Skip to content

fix(maths): correct float rounding in num_digits_fast for large integers - #15754

Open
SaifAli-S wants to merge 1 commit into
TheAlgorithms:masterfrom
SaifAli-S:fix-num-digits-fast-rounding
Open

SaifAli-S wants to merge 1 commit into
TheAlgorithms:masterfrom
SaifAli-S:fix-num-digits-fast-rounding

Conversation

@SaifAli-S

Copy link
Copy Markdown

Describe your change

  • Add an algorithm?
  • Fix a bug or typo in an existing algorithm?
  • Add or change doctests? -- Note: Please avoid changing both code and tests in a single pull request.
  • Documentation change?

Fixes #15753

num_digits_fast() in maths/number_of_digits.py previously evaluated math.floor(math.log10(abs(n)) + 1). For integers where abs(n) >= 10**16 - 1 (exceeding IEEE 754 53-bit float mantissa precision), floating-point rounding near powers of ten caused an off-by-one error (e.g. num_digits_fast(10**16 - 1) evaluated to 17 instead of 16).

This fix verifies the logarithmic estimate against integer powers of ten (10 ** (digits - 1) and 10 ** digits), correcting any floating-point rounding discrepancies while maintaining $O(1)$ fast calculation speed. Added regression doctests covering boundary values ($10^{16}-1$, $10^{16}$, $10^{50}-1$, and negative integers).

Checklist

  • I have read CONTRIBUTING.md.
  • This pull request is all my own work -- I have not plagiarized.
  • I know that pull requests will not be merged if they fail the automated tests.
  • This PR only changes one algorithm file. To ease review, please open separate PRs for separate algorithms.
  • All new Python files are placed inside an existing directory.
  • All filenames are in all lowercase characters with no spaces or dashes.
  • All functions and variable names follow Python naming conventions.
  • All function parameters and return values are annotated with Python type hints.
  • All functions have doctests that pass the automated testing.
  • All new algorithms include at least one URL that points to Wikipedia or another similar explanation.
  • If this pull request resolves one or more open issues, then the description above includes the issue number(s) with a closing keyword: "Fixes #ISSUE-NUMBER".

@algorithms-keeper algorithms-keeper Bot added awaiting reviews This PR is ready to be reviewed enhancement This PR modified some existing files labels Nov 13, 2026
ssilva04 pushed a commit to ssilva04/Python that referenced this pull request Nov 15, 2026
ssilva04 pushed a commit to ssilva04/Python that referenced this pull request Nov 15, 2026

@jgohel9902 jgohel9902 left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the update. I checked the new version against len(str(abs(n))) up to 599 digits and 200,000 random values with no mismatches, so the correction is solid. A few small suggestions: add a 10**15 - 1 doctest, and add a comment explaining why the exact check is needed.

Comment thread maths/number_of_digits.py
4
>>> num_digits_fast(10**15)
16
>>> num_digits_fast(10**16 - 1)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The new doctests look good, including 1050 - 1 and the negative case. One more worth adding is 1015 - 1 (expected 15): in my check the plain floor(log10(n) + 1) formula is already wrong from that value onward, so it is the first value the correction step actually has to fix. Also 10**15 itself is covered by the existing doctest, which is a good neighbour for it.

Comment thread maths/number_of_digits.py
return 1

abs_n = abs(n)
digits = math.floor(math.log10(abs_n) + 1)

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The correction below makes this safe: I compared the function with len(str(abs(n))) for 10k - 1, 10k and 10**k + 1 for every k up to 599, and for 200,000 random values with up to 300 digits, with no mismatches. One small thought: floor(log10(abs_n)) + 1 and floor(log10(abs_n) + 1) can differ by float rounding right before a power of ten, so the code relies entirely on the correction. A one-line comment saying so would help readers.

Comment thread maths/number_of_digits.py

abs_n = abs(n)
digits = math.floor(math.log10(abs_n) + 1)
if abs_n < 10 ** (digits - 1):

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Could you add a short comment above the correction explaining why it exists? For example: "log10 on a float can be off by one near powers of ten, since ints above 2**53 lose precision when converted to float; correct it exactly with integer comparisons." The explanation is easy to lose later, and without it the next reader may remove the check as redundant.

Comment thread maths/number_of_digits.py
1
>>> num_digits_faster(-123456)
6
>>> num_digits_faster('123') # Raises a TypeError for non-integer input

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Out of scope for this PR, but worth a follow-up: after this change num_digits_fast is no longer faster than len(str(abs(n))) in a quick timing (about 3.5x slower on 2**100). Could you paste benchmark() output before and after in the description, so the name still matches the numbers?

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

awaiting reviews This PR is ready to be reviewed enhancement This PR modified some existing files

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Bug: num_digits_fast() returns incorrect digit count for large integers

2 participants