Approximation of Sine function in Integer Forth

[Thomas Klemm]

(03-12-2022 04:25 PM)Martin Hepperle Wrote:  Maybe there is another more elegant or more accurate approach?

You could use Bhaskara's Sine and Cosine Approximations.
They are accurate to about 1‰.

Functions in Python

Code:

def sin(x):
    u = (180 - x) * x
    return 4000 * u // (40500 - u)

def cos(x):
    u = x**2
    return 4000 * (8100 - u) // (32400 + u)

Plot the Difference

Code:

import matplotlib.pyplot as plt
import numpy as np

x = np.linspace(0, 90, 91)

Sine

Code:

plt.plot(x, sin(x) -  1000 * np.sin(np.radians(x)), 'b')

Cosine

Code:

plt.plot(x, cos(x) -  1000 * np.cos(np.radians(x)), 'r')

Cheers
Thomas