@@ -221,19 +221,21 @@ Classification functions
221221 ``False `` otherwise.
222222
223223 Whether or not two values are considered close is determined according to
224- given absolute and relative tolerances.
224+ given absolute and relative tolerances. If no errors occur, the result will
225+ be: ``abs(a-b) <= max(rel_tol * max(abs(a), abs(b)), abs_tol) ``.
225226
226227 *rel_tol * is the relative tolerance -- it is the maximum allowed difference
227228 between *a * and *b *, relative to the larger absolute value of *a * or *b *.
228229 For example, to set a tolerance of 5%, pass ``rel_tol=0.05 ``. The default
229230 tolerance is ``1e-09 ``, which assures that the two values are the same
230- within about 9 decimal digits. *rel_tol * must be greater than zero.
231-
232- *abs_tol * is the minimum absolute tolerance -- useful for comparisons near
233- zero. *abs_tol * must be at least zero.
234-
235- If no errors occur, the result will be:
236- ``abs(a-b) <= max(rel_tol * max(abs(a), abs(b)), abs_tol) ``.
231+ within about 9 decimal digits. *rel_tol * must be nonnegative and less
232+ than ``1.0 ``.
233+
234+ *abs_tol * is the absolute tolerance; it defaults to ``0.0 `` and it must be
235+ nonnegative. When comparing ``x `` to ``0.0 ``, ``isclose(x, 0) `` is computed
236+ as ``abs(x) <= rel_tol * abs(x) ``, which is ``False `` for any ``x `` and
237+ rel_tol less than ``1.0 ``. So add an appropriate positive abs_tol argument
238+ to the call.
237239
238240 The IEEE 754 special values of ``NaN ``, ``inf ``, and ``-inf `` will be
239241 handled according to IEEE rules. Specifically, ``NaN `` is not considered
0 commit comments