浮点数超大数值处理及除法结果相等性比较异常问题咨询

It’s a classic floating-point precision gotcha! Let’s break down exactly what’s happening here and how to fix it.

The Root Cause #

Floating-point numbers (like Python’s float) rely on binary representation, which can’t perfectly capture every decimal fraction—especially extremely small or large values. Here’s the step-by-step breakdown:

1. Your tiny denominator isn’t exact: When you write 0.00000000000000000000007, Python converts this to the closest possible 64-bit float. That stored value is actually slightly larger than the exact 7e-23 you intended.
2. Division gives a near-but-not-exact result: Dividing 7 by this approximated denominator produces a value that’s very close to 1e+23, but not exactly equal to the literal 1e+23 or your variable tt (which holds the exact float representation of 1e+23).
3. String output hides the difference: When printed, Python rounds floats to a human-readable 15 significant digits. Both values look like 1e+23 in output, but their underlying binary values are subtly different.

To confirm this, let’s peek at the exact values using Python’s decimal module:

import decimal

# Exact value of your tiny denominator as a float
print(decimal.Decimal(0.00000000000000000000007))
# Output: 7.00000000000000020816681711721685132943093776702880859375E-23

# Exact value of temp_ans
temp_ans = 7 / 0.00000000000000000000007
print(decimal.Decimal(temp_ans))
# Output: 99999999999999991611392.0

# Exact value of 1e+23 as a float
print(decimal.Decimal(1e+23))
# Output: 100000000000000000000000.0

As you can see, temp_ans is a tiny bit smaller than 1e+23—which is why the equality checks fail.

How to Fix This #

Instead of checking for exact equality with ==, compare if the values are within a small tolerance (called epsilon) of each other. This is the standard practice for floating-point comparisons.

Example:

temp_ans = 7 / 0.00000000000000000000007
tt = 1e+23

# Choose a small epsilon (adjust based on your precision needs)
epsilon = 1e-9

if abs(tt - temp_ans) < epsilon:
    print("YES1")
if abs(1e+23 - temp_ans) < epsilon:
    print("YES2")
if tt == 1e+23:
    print("YES3")

This will output all three "YES" messages as you expected.

Alternatively, if you need precise decimal arithmetic (no floating-point approximations), use Python’s decimal module to work with exact decimal values instead of floats.

内容的提问来源于stack exchange，提问作者Rahul Verma