numpy matrix subtraction gives very small number where it should give zero

Asked Nov 07 '20 at 16:01

Active Nov 07 '20 at 16:01

Viewed 149 times

I have the following code:

A = np.random.randint(-10,10,(4,4))
B = np.random.randint(-10,10,(4,4))
A1, A2, A3, A4 = inv(A@B), inv(A)@inv(B), inv(B@A), inv(B)@inv(A)
print(A1 - A4)

since mathematically A1 = A4, subtracting them should give back zero matrix, but I get

[[-1.73472348e-18  1.73472348e-18  3.46944695e-18  3.46944695e-18]
 [-8.67361738e-19  0.00000000e+00  4.01154804e-18  3.46944695e-18]
 [ 3.46944695e-18  1.73472348e-18  3.46944695e-18 -3.46944695e-18]
 [-1.73472348e-18 -1.73472348e-18 -1.73472348e-18  1.73472348e-18]]

its very close to zero, but not really zero - why is that?

asked Nov 07 '20 at 16:01

moran Reznik

1

My guess would be floating point precision https://docs.python.org/3/tutorial/floatingpoint.html – Leonard Brünings Nov 07 '20 at 16:04
1

Does this answer your question? [Is floating point math broken?](https://stackoverflow.com/questions/588004/is-floating-point-math-broken) – fdermishin Nov 07 '20 at 18:24

numpy matrix subtraction gives very small number where it should give zero

0 Answers0