of pixels and pixels only background_corrected_data = ( raw_data[self.pixel_idx] - self.background[self.pixel_idx, np.newaxis] ) # Linearize response of Pixels (and pixels only) intensities = self.prl.linearize(background_corrected_data) # Calculate transmission/ relative intensity # (probe intensity / reference intensity) transmission = ( intensities[self.probe_pixel_idx] / intensities[self.ref_pixel_idx] ) # Get the corresponding chopper state for each shot chopper_states = np.digitize( raw_data[self.index_dict[self.chopper_name]], self.chopper_voltage_level ) # Get Wobbler States wobbler_states = dp.get_wobbler_states( raw_data[self.index_dict["wobbler"]], self.laser_freq, wobbler_freq=self.wobbler_freq, ) # Generate states array by stacking chopper states and wobbler states # ? There is proably a faster way to do this than stacking by preallocating arrays? Nvm for now states = np.vstack((chopper_states, wobbler_states)) # Sort and average data self.data[:], self.weights[:], self.counts[:], statistics = dp.sort_data( transmission, states, self.number_of_possible_states ) # --------- Phase cycling ---------- # phase cycled = scatter free # -- Phase cycle transmissions -- # Remove scattering by averaging the different wobbler states # * In VB6 the interleaves are averaged # * once the difference spectra are calculated # * this (below) first phase cycles and then # * offers data for signal calculating # ** Update: We tested/compared the difference # ** for one data set and the absolute difference # ** was consistently smaller 10e-3 smaller # ** than the signal (this was done with interleaves # ** but the same logic applies to the wobbler) # ** We (Jens and us) agree that it conceptually # ** makes more sense to average the relative # ** intensities. #! It should be tested whether it makes sense to first average #! scans and then phase cycle (we are first phase cycling then averaging here) # * In this scenario we don't actually need to compute the wobbler states, # * which wobbler position came first etc. # * It would suffice to separate the data in 4 different states like # * this: 0,1,2,3,0,1,2,3,0,1.... # * When averaging the way it is done here the actual position of the # * does not matter because they are averaged out before we average # * two different scans. self.phase_cycled_data[1] = np.average( self.data, # Do not use weights to phase cycle because each wobbler position needs to be weighted equally axis=-1, # The last axis of the array is the wobbler axis ) # Now we need to calculate the total counts # that were observed in each chopper state # for all wobbler states self.phase_cycled_counts[1] = self.counts.sum(axis=-1) # We also need to update the weights (inverse variance of each state) self.phase_cycled_weights[1] = self.weights.sum(axis=-1) # Add everything to data container # and give it to processing queue data_container["sorted data"] = self.data.copy() data_container["phase cycled data"] = self.phase_cycled_data[1].copy() data_container["intensities"] = intensities data_container["transmission"] = transmission data_container["statistics"] = statistics data_container["chopper states"] = chopper_states self.processing_queue.put(data_container.copy()) # ---------------------------------------- # Save data (if specified) if self.saver: # ----- Average data ---- # Create / Average "averaged" phase cycled data by using # weighted average between the phase cycled temp data # (weight = counts for each state in "temp" data set) # and the already existing average data # (weight = counts for each state in average data set) self.phase_cycled_data[0] = np.average( self.phase_cycled_data[:], axis=0, weights=self.phase_cycled_counts[:], ) # Use saver class to save data self.saver.save_scan(self.phase_cycled_data[1], scan_idx) self.saver.save_avg(self.phase_cycled_data[0]) self.saver.save_counts(self.phase_cycled_counts[1], scan_idx) self.saver.save_weights(self.phase_cycled_weights[1], scan_idx) if self.saver.raw_data: self.saver.save_raw_data(raw_data, scan_idx) # Update counts self.phase_cycled_counts[0] += self.phase_cycled_counts[1] # Once stop signal was received the function breaks out of the loop. # Now we need to tell the secondary processing to stop. self.processing_queue.put("stop")
[docs]class SecondaryProcessing(multiprocessing.Process): """ Secondary processing class (python multiprocess). This class is used to process the data from the primary processing class such that it can be displayed on the GUI within plots and lineEdits. This generally implies (if applicable): * calculation of signals * calculation of statistics like standard deviation of intensities and shot to shot standard deviation of signal * interpolation for 2D / heatmap plots (see comments in code why this is necessary) * Fourier transform and phasing for time domain data This data is handed over to the plotting thread. Note: The feature of saving figures/plots to the hard drive should be implemented here if it is needed. Args: processing_queue (Queue): Multiprocessing queue object that the primary processing process uses to pass data to the secondary processing process. plot_queue (Queue): Multiprocessing queue object that the secondary processing process uses to pass data to the plot thread. info_queue (Queue): Multiprocessing queue object which is used to transfer/hand over information to lineEdits on GUI. Contains (if applicable) scan index, delay index, interleave index, values for statistics groupBox etc. """ def __init__( self, processing_queue: Queue, plot_queue: Queue, info_queue: Queue, ): super(multiprocessing.Process, self).__init__() # Multiprocessing Queue # Gets data from acquisition class self.processing_queue = processing_queue # Gives data to secondary processing class self.plot_queue = plot_queue # Give experimental status and statistics to # GUI self.info_queue = info_queue
[docs] def run(self): while True: # Get data / information from acquisition process data_container = self.processing_queue.get() if type(data_container) == str: if data_container == "stop": break # ---------------------------------------- # Calculate information for plotting and GUI # We first calculate everything that is required # for plotting and pass it to the plot queue because # plotting can also cost time. Only then we # calculate the values that we want to display on # line edits on the GUI # --- Note considering naming: # For experiments with interleaves we choose # to call the non phase cycled signals="signal" # and the phase cycled signal="phase_cycled_signal" # here we reverse this order. So non phase cycled # signals = "npc_signal" and phase cycled signal # = "signal". # We do this because with interleaves it takes # time to measure all delay stage positions. # Here all different phase states are collected # within one acquisition. --- # Calculate the pump probe signal from the sorted # and phase cycled data # phase cycled = scatter free phase_cycled_data = data_container["phase cycled data"] absorption = -np.log10(phase_cycled_data) signal = absorption[:, 1] - absorption[:, 0] # Additionally calculate the non phased cycled signal # for each wobbler state # npc = non-phase cycled sorted_data = data_container["sorted data"] npc_absorption = -np.log10(sorted_data) npc_signal = npc_absorption[:, 1, :] - npc_absorption[:, 0, :] # Calculate the average intensity for each pixel avg_intensities = np.average(data_container["intensities"], axis=1) # Calculate the standard deviation of the # intensities std_intensities = np.std(data_container["intensities"], axis=1, ddof=1) # Add everything to data container # and give to plot queue data_container["signal"] = signal data_container["non phase cycled signal"] = npc_signal data_container["average intensity"] = avg_intensities data_container["std intensity"] = std_intensities self.plot_queue.put(data_container.copy()) # ---------------------------------------- # Calculate everything that is needed for statistical information # on GUI and add it to data container and give it to info queue # Hand over the mean state information for every pixel data_container["mean state intensity"] = np.average( data_container["transmission"], axis=1 ) data_container["mean state std"] = data_container["statistics"][1] self.info_queue.put(data_container.copy()) # Once stop signal was received the function breaks out of the loop. # Now we need to tell the plotting and updating of info on GUI to stop. self.plot_queue.put("stop") self.info_queue.put("stop")
[docs]class PyqtPlotting: """ Pyqt Plotting class (Qt multithreaded). This class is necessary for displaying plots on the GUI. PyQtGraph is used as the plotting engine. Generally the plots are set up first (type of plot, layout, title etc.). On the first run, the plots are drawn for the first time. Then the plots are updated every iteration. We update the same plot references every time to make it more efficient. Args: widget_pyqtgraph (QWidget): WidgetPyqtgraph object on which the plots are going to be displayed. Has methods for plot manipulation (i.e. removal of plots, autoscale). adc (ADC): Analog to digital converter hardware object which is used to communicate with and read data from the ADC. plot_queue (Queue): Multiprocessing queue object that the secondary processing process uses to pass data to the plot thread. """
[docs] class Signals(QObject): new_data = pyqtSignal(dict)
def __init__(self, widget_pyqtgraph, adc: ADC, plot_queue: Queue): # Assign attributes self.adc = adc self.widget_pyqtgraph = widget_pyqtgraph self.graphics_layout = widget_pyqtgraph.graphics_layout self.plot_queue = plot_queue # Setup signals and threadpool self.threadpool = QThreadPool() self.signals = self.Signals() # Clear old plots self.widget_pyqtgraph.remove_plots() # Setup plot that displays pump probe signal self.widget_pyqtgraph.plots["signal"] = self.graphics_layout.addPlot( row=0, col=0, colspan=2 ) self.widget_pyqtgraph.plots["signal"].setTitle("Average signal") self.widget_pyqtgraph.plots["signal"].setLabel( "bottom", "wavenumber [cm<sup>-1</sup>]" ) self.widget_pyqtgraph.plots["signal"].setLabel("left", "difference signal[OD]") self.widget_pyqtgraph.set_style(self.widget_pyqtgraph.plots["signal"]) # Setup plot that displays intensities self.widget_pyqtgraph.plots["intensities"] = self.graphics_layout.addPlot( row=2, col=0 ) self.widget_pyqtgraph.plots["intensities"].setTitle("Average intensities") self.widget_pyqtgraph.plots["intensities"].setLabel( "bottom", "wavenumber [cm<sup>-1</sup>]" ) self.widget_pyqtgraph.plots["intensities"].setLabel("left", "intensity [a.u.]") self.widget_pyqtgraph.set_style(self.widget_pyqtgraph.plots["intensities"]) # Create dictionary that holds reference to lines etc. self.plot_ref = {} # Connect signal that data has arrived to update the plot self.signals.new_data.connect( lambda data_container: self.update_plot(data_container) ) # Start loop that gets data from queue in Qt Thread self.work = Worker(self.run)
[docs] def run(self): while True: data_container = self.plot_queue.get() if type(data_container) == str: if data_container == "stop": break self.signals.new_data.emit(data_container)
[docs] def update_plot(self, data_container): # Plot for the first time to get line references signal = data_container["signal"] npc_signal = data_container["non phase cycled signal"] probe_axis = data_container["probe axis"] avg_intensities = data_container["average intensity"] std_intensities = data_container["std intensity"] if self.plot_ref: # Update non phase cycled signal / Signal for each wobbler state for i in range(npc_signal.shape[-1]): self.plot_ref["signal wobbler state {}".format(i)].setData( x=probe_axis, y=npc_signal[:, i] ) # Update signal plot self.plot_ref["signal"].setData(x=probe_axis, y=signal) # Update intensity error bars for probe array self.plot_ref["probe error bars"].setData( x=probe_axis, y=avg_intensities[self.adc.probe_pixel_idx], height=5 * std_intensities[self.adc.probe_pixel_idx], ) # Update intensity error bars for reference array self.plot_ref["ref error bars"].setData( x=probe_axis, y=avg_intensities[self.adc.reference_pixel_idx], height=5 * std_intensities[self.adc.reference_pixel_idx], ) # Update intensities self.plot_ref["probe intensities"].setData( x=probe_axis, y=avg_intensities[self.adc.probe_pixel_idx] ) self.plot_ref["ref intensities"].setData( x=probe_axis, y=avg_intensities[self.adc.reference_pixel_idx] ) else: # First time plotting signal_pen = pg.mkPen(color="#d62728", width=2.5, style=QtCore.Qt.SolidLine) npc_signal_pen = pg.mkPen( color="#A9A9A9", width=1.2, style=QtCore.Qt.SolidLine ) probe_pen = pg.mkPen(color="#1f77b4", width=2.5, style=QtCore.Qt.SolidLine) reference_pen = pg.mkPen( color="#ff7f0e", width=2.5, style=QtCore.Qt.SolidLine ) std_intensity_pen = pg.mkPen( color="#7f7f7f", width=1.5, style=QtCore.Qt.SolidLine ) # ? dashed lines? std_signal_pen = pg.mkPen( color="#2ca02c", width=2.5, style=QtCore.Qt.SolidLine ) # Plot non phase cycled signal / Signal for each wobbler state for i in range(npc_signal.shape[-1]): self.plot_ref[ "signal wobbler state {}".format(i) ] = self.widget_pyqtgraph.plots["signal"].plot( x=probe_axis, y=npc_signal[:, i], name="Pump-probe signal wobbler state {}".format(i), pen=npc_signal_pen, ) # Plot signal self.plot_ref["signal"] = self.widget_pyqtgraph.plots["signal"].plot( x=probe_axis, y=signal, name="Phase cycled pump-probe signal", pen=signal_pen, ) # Create intensity error bars for probe array self.plot_ref["probe error bars"] = pg.ErrorBarItem( x=probe_axis, y=avg_intensities[self.adc.probe_pixel_idx], height=5 * std_intensities[self.adc.probe_pixel_idx], beam=0.3, pen=std_intensity_pen, ) self.widget_pyqtgraph.plots["intensities"].addItem( self.plot_ref["probe error bars"] ) # Create intensity error bars for reference array self.plot_ref["ref error bars"] = pg.ErrorBarItem( x=probe_axis, y=avg_intensities[self.adc.reference_pixel_idx], height=5 * std_intensities[self.adc.reference_pixel_idx], beam=0.3, pen=std_intensity_pen, ) self.widget_pyqtgraph.plots["intensities"].addItem( self.plot_ref["ref error bars"] ) # Plot intensities self.plot_ref["probe intensities"] = self.widget_pyqtgraph.plots[ "intensities" ].plot( x=probe_axis, y=avg_intensities[self.adc.probe_pixel_idx], name="Average intensities on probe array", pen=probe_pen, ) self.plot_ref["ref intensities"] = self.widget_pyqtgraph.plots[ "intensities" ].plot( x=probe_axis, y=avg_intensities[self.adc.reference_pixel_idx], name="Average intensities on reference array", pen=reference_pen, ) self.widget_pyqtgraph.disable_autoscale()