Skip to main content

Min Max Position

Summary​

Finding the position of the minimum or maximum value in a given axis

  • tf.argmax()
  • tf.argmin()

Content​

Position of Minimum Value​

X = tf.random.uniform(shape=[50], minval=1, maxval=100)

# to get the position of the minimum value
X, tf.argmin(X), X[tf.argmin(X)]

"""
(<tf.Tensor: shape=(50,), dtype=float32, numpy=
 array([17.734013 , 95.48472  , 58.482388 , 78.88151  , 64.40895  ,
        38.730816 ,  8.319603 , 94.95384  , 52.413155 , 12.9239025,
        86.02268  ,  7.4113903, 67.88889  , 35.025444 , 23.60047  ,
        82.81168  , 57.993065 , 82.2186   , 85.74505  , 54.120884 ,
        69.66284  , 27.195156 , 80.53625  , 31.74381  , 24.519695 ,
        30.33681  , 54.3078   , 89.222694 , 23.813492 , 85.804405 ,
        54.906315 , 66.093094 , 66.674576 , 83.43016  , 58.683666 ,
        96.25571  , 55.63027  , 26.138855 , 65.98061  , 93.25791  ,
        74.37076  , 89.52693  , 71.55578  , 22.188017 , 31.843405 ,
        32.506294 , 93.34212  , 52.360386 , 68.41451  ,  6.997858 ],
       dtype=float32)>,
 <tf.Tensor: shape=(), dtype=int64, numpy=49>,
 <tf.Tensor: shape=(), dtype=float32, numpy=6.997858>)
"""

Position of Maximum Value​

X = tf.random.uniform(shape=[50], minval=1, maxval=100)

X, tf.argmax(X), X[tf.argmax(X)]
"""
(<tf.Tensor: shape=(50,), dtype=float32, numpy=
 array([13.832081 , 34.121017 , 88.644684 , 16.066666 , 98.641365 ,
        48.601196 , 40.79431  , 58.915146 , 16.614832 , 27.061255 ,
        66.605515 , 14.059157 , 68.04469  , 25.745338 , 46.18855  ,
        42.901367 , 68.76748  , 89.02549  , 53.88989  ,  2.517536 ,
        53.744823 , 57.705093 , 30.424484 , 17.591778 , 79.91901  ,
        27.821213 , 61.605904 , 51.709534 , 72.70406  , 70.62744  ,
         9.9101925, 63.1496   , 23.855328 , 71.65855  , 84.96536  ,
        28.362629 , 57.578    , 56.093754 , 63.568623 , 45.199787 ,
        52.54676  , 35.93845  , 44.717003 , 52.629868 , 33.033478 ,
        84.39091  , 37.273    , 63.246048 , 10.915096 , 49.832104 ],
       dtype=float32)>,
 <tf.Tensor: shape=(), dtype=int64, numpy=4>,
 <tf.Tensor: shape=(), dtype=float32, numpy=98.641365>)
"""