Some corner cases in the mean, var, std.<br><br><b>Empty arrays</b><br><br>I think these cases should either raise an error or just return nan. Warnings seem ineffective to me as they are only issued once by default.<br><br>
In [3]: ones(0).mean()<br>/home/charris/.local/lib/python2.7/site-packages/numpy/core/_methods.py:61: RuntimeWarning: invalid value encountered in double_scalars<br>  ret = ret / float(rcount)<br>Out[3]: nan<br><br>In [4]: ones(0).var()<br>
/home/charris/.local/lib/python2.7/site-packages/numpy/core/_methods.py:76: RuntimeWarning: invalid value encountered in true_divide<br>  out=arrmean, casting=&#39;unsafe&#39;, subok=False)<br>/home/charris/.local/lib/python2.7/site-packages/numpy/core/_methods.py:100: RuntimeWarning: invalid value encountered in double_scalars<br>
  ret = ret / float(rcount)<br>Out[4]: nan<br><br>In [5]: ones(0).std()<br>/home/charris/.local/lib/python2.7/site-packages/numpy/core/_methods.py:76: RuntimeWarning: invalid value encountered in true_divide<br>  out=arrmean, casting=&#39;unsafe&#39;, subok=False)<br>
/home/charris/.local/lib/python2.7/site-packages/numpy/core/_methods.py:100: RuntimeWarning: invalid value encountered in double_scalars<br>  ret = ret / float(rcount)<br>Out[5]: nan<br><br><b>ddof &gt;= number of elements</b><br>
<br>I think these should just raise errors. The results for ddof &gt;= #elements is happenstance, and certainly negative numbers should never be returned.<br><br>In [6]: ones(2).var(ddof=2)<br>/home/charris/.local/lib/python2.7/site-packages/numpy/core/_methods.py:100: RuntimeWarning: invalid value encountered in double_scalars<br>
  ret = ret / float(rcount)<br>Out[6]: nan<br><br>In [7]: ones(2).var(ddof=3)<br>Out[7]: -0.0<br><b><br>nansum</b><br><br>Currently returns nan for empty arrays. I suspect it should return nan for slices that are all nan, but 0 for empty slices. That would make it consistent with sum in the empty case.<br>
<br>Chuck<br>