Python matplotlib showing jagged graphs for constant value functions

Question

I was trying to plot a function with absolute values and it's piecewise counterpart, but matplotlib was ploting the function with jagged corners.

Here is my full code:

import matplotlib.pyplot as plt
import numpy as np

plt.style.use("dark_background")

def f(x):
    return abs(2 * x - 3) + abs(3 * x + 1) - abs(5 * x - 3)

def g(x):
    if x < -1/3:
        return -1
    if -1/3 < x < 3/5:
        return 6 * x + 1
    if 3/5 < x < 3/2:
        return -4 * x + 7
    if x > 3/2:
        return 1

gf = np.vectorize(g)
x = np.linspace(-1, 2, 300)

plt.figure(figsize = (10, 4))

plt.subplot(1, 2, 1)
plt.plot(x, f(x), color = "blue")

plt.subplot(1, 2, 2)
plt.plot(x, gf(x), color = "red")

plt.show()

And here is the plot that was generated: Plot of the functions

The solution to this problem I've found was to add a small of x to the constant output:

def g(x):
    if x < -1/3:
        return -1 + 0.000001 * x
    if -1/3 < x < 3/5:
        return 6 * x + 1
    if 3/5 < x < 3/2:
        return -4 * x + 7
    if x > 3/2:
        return 1 + 0.000001 * x

This solved my problem, but I still don't know why it works, or what's wrong with the first plot I've made.

Fixed jagged plot of piecewise function

If anyone knows a simpler solution or the source of the problem, let me know.

jared · Accepted Answer · 2025-09-23 18:34:11Z

This is caused by np.vectorize. Instead, just use np.piecewise.

def g(x):
    return np.piecewise(x,
                        [x < -1/3, 
                         (-1/3 < x) & (x < 3/5), 
                         (3/5 < x) & (x < 3/2), 
                         (x > 3/2)],
                        [lambda x: -np.ones_like(x), 
                         lambda x: 6*x + 1, 
                         lambda x: -4*x + 7, 
                         lambda x: np.ones_like(x)])

You can then just do g(x) for plotting.

As for why np.vectorize is causing a problem here, it's because it infers the output type from the first input and it's incorrectly infering an integer output (if you check gf(x).dtype you'll see it is a vector of integers). If you feel strongly about using np.vectorize, make sure to specify the output type by doing gf = np.vectorize(g, otypes=[float]).

JRiggles · Accepted Answer · 2025-09-23 17:50:43Z

This may not exactly be the answer, but converting the values returned by g to floats also seems to work:

import matplotlib.pyplot as plt
import numpy as np

plt.style.use("dark_background")

def f(x):
    return abs(2 * x - 3) + abs(3 * x + 1) - abs(5 * x - 3)

def g(x):
    if x < -1/3:
        return -1.0  # note the additional decimal place
    if -1/3 < x < 3/5:
        return float(6 * x + 1)
    if 3/5 < x < 3/2:
        return float(-4 * x + 7)
    if x > 3/2:
        return 1.0  # note the additional decimal place

gf = np.vectorize(g)
x = np.linspace(-1, 2, 300)

plt.figure(figsize = (10, 4))

plt.subplot(1, 2, 1)
plt.plot(x, f(x), color = "blue")

plt.subplot(1, 2, 2)
plt.plot(x, gf(x), color = "red")

plt.show()

That said, there may be a more elegant solution...I'm just not sure what it is offhand.

FWIW: your approach (i.e. adding 0.000001 * x to each value) also works because adding that small decimal value forces the return values of g into float; it's necessarily less accurate than directly using float(), but that's why it still mostly does what you expect!

Collectives™ on Stack Overflow

Python matplotlib showing jagged graphs for constant value functions

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related