Imprecise Rounding in Python

Question

Is there an easy way to make Python round using 32 bits instead of 64 bits? It would work in the same way as JavaScript's Math.fround, but I've ran circles in my head failing to come up with a solution.

It would help giving examples of what you are trying to do, does this help: https://stackoverflow.com/a/11523128/5236575? — shoaib30, Sep 10 '21 at 06:36
Is there a reason you would need to round from 32 bits instead of 64 bits? — Jason Chia, Sep 10 '21 at 06:41
I'm using a seeded pseudorandom generator from a video game and the values were off because it wasn't being rounded from 32 bits. — TrueGrace, Sep 10 '21 at 12:07

blhsing · Accepted Answer · 2021-09-10T06:55:46.977

3

One quick way to do it is to assign the number to the value of a ctypes.c_float object:

from ctypes import c_float

d = 1.2345678901234567890
f = c_float()
f.value = d
print(d) # 64-bit double
print(f.value) # 32-bit float

This outputs:

1.2345678901234567
1.2345678806304932

Compared to Math.fround in JavaScript:

console.log(Math.fround(1.2345678901234567890))

edited Sep 10 '21 at 06:55

answered Sep 10 '21 at 06:43

blhsing

91,368
6
71
106

score 2 · Answer 2 · answered Sep 10 '21 at 06:55

2

try using numpy float32

np.float32(6.33333333333)

you should read this other question.

answered Sep 10 '21 at 06:55

JPcode

90
7

Unfortunately, np.round() and np.around() don't function the same as Math.fround(). Thank you though! – TrueGrace Sep 10 '21 at 12:00

Imprecise Rounding in Python

2 Answers2