[Numpy-discussion] How exactly ought 'dot' to work?

Alan G Isaac alan.isaac at gmail.com
Sat Feb 22 18:30:26 EST 2014


It would help me follow this discussion if it were broken up
into pieces:

- what is being asserted as first principles for `dot`
   (e.g., what mathematical understanding)?
- to what extent do other important implementations
   (e.g., Mathematica and Julia) deviate from the
   proposed mathematical understanding?
- were the motivations for any deviations adequate
   (e.g., supported by strong use cases)?

An example is the discussion of whether scalar
multiplication of a matrix should be represented
by * or by a new operator (e.g., @).  I am
personally most comfortable with the idea that
a new matrix multiplication operator would not handle
scalar multiplication or violate tensor product rules
(perhaps I am influenced by Mathematica), but I am not
prepared to argue the principles of such choices, and
would appreciate hearing from those who are.

Thanks,
Alan Isaac





More information about the NumPy-Discussion mailing list