matplotlib/matplotlib · error · ValueError

Axis limits cannot be NaN or Inf

Error message

Axis limits cannot be NaN or Inf

What it means

When view limits are set, matplotlib converts each limit to a number and rejects non-finite Real values (NaN, +Inf, -Inf), because such limits cannot define a usable view interval. Conversion (e.g. datetime to float) happens first, so only numeric results that are NaN or Inf after conversion raise this ValueError.

Source

Thrown at lib/matplotlib/axes/_base.py:3844

        return tuple(self.viewLim.intervalx)

    def _validate_converted_limits(self, limit, convert):
        """
        Raise ValueError if converted limits are non-finite.

        Note that this function also accepts None as a limit argument.

        Returns
        -------
        The limit value after call to convert(), or None if limit is None.
        """
        if limit is not None:
            converted_limit = convert(limit)
            if isinstance(converted_limit, np.ndarray):
                converted_limit = converted_limit.squeeze()
            if (isinstance(converted_limit, Real)
                    and not np.isfinite(converted_limit)):
                raise ValueError("Axis limits cannot be NaN or Inf")
            return converted_limit

    def set_xlim(self, left=None, right=None, *, emit=True, auto=False,
                 xmin=None, xmax=None):
        """
        Set the x-axis view limits.

        Parameters
        ----------
        left : float, optional
            The left xlim in data coordinates. Passing *None* leaves the
            limit unchanged.

            The left and right xlims may also be passed as the tuple
            (*left*, *right*) as the first positional argument (or as
            the *left* keyword argument).

            .. ACCEPTS: (left: float, right: float)

View on GitHub (pinned to b379c1b69e)

Solutions

1. Use NaN-aware aggregates: ax.set_xlim(np.nanmin(d), np.nanmax(d)).
2. Filter invalid values before plotting: d = d[np.isfinite(d)].
3. Validate computed limits: if not np.isfinite(lo) or not np.isfinite(hi), fall back to ax.relim(); ax.autoscale_view() or sensible defaults.

Example fix

# before
ax.set_xlim(d.min(), d.max())  # d contains NaN

# after
ax.set_xlim(np.nanmin(d), np.nanmax(d))

import numpy as np

lo, hi = compute_limits(data)
if not (np.isfinite(lo) and np.isfinite(hi)):
    lo, hi = np.nanmin(data), np.nanmax(data)  # or sensible defaults
ax.set_xlim(lo, hi)

try:
    ax.set_xlim(lo, hi)
except ValueError as e:
    if 'NaN or Inf' in str(e):
        finite = values[np.isfinite(values)]
        ax.set_xlim(finite.min(), finite.max())
    else:
        raise

When it happens

Trigger: ax.set_xlim(float('nan'), 1) or ax.set_ylim(np.inf, 5); most commonly limits computed from data: d.min()/d.max() on arrays containing NaN; 0/0 or x/0 producing inf/nan in limit arithmetic; limits taken from empty aggregates (np.min([]) -> nan with warning).

Common situations: Real-world datasets with missing values fed straight into limit computation; empty slices after filtering returning nan; unit conversions or scalings introducing inf; interactive apps computing limits from unvalidated user ranges.

Related errors

• Can only output finite numbers in PDF
• Wedge sizes must be finite numbers
• ecdf() does not support NaNs
• aspect must be finite and positive
• z array must not contain non-finite values within the triang

AI-assisted analysis of matplotlib/matplotlib@b379c1b69e (2026-08-21). Data as JSON: /api/errors/f658884c941cab8f. Report an issue: GitHub.