On Sat, Mar 15, 2014 at 12:40 PM, Nathaniel Smith <n...@pobox.com> wrote:
> On Sat, Mar 15, 2014 at 6:33 PM, Joe Kington <joferking...@gmail.com> > wrote: > > On Sat, Mar 15, 2014 at 1:28 PM, Nathaniel Smith <n...@pobox.com> wrote: > >> > >> On Sat, Mar 15, 2014 at 3:41 AM, Nathaniel Smith <n...@pobox.com> wrote: > >> > Hi all, > >> > > >> > Here's the main blocker for adding a matrix multiply operator '@' to > >> > Python: > >> > we need to decide what we think its precedence and associativity > should > >> > be. > >> > >> Another data point that might be useful: > >> > >> Matlab: same-left > >> > >> > >> R: tight-left > > > > > > > > I was going to ask this earlier, but I was worried I was missing > something > > major. > > > > Why was "tight-left" not an option? > > > > > > This means that if you don't use parentheses, you get: > > a @ b @ c -> (a @ b) @ c > > a * b @ c -> a * (b @ c) > > a @ b * c -> (a @ b) * c > > > > > > In my (very inexperienced) opinion, it seems like the most intuitive > option. > > Because tight-left doesn't seem to have much to recommend it over > same-left, and all else being equal having fewer levels of precedence > is usually considered a good thing. Unless I'm missing something. If > we do decide that tight-left is best then we could certainly advocate > for it. > > I wouldn't read too much into R's choice; they don't actually define a > separate precedence level for matrix multiplication specifically. They > have a single precedence level for all "special" (user-defined) > operators, and matrix multiplication happens to be one of these. > (Their versions of // and % are also "special", but I don't think > anyone would expect // to bind more tightly than / if one were > choosing precedences on a case-by-case basis.) > > Just to throw something new into the mix u@v@w = u@(v@w) -- u@v is a dyadic matrix u@v -- is a scalar It would be nice if u@v@None, or some such, would evaluate as a dyad. Or else we will still need the concept of row and column 1-D matrices. I still think v.T should set a flag so that one can distinguish u@v.T (dyad) from u.T@v (inner product), where 1-D arrays are normally treated as column vectors. Chuck
_______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion