4

I created a widget using interact() in the Jupyter notebook which plots the data (see code below). However, the widget updates very slowly (around 1 fps) when I move the slider.

Evaluation of the data inside the functions neg(T) and pos(T) takes around 1e-4 seconds and plotting of the data inside the function conc(T) - around 1e-1 seconds. Then why does my widget have such a slow response? How could I improve it?

from scipy import constants
from numpy import *
import matplotlib.pyplot as plt
from ipywidgets import interact
import time

%matplotlib inline

m_e = m_h = 1.5 * 10**-49
hbar = constants.hbar
k = constants.k / constants.e

E_g = 1.1242
E_v = 0
E_c = E_v + E_g
E_a = E_v + 0.045
E_d = E_c - 0.045
n_a = 10**14
n_d = 10**12
g_a = 4
g_d = 2

E_f = linspace(E_v, E_c, 100)

def neg(T):
    start = time.time()

    N_c = 2 * 10**-6 * (m_e*k*T/(2*pi*hbar**2))**1.5
    n = N_c * exp((E_f-E_c)/(k*T))
    N_a_minus = n_a/(1+g_a*exp((E_a-E_f)/(k*T)))

    print(time.time()-start)

    return n + N_a_minus

def pos(T):
    start = time.time()

    N_v = 2 * 10**-6 * (m_h*k*T/(2*pi*hbar**2))**1.5
    p = N_v * exp((E_v-E_f)/(k*T))
    N_d_plus = n_d/(1+g_d*exp((E_f-E_d)/(k*T)))

    print(time.time()-start)

    return p + N_d_plus

def conc(T):
    start = time.time()

    plt.plot(E_f, neg(T), E_f, pos(T))
    plt.axis([E_v,E_g,10.0**6,10.0**21])
    plt.yscale('log')

    print(time.time()-start)

interact(conc, T=(50,800,50));
Pavlo Fesenko
  • 187
  • 9
  • Do you lose most of the time in evaluating your function, or in the plotting itself? – cel Jan 27 '16 at 06:30
  • @cel I have rephrased my question and hopefully it's more clear now. Could you suggest how I could improve the response of my widget? Thanks! – Pavlo Fesenko Feb 01 '16 at 05:45
  • Actually, I don't know. Maybe the response time of the notebook is to blame - you could try asking the developers in jupyter's gitter channel. – cel Feb 01 '16 at 05:56

0 Answers0