How about using MatrixExpr as an unevaluated proxy, and then substitute ImmutibleMatrix when you want to evaluate it? It really seems to me that sums, producs, and powers of ImmutibleMatrix should evaluate by default.
Would you ever want X + X to not reduce to 2*X? That's kind of how the core works. Aaron Meurer On Sun, Mar 11, 2012 at 6:20 PM, Matthew Rocklin <[email protected]> wrote: > @Aaron - Will do. In general my point was that here is a case where we want > to write down X*Y in one way potentially but act on the syntax in two > different ways. We want to build two different mechanisms to consume the > expression X+Y > > Return an unevaluated but simplified Matrix Expr > Return a fully evaluated ImmutableMatrix > > In our current design we have to make a single choice. It'd be nice if we > could plug in either these two choices in a systematic way. It would also be > nice if we could chain them. For example we might want to simplify the X+X > into 2*X before turning it into an ImmutableMatrix. > > @Krastanov - you bring up a good point that it would be nice if > ImmutableMatrix(X*Y) worked. I think that users are more likely to try this > before looking for and finding .as_explicit() > > > > On Sun, Mar 11, 2012 at 3:27 PM, Aaron Meurer <[email protected]> wrote: >> >> On Sun, Mar 11, 2012 at 3:55 PM, Matthew Rocklin <[email protected]> >> wrote: >> > Right i**2 is a MatrixExpr. At some point someone suggested that we have >> > unevaluated dense matrix expressions. This choice was made to satisfy >> > that >> > desire. We can change things around though. >> > >> > In [1]: X = ImmutableMatrix(3,3, range(9)) >> > In [4]: Y = ImmutableMatrix(ones(3, 3)) >> > >> > In [6]: X*Y >> > Out[6]: >> > ⎡0 1 2⎤⋅⎡1 1 1⎤ >> > ⎢ ⎥ ⎢ ⎥ >> > ⎢3 4 5⎥ ⎢1 1 1⎥ >> > ⎢ ⎥ ⎢ ⎥ >> > ⎣6 7 8⎦ ⎣1 1 1⎦ >> > >> > In [7]: (X*Y)[0,0] # indexing works >> > Out[7]: 3 >> > >> > In [8]: (X*Y).as_explicit() >> > Out[8]: >> > ⎡3 3 3 ⎤ >> > ⎢ ⎥ >> > ⎢12 12 12⎥ >> > ⎢ ⎥ >> > ⎣21 21 21⎦ >> > >> > This isn't necessarily the way that things should be done though. One >> > could >> > imagine a number of alternatives. This conversation ties in well to the >> > container vs. canonicalizer idea happening in this thread. >> >> I added stub sections for MatrixExpr and ImmutableMatrix at >> https://github.com/sympy/sympy/wiki/canonicalization. It would be >> great if you could write some stuff there, as I really don't have a >> good idea of what it should be able to do. >> >> Aaron Meurer >> >> > >> > -Matt >> > >> > On Sun, Mar 11, 2012 at 1:18 PM, [email protected] >> > <[email protected]> wrote: >> >> >> >> Actually one can use as_explicit. I did not saw it in the >> >> documentations, only when I checked the code I found about it. >> >> >> >> On 11 March 2012 21:14, [email protected] >> >> <[email protected]> wrote: >> >> > n [8]: i = ImmutableMatrix(eye(3)) >> >> > >> >> > In [9]: i**2 >> >> > Out[9]: >> >> > >> >> > 2 >> >> > ⎡1 0 0⎤ >> >> > ⎢ ⎥ >> >> > ⎢0 1 0⎥ >> >> > ⎢ ⎥ >> >> > ⎣0 0 1⎦ >> >> > >> >> > In [10]: _.doit() >> >> > TypeError >> >> > >> >> > Casting to Matrix does not work either. How to actually calculate the >> >> > product with Immutable matrices? >> >> >> >> -- >> >> You received this message because you are subscribed to the Google >> >> Groups >> >> "sympy" group. >> >> To post to this group, send email to [email protected]. >> >> To unsubscribe from this group, send email to >> >> [email protected]. >> >> For more options, visit this group at >> >> http://groups.google.com/group/sympy?hl=en. >> >> >> > >> > -- >> > You received this message because you are subscribed to the Google >> > Groups >> > "sympy" group. >> > To post to this group, send email to [email protected]. >> > To unsubscribe from this group, send email to >> > [email protected]. >> > For more options, visit this group at >> > http://groups.google.com/group/sympy?hl=en. >> >> -- >> You received this message because you are subscribed to the Google Groups >> "sympy" group. >> To post to this group, send email to [email protected]. >> To unsubscribe from this group, send email to >> [email protected]. >> For more options, visit this group at >> http://groups.google.com/group/sympy?hl=en. >> > > -- > You received this message because you are subscribed to the Google Groups > "sympy" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/sympy?hl=en. -- You received this message because you are subscribed to the Google Groups "sympy" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/sympy?hl=en.
