*   <li>b → 3
 * </ul>
 *
 * ... or as a single "flattened" collection of key-value pairs:
 *
 * <ul>
 *   <li>a → 1
 *   <li>a → 2
 *   <li>b → 3
 * </ul>
 *
 * <p><b>Important:</b> although the first interpretation resembles how most multimaps are
 * <i>implemented</i>, the design of the {@code Multimap} API is based on the <i>second</i> form.

   * #MANY_VALUES_MEAN} and the sum-of-squares-of-deltas is {@link
   * #MANY_VALUES_SUM_OF_SQUARES_OF_DELTAS}. The smallest and largest finite values are always
   * {@link #MANY_VALUES_MIN} and {@link #MANY_VALUES_MAX}, although setting non-finite values will
   * change the true min and max.
   */
  static class ManyValues {

    private final ImmutableList<Double> values;

    ManyValues(double[] values) {

            item_def["description"] = item_description
    # definitions: dict[DefsRef, dict[str, Any]]
    # but mypy complains about general str in other places that are not declared as
    # DefsRef, although DefsRef is just str:
    # DefsRef = NewType('DefsRef', str)
    # So, a cast to simplify the types here
    return field_mapping, cast(dict[str, dict[str, Any]], definitions)

    externalDocs: ExternalDocumentation | None = None
    example: Annotated[
        Any | None,
        typing_deprecated(
            "Deprecated in OpenAPI 3.1.0 that now uses JSON Schema 2020-12, "
            "although still supported. Use examples instead."
        ),
    ] = None


# Ref: https://json-schema.org/draft/2020-12/json-schema-core.html#name-json-schema-documents
# A JSON Schema MUST be an object or a boolean.

          .isWithin(ALLOWED_ERROR)
          .of(expectedLargeDatasetPercentile(index));
    }

    // Assert that the dataset contains the same elements after the in-place computation (although
    // they may be reordered). We only do this for one index rather than for all indexes, as it is
    // quite expensive (quadratic in the size of PSEUDORANDOM_DATASET).

          .isWithin(ALLOWED_ERROR)
          .of(expectedLargeDatasetPercentile(index));
    }

    // Assert that the dataset contains the same elements after the in-place computation (although
    // they may be reordered). We only do this for one index rather than for all indexes, as it is
    // quite expensive (quadratic in the size of PSEUDORANDOM_DATASET).

   * latter case, tasks will run serially on the calling thread. Tasks are run to completion before
   * a {@code Future} is returned to the caller (unless the executor has been shutdown).
   *
   * <p>Although all tasks are immediately executed in the thread that submitted the task, this
   * {@code ExecutorService} imposes a small locking overhead on each task submission in order to
   * implement shutdown and termination behavior.

   * #MANY_VALUES_MEAN} and the sum-of-squares-of-deltas is {@link
   * #MANY_VALUES_SUM_OF_SQUARES_OF_DELTAS}. The smallest and largest finite values are always
   * {@link #MANY_VALUES_MIN} and {@link #MANY_VALUES_MAX}, although setting non-finite values will
   * change the true min and max.
   */
  static class ManyValues {

    private final ImmutableList<Double> values;

    ManyValues(double[] values) {

    * You can also [follow @fastapi on X (Twitter)](https://x.com/fastapi) (a separate account).
* [Follow me on **LinkedIn**](https://www.linkedin.com/in/tiangolo/).
    * Hear when I make announcements or release new tools (although I use X (Twitter) more often 🤷‍♂).
* Read what I write (or follow me) on [**Dev.to**](https://dev.to/tiangolo) or [**Medium**](https://medium.com/@tiangolo).
    * Read other ideas, articles, and read about tools I have created.

 * appropriate {@code copyOf} method itself.
 *
 * <p>Expressing the immutability guarantee directly in the type that user code references is a
 * powerful advantage. Although Java offers certain immutable collection factory methods, such as
 * {@link Collections#singleton(Object)} and <a
 * href="https://docs.oracle.com/en/java/javase/25/docs/api/java.base/java/util/Set.html#unmodifiable">{@code