Plotting inside a thread and update it (python)

Question

I'm facing a problem working with multiple threads in Python. Let me give the context, I'm trying to read values from an oscilloscope and make a histogram in real time. But the problem is the plotting in real time make my measurements very slow.

So I thought to read the values from the oscilloscope in one thread and then plot the histogram in another thread - which should fix the slowness. Since I'm a beginner with python and I'm getting a lot of errors.

This is my code I wrote:

    import time
    import matplotlib.pyplot as plt
    import numpy as np
    from matplotlib.ticker import PercentFormatter
    from numpy.core.shape_base import block
    import pyvisa
    import os
    from matplotlib.pyplot import  figure, step
    from scipy.signal import find_peaks
    import queue
    import threading

    def acquisition(queue, name):
        rm = pyvisa.ResourceManager()
        rta = rm.open_resource('TCPIP::192.168.100.101::INSTR')

        path='Desktop/Laboratorio/Programacion- 
        Automatizacion/Pyvisa/Output/'

        rta.write("MEASurement1:TIMeout:AUTO")

        valuep = []
        aux = 0

        #Comprobamos que no existen ya esos ficheros
        if os.path.isfile(path + name + '.txt'):
            os.remove(path + name + '.txt') 

        if os.path.isfile(path + name + '.png'):
            os.remove(path + name + '.png') 

        start_time = time.time()

        aux
        entries=1000
        gain=0  
        completed = 0

        while len(valuep) < entries:

            if rta.query("*OPC?"):
                p1=float(rta.query("CURSor1:Y1Position?"))
            if rta.query("*OPC?"):
                p2=float(rta.query("CURSor1:Y2Position?"))

            r=(p1-p2)
            valuep.append(r)
            a = np.array(valuep)

            completed = len(valuep)/entries*100
            print(str(completed) + ' %\n')

            hist, bin_edges = np.histogram(a, 300)
            bin_edges = bin_edges[1:]
            peaks, _ = find_peaks(hist, distance=10, prominence=10)
            
            sumDelta = 0

            if len(peaks) >= 2:
                sumDelta = 0
                for i in range(len(peaks)-1):
                    deltaV=bin_edges[peaks[i+1]]-bin_edges[peaks[i]]
                    sumDelta += deltaV
                    gain=(sumDelta/(len(peaks)-1))/(50*1.602e-19)

            if completed == 10 or completed == 20 or completed == 30 or 
            completed == 40 or completed == 50 or completed == 60 or 
            completed == 70 or completed == 80 or completed == 90 or 
            completed == 100:
                auxLen = np.arange(0 , len(valuep) , 1)
                with open(path + name + '.txt', 'w') as f:
                    f.write(str(gain) + ' | ' + str(round((time.time() - 
                    start_time)/60, 3)) + '\n')
                    for i in auxLen:
                        f.write(str(valuep[i]))
                        f.write('\n')
            message = a
            queue.put(message)

        rta.close()

    def histogram(queue, name):
        while True:
            message = queue.get()
            plt.cla()
            # plt.xlabel("Vs | %=" + str(round(len(valuep)/entries*100, 3)) + " | Time: " + str(round((time.time() - start_time)/60, 3)) + " min")
            plt.ylabel("Bins")
            plt.yscale('log')
            hist, bin_edges = np.histogram(message, 300)
            bin_edges = bin_edges[1:]
            plt.plot(bin_edges, hist)
            # plt.plot(bin_edges[peaks], hist[peaks], "x")
            plt.show()
            plt.tight_layout()
    
    
    if __name__ == "__main__":
        print('Input code name: ')
        name = input()
        q = queue.Queue()
        acquire = threading.Thread(target=acquisition, args=(q, name,))
        hist = threading.Thread(target=histogram, args=(q, name,))
        acquire.start()
        hist.start()
        acquire.join()
        hist.join()
        q.join()
        plt.plot()

The function acquire reads values from oscilloscope and return it to a queue as a numpy array.

The error is:

UserWarning: Starting a Matplotlib GUI outside of the main thread will likely fail.

I really appreciate your help and any comments, to how should I proceed. Thanks in advance.

This is the code I had at first:

#librerias
import matplotlib.pyplot as plt
import numpy as np
from matplotlib.ticker import PercentFormatter
import pyvisa
import time
import os
from itertools import count
from matplotlib.animation import FuncAnimation
from matplotlib.pyplot import  figure, step
from scipy.signal import find_peaks

#Abrir sesion VISA
rm = pyvisa.ResourceManager()
rta = rm.open_resource('TCPIP::192.168.100.101::INSTR')

print('Input code name: ')
name=input()

path='Desktop/Laboratorio/Programacion-Automatizacion/Pyvisa/Output/'

rta.write("MEASurement1:TIMeout:AUTO")
rta.write("SYSTem:COMMunicate:INTerface:ETHernet:TRANsfer FD100")
rta.write("FORM BIN") # // Set BIN data format

valuep = []
aux = 0

#Comprobamos que no existen ya esos ficheros
if os.path.isfile(path + name + '.txt'):
    os.remove(path + name + '.txt') 

if os.path.isfile(path + name + '.png'):
    os.remove(path + name + '.png') 

start_time = time.time()

def animate(i):
    global aux
    entries=1000
    gain=0  

    if rta.query("*OPC?"):
        p1=float(rta.query("CURSor1:Y1Position?"))
    if rta.query("*OPC?"):
        p2=float(rta.query("CURSor1:Y2Position?"))

    r=(p1-p2)
    valuep.append(r)
    a = np.array(valuep)

    plt.cla()
    plt.xlabel("Vs | %=" + str(round(len(valuep)/entries*100, 3)) + " | Time: " + str(round((time.time() - start_time)/60, 3)) + " min")
    plt.ylabel("Bins")
    plt.yscale('log')

    hist, bin_edges = np.histogram(a, 300)
    bin_edges = bin_edges[1:]
    plt.plot(bin_edges, hist)
    peaks, _ = find_peaks(hist, distance=10, prominence=10)
    
    sumDelta = 0

    if len(peaks) >= 2:
        sumDelta = 0
        for i in range(len(peaks)-1):
            deltaV=bin_edges[peaks[i+1]]-bin_edges[peaks[i]]
            sumDelta += deltaV
        gain=(sumDelta/(len(peaks)-1))/(50*1.602e-19)
        #print(gain)

    plt.plot(bin_edges[peaks], hist[peaks], "x")
    plt.show()

    plt.tight_layout() 

    #Escritura fichero CSV
    if len(valuep) == entries:
        auxLen = np.arange(0 , len(valuep) , 1)
        with open(path + name + '.txt', 'w') as f:
            f.write(str(gain) +'\n')
            for i in auxLen:
                f.write(str(valuep[i]))
                f.write('\n')
        plt.savefig(path + name + '.png')

ani = FuncAnimation(plt.gcf(), animate, interval=0.013)

plt.tight_layout()

plt.show()
rta.close()

This code work just like I want, but very very slow, Im speaking that plotting with aniFunction need 96min for 15000 acquisition and without anifunc need 26min for 30000 acquisition, thats why Im trying with threads.

Tomorrow I will make a example of "def acquisition" to send it to a friend, for replicate without the scope, and then I will make a edit with this code. — FERNANDO FUENTES-GUERRA FERNND, Oct 19 '21 at 12:17
UserWarning is not actually an error, it says it could fail. But what is the issue you are facing ? the plot is happening ? — Kris, Oct 19 '21 at 12:21
@kris, Hi Kris, the plot is realizing, but not updating in real time, do u said that this warning isnt a important isssue? Do u know how can I use the aniFunction here? And thanks a lot for edit my english and comment. — FERNANDO FUENTES-GUERRA FERNND, Oct 19 '21 at 12:27
Can you post the complete code ? I'm not clear on your need to join all threads and plot at last. IMO, you should also consider a possibility of multi-process instead of multi-threading, as its not so promising for performance improvement — Kris, Oct 19 '21 at 12:35
@Kris can I upload my first code that run slow, and the code I have now? Or is it a lot? — FERNANDO FUENTES-GUERRA FERNND, Oct 20 '21 at 06:12
Does this answer your question? [Why can't matplotlib plot in a different thread?](https://stackoverflow.com/questions/34764535/why-cant-matplotlib-plot-in-a-different-thread) — moooeeeep, Oct 20 '21 at 06:41
@moooeeeep Hi, I have see this post and I have try to implement it but I dont know/understand TkAgg to replicate my code. — FERNANDO FUENTES-GUERRA FERNND, Oct 20 '21 at 07:55

score 0 · Answer 1 · answered Oct 20 '21 at 10:44

OK, I have fix it. I have use:

warnings.filterwarnings("ignore")

for warnings and:

ani = FuncAnimation(plt.gcf(), animate, interval=0.013)

to make the animation like before. Now I have a problem with check is queue is empty, if I use:

if not queue.empty():

Dont work, and if I dont use it the program stop bad. I can post the code if u want. Thanks to all.

Plotting inside a thread and update it (python)

1 Answers1