operator overloading: class*number vs number*class

Hello,

I've defined a class (lets say vector) and I want to be able to overload the following binary operations

1) >> my_class*number

2) >> number*my_class

I have overloaded mtimes, and then case 1) works (I detect the type of the second operator and proceed accordingly), but case 2) does not. In fact the mtimes function of the class does not get called in case 2).

This may seem as logical, but in cannot figure out a work arround for 2).

How do I program behavior for case 2), that is number*my_class ?.

Thanks,

Javier

Answer 1

[Javier Ros]

Solution:

It seems that it is a question of object precedence when invoking the mtimes defined operator.

Changing the definition of my_Class from

classdef  my_Class

to

classdef (InferiorClasses = {?sym}) my_Class

gives precedence to my_Class mtimes method against the one of sym class

[Steven Lord]

That is correct. See the first section on this documentation page for an explanation.

Answer 2

[Javier Ros]

After some more trials I presume this is a BUG related to the symbolic toolbox.

If I do

2.1) >> 5*my_class ---->>>> WORKS it enters the overloaded class function mtimes

2.2) >> syms g real, g*my_class --->>>> FAILS it does not enter the class function mtimes

The error is:

>> g*ez %g is syms g real, and ez is of type my_class

    Error using sym>tomupad (line 1229)
    Unable to convert 'my_class' to 'sym'.

    Error in sym (line 211)
                    S.s = tomupad(x);

    Error in sym/privResolveArgs (line 966)
                        argout{k} = sym(arg);

    Error in sym/privBinaryOp (line 990)
                args = privResolveArgs(A, B);

    Error in  *  (line 312)
            X = privBinaryOp(A, B, 'symobj::mtimes');

[Steven Lord]

No, this is not a bug.

I assume you posted this answer before posting the one that you ultimately accepted, as in the accepted answer you successfully identified how to obtain the behavior you wanted. The comment I posted to that accepted answer points to the documentation page that explains why you see this behavior.