Repository navigation
BUG: Dot product does not check for floating point warning flags #14925
Description
The dot function (unlike ufuncs, and the ufunc based matmul, @) does not check for floating point warning/error flags. It would probably better if it would, although it is likely not high priority. The example being:
np.dot([1,1], [-np.inf, np.inf])
Which should be setting the CPU invalid value produced flag on the NaN result, but it is not caught by NumPy.
(phased out from gh-14788)
Activity
BUG: Check for floating point exceptions in dot
BUG: Check for floating point exceptions in dot
BUG: Check for floating point exceptions in dot
ENH: Check for floating point exceptions in dot (#28442)
ENH: Check for floating point exceptions in dot (numpy#28442)