.. _op_ai_onnx_Min:

Min
===

- **Domain**: ``ai.onnx``
- **Since version**: 13

Element-wise min of each of the input tensors (with Numpy-style broadcasting support).
All inputs and outputs must have the same data type.

This operator supports **multidirectional (i.e., Numpy-style) broadcasting**; for more details please check `the doc <Broadcasting.md>`_.

**Inputs**

- **data_0** (*T*): List of tensors for min.

**Outputs**

- **min** (*T*): Output tensor.

**Type Constraints**

- **T**: Constrain input and output types to numeric tensors.
  Allowed types: tensor(bfloat16), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8).

Examples
--------

**test_cc_min_bcast**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(2, 2), dtype=float32
        [[1., 2.],
         [3., 4.]]
      data_1: shape=(), dtype=float32
        2.5

    Outputs:
      result: shape=(2, 2), dtype=float32
        [[1. , 2. ],
         [2.5, 2.5]]

**test_min_example**

.. code-block:: text

    Node:
      Min(data_0, data_1, data_2) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=float32
        [3., 2., 1.]
      data_1: shape=(3,), dtype=float32
        [1., 4., 4.]
      data_2: shape=(3,), dtype=float32
        [2., 5., 0.]

    Outputs:
      result: shape=(3,), dtype=float32
        [1., 2., 0.]

**test_min_float16**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=float16
        [1., 4., 3.]
      data_1: shape=(3,), dtype=float16
        [3., 2., 5.]

    Outputs:
      result: shape=(3,), dtype=float16
        [1., 2., 3.]

**test_min_float32**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=float32
        [3., 2., 1.]
      data_1: shape=(3,), dtype=float32
        [1., 4., 4.]

    Outputs:
      result: shape=(3,), dtype=float32
        [1., 2., 1.]

**test_min_float64**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=float64
        [3., 2., 1.]
      data_1: shape=(3,), dtype=float64
        [1., 4., 4.]

    Outputs:
      result: shape=(3,), dtype=float64
        [1., 2., 1.]

**test_min_int16**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=int16
        [3, 2, 1]
      data_1: shape=(3,), dtype=int16
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=int16
        [1, 2, 1]

**test_min_int32**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=int32
        [3, 2, 1]
      data_1: shape=(3,), dtype=int32
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=int32
        [1, 2, 1]

**test_min_int64**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=int64
        [3, 2, 1]
      data_1: shape=(3,), dtype=int64
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=int64
        [1, 2, 1]

**test_min_int8**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=int8
        [3, 2, 1]
      data_1: shape=(3,), dtype=int8
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=int8
        [1, 2, 1]

**test_min_one_input**

.. code-block:: text

    Node:
      Min(data_0) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=float32
        [3., 2., 1.]

    Outputs:
      result: shape=(3,), dtype=float32
        [3., 2., 1.]

**test_min_two_inputs**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=float32
        [3., 2., 1.]
      data_1: shape=(3,), dtype=float32
        [1., 4., 4.]

    Outputs:
      result: shape=(3,), dtype=float32
        [1., 2., 1.]

**test_min_uint16**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=uint16
        [3, 2, 1]
      data_1: shape=(3,), dtype=uint16
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=uint16
        [1, 2, 1]

**test_min_uint32**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=uint32
        [3, 2, 1]
      data_1: shape=(3,), dtype=uint32
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=uint32
        [1, 2, 1]

**test_min_uint64**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=uint64
        [3, 2, 1]
      data_1: shape=(3,), dtype=uint64
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=uint64
        [1, 2, 1]

**test_min_uint8**

.. code-block:: text

    Node:
      Min(data_0, data_1) -> (result)

.. code-block:: text

    Inputs:
      data_0: shape=(3,), dtype=uint8
        [3, 2, 1]
      data_1: shape=(3,), dtype=uint8
        [1, 4, 4]

    Outputs:
      result: shape=(3,), dtype=uint8
        [1, 2, 1]

Differences with previous version (12)
--------------------------------------

**SchemaDiff**: ``Min`` (domain ``'ai.onnx'``)

* old version: 12
* new version: 13
* breaking: no

**Type constraints:**

* changed 'T': added types: ['tensor(bfloat16)']

Version History
---------------

- :doc:`Version 12 <Min-12>`
- :doc:`Version 8 <Min-8>`
- :doc:`Version 6 <Min-6>`
- :doc:`Version 1 <Min-1>`