mars.tensor.mean¶

mars.tensor.
mean
(a, axis=None, dtype=None, out=None, keepdims=None, combine_size=None)[source]¶ Compute the arithmetic mean along the specified axis.
Returns the average of the array elements. The average is taken over the flattened tensor by default, otherwise over the specified axis. float64 intermediate and return values are used for integer inputs.
 Parameters
a (array_like) – Tensor containing numbers whose mean is desired. If a is not an tensor, a conversion is attempted.
axis (None or int or tuple of ints, optional) –
Axis or axes along which the means are computed. The default is to compute the mean of the flattened array.
If this is a tuple of ints, a mean is performed over multiple axes, instead of a single axis or all the axes as before.
dtype (datatype, optional) – Type to use in computing the mean. For integer inputs, the default is float64; for floating point inputs, it is the same as the input dtype.
out (Tensor, optional) – Alternate output tensor in which to place the result. The default is
None
; if provided, it must have the same shape as the expected output, but the type will be cast if necessary. See doc.ufuncs for details.keepdims (bool, optional) –
If this is set to True, the axes which are reduced are left in the result as dimensions with size one. With this option, the result will broadcast correctly against the input tensor.
If the default value is passed, then keepdims will not be passed through to the mean method of subclasses of Tensor, however any nondefault value will be. If the subclasses sum method does not implement keepdims any exceptions will be raised.
combine_size (int, optional) – The number of chunks to combine.
 Returns
m – If out=None, returns a new tensor containing the mean values, otherwise a reference to the output array is returned.
 Return type
Tensor, see dtype parameter above
Notes
The arithmetic mean is the sum of the elements along the axis divided by the number of elements.
Note that for floatingpoint input, the mean is computed using the same precision the input has. Depending on the input data, this can cause the results to be inaccurate, especially for float32 (see example below). Specifying a higherprecision accumulator using the dtype keyword can alleviate this issue.
By default, float16 results are computed using float32 intermediates for extra precision.
Examples
>>> import mars.tensor as mt
>>> a = mt.array([[1, 2], [3, 4]]) >>> mt.mean(a).execute() 2.5 >>> mt.mean(a, axis=0).execute() array([ 2., 3.]) >>> mt.mean(a, axis=1).execute() array([ 1.5, 3.5])
In single precision, mean can be inaccurate:
>>> a = mt.zeros((2, 512*512), dtype=mt.float32) >>> a[0, :] = 1.0 >>> a[1, :] = 0.1 >>> mt.mean(a).execute() 0.54999924
Computing the mean in float64 is more accurate:
>>> mt.mean(a, dtype=mt.float64).execute() 0.55000000074505806