It's much better to define the methods when you define the class. When I do it your way, MyMatrix([1, 2, 3])*Matrix([1, 2, 3]) gives ShapeError, but when I do it my way, it works just fine.
In fact, I don't understand why your example works at all. You are taking a method on an instance (K), which already has the implicit argument of self, and putting it on the class, MyMatrix, which doesn't. Aaron Meurer On Sun, Feb 2, 2014 at 11:18 AM, Jason Moore <[email protected]> wrote: > Here is one example of class overload that works: > > In [1]: from sympy import Matrix, MatrixSymbol > > In [2]: k = MatrixSymbol('k', 2, 2) > > In [3]: m = MatrixSymbol('m', 2, 2) > > In [4]: K = Matrix(k) > > In [5]: M = Matrix(m) > > In [6]: K * M > Out[6]: > Matrix([ > [k[0, 0]*m[0, 0] + k[0, 1]*m[1, 0], k[0, 0]*m[0, 1] + k[0, 1]*m[1, 1]], > [k[1, 0]*m[0, 0] + k[1, 1]*m[1, 0], k[1, 0]*m[0, 1] + k[1, 1]*m[1, 1]]]) > > In [7]: class MyMatrix(Matrix): > pass > ...: > > In [8]: MyMatrix.__mul__ = K.multiply_elementwise > > In [9]: K = MyMatrix(k) > > In [10]: M = MyMatrix(m) > > In [11]: K * M > Out[11]: > Matrix([ > [k[0, 0]*m[0, 0], k[0, 1]*m[0, 1]], > [k[1, 0]*m[1, 0], k[1, 1]*m[1, 1]]]) > > > > Jason > moorepants.info > +01 530-601-9791 > > > On Sun, Feb 2, 2014 at 12:16 PM, Aaron Meurer <[email protected]> wrote: >> >> Or, if you don't use matrix multiplication at all, then you could >> overload __mul__. >> >> And I forgot to mention that you'll want to overload both __op__ and >> __rop__ for whatever operator you choose. >> >> Aaron Meurer >> >> On Sun, Feb 2, 2014 at 11:15 AM, Aaron Meurer <[email protected]> wrote: >> > To elaborate on what Sergey is saying, you need to understand that >> > SymPy is just Python, so it is limited by what is available in Python. >> > .* is not a valid operator in Python, so there is no way to make it >> > work. >> > >> > In addition to this, SymPy philosophy is to make the Python operators >> > mean what they mean in normal Python. % is the modulo operator, so it >> > may not make sense to use it for something completely different. On >> > the other hand, I'm not sure if % can ever make sense for matrices, so >> > maybe it would be possible. >> > >> > I would recommend just creating a light subclass of Matrix for your >> > work, and overload an operator to do elementwise multiplication. There >> > is a list of all Python operators here >> > >> > http://docs.python.org/3.3/reference/expressions.html#operator-precedence. >> > You should be aware of the precedence of the operator you choose. % >> > actually probably is a good choice, because it has the same precedence >> > as *. >> > >> > Something like >> > >> > class MyMatrix(Matrix): >> > def __mod__(self, other): >> > return self.multiply_elementwise(other) >> > >> > You'll want to be sure to document this nicely in your thesis, so that >> > people don't think that % is SymPy behavior. >> > >> > If you come across an instance where something returns a normal Matrix >> > instead of a MyMatrix, that's most likely a bug that you should >> > report. >> > >> > Aaron Meurer >> > >> > On Sun, Feb 2, 2014 at 11:04 AM, Sergey Kirpichev <[email protected]> >> > wrote: >> >> On Sunday, February 2, 2014 8:30:45 PM UTC+4, David Kremer wrote: >> >>> >> >>> What I would simply ask, is if it wouldn't be possible to overload an >> >>> operator such as "%" or ".*" to >> >>> implement matrix elementwise multiplication in sympy. It would greatly >> >>> improve the readability of my >> >>> code. >> >> >> >> >> >> First, there is no ".*" operator at all: >> >> http://docs.python.org/dev/reference/lexical_analysis.html#operators >> >> >> >> % - stands for integer division (yields the remainder). While it's >> >> possible >> >> to overload >> >> it, I doubt if that's natural and pythonic (to include in sympy). In >> >> your >> >> words, it's too hackish. >> >> >> >>> >> >>> About the second question, it is about the point if defining a single >> >>> operator such as ".*" in matlab >> >>> could be done in sympy >> >> >> >> >> >> Probably, this can be done only with some syntax extension. >> >> >> >> BTW, see also PEP 225. >> >> >> >> -- >> >> You received this message because you are subscribed to the Google >> >> Groups >> >> "sympy" group. >> >> To unsubscribe from this group and stop receiving emails from it, send >> >> an >> >> email to [email protected]. >> >> To post to this group, send email to [email protected]. >> >> Visit this group at http://groups.google.com/group/sympy. >> >> For more options, visit https://groups.google.com/groups/opt_out. >> >> -- >> You received this message because you are subscribed to the Google Groups >> "sympy" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> To post to this group, send email to [email protected]. >> Visit this group at http://groups.google.com/group/sympy. >> For more options, visit https://groups.google.com/groups/opt_out. > > > -- > You received this message because you are subscribed to the Google Groups > "sympy" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at http://groups.google.com/group/sympy. > For more options, visit https://groups.google.com/groups/opt_out. -- You received this message because you are subscribed to the Google Groups "sympy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/sympy. For more options, visit https://groups.google.com/groups/opt_out.
