sigfig - Scientific Rounding

1.3.19 · active · verified Thu Apr 16

sigfig is a Python library (current version 1.3.19) designed for precise rounding of numbers based on significant figures, decimal places, or uncertainty. It also provides versatile formatting options and can interpret various numeric input types (e.g., strings, floats, Decimals). The library aims to provide expected rounding results in scientific and engineering contexts, often differing from Python's built-in `round()` function. It is actively maintained with releases as recent as March 2025.

Common errors

```
UserWarning: warning: X significant figures requested from number with only Y significant figures
```
cause You are attempting to round a number to more significant figures than it inherently possesses or than `sigfig` can infer from its type. For example, passing the integer `1` and requesting `sigfigs=3`.

fix Provide the number with sufficient precision (e.g., as a string `'1.00'`). Alternatively, if the intent is to enforce a minimum number of significant figures, suppress the warning: `from sigfig import round; round(1, sigfigs=3, warn=False)`.
```
Incorrect rounding or significant figure count for numbers with trailing zeros (e.g., 10.0, 1200)
```
cause Python's `float` type often discards trailing zeros that are significant in scientific notation (e.g., `float('10.00')` might be `10.0`). When `sigfig` receives such a float, it loses the original precision information.

fix Always pass numbers as strings (e.g., `'10.00'`, `'1200.'`) or `decimal.Decimal` objects to `sigfig.round()` to preserve all significant digits, including trailing zeros. Example: `from sigfig import round; round('10.00', sigfigs=3)`.

Warnings

gotcha Python's native float type can lead to precision loss, especially with trailing zeros (e.g., `10.0` might internally be `10.0` with 2 sigfigs, not `10.00` with 3). This can cause `sigfig` to misinterpret the intended precision.
Fix: Pass numbers to `sigfig.round()` as strings (e.g., `'10.00'`) or `decimal.Decimal` objects when precise control over significant figures and trailing zeros is critical.
gotcha Importing `from sigfig import round` directly overwrites the built-in `round()` function. This can lead to unexpected behavior if other parts of your code rely on the standard `round()` behavior.
Fix: Import `sigfig` with an alias (e.g., `import sigfig as sf`) and call its functions explicitly (e.g., `sf.round()`) to avoid conflicts.
gotcha The library may issue a `UserWarning` if you request more significant figures than the input number inherently possesses, indicating potential 'invented numbers'.
Fix: Ensure the input number has sufficient precision (e.g., by providing it as a string with enough digits, like `'1.000'`). If you intentionally want to format a number to a minimum number of significant figures, you can suppress the warning with `warn=False` (e.g., `round(1, sigfigs=3, warn=False)`).

Install

pip install sigfig Install latest version

Imports

round
wrong:
```
round(0.25, 1)
```
correct:
```
from sigfig import round
```
The `sigfig.round` function provides different (often more scientifically accurate) rounding behavior than Python's built-in `round()`, especially concerning decimal places and significant figures. The built-in `round(0.25, 1)` yields `0.2`, while `sigfig.round(0.25, decimals=1)` yields `0.3`.

Quickstart

This quickstart demonstrates how to import `sigfig.round` and use it for common rounding operations: by significant figures, by decimal places, and by uncertainty. It also highlights the recommendation to pass numbers as strings to maintain precision.

from sigfig import round

# Round to 4 significant figures
result_sigfigs = round('123.456', sigfigs=4)
print(f"Rounded to significant figures: {result_sigfigs}")

# Round to 2 decimal places
result_decimals = round('3.14159', decimals=2)
print(f"Rounded to decimal places: {result_decimals}")

# Round by uncertainty
result_uncertainty = round('3.14159', uncertainty=2)
print(f"Rounded by uncertainty: {result_uncertainty}")

view raw JSON →