known_bugs.py

import numpy

class adouble:
    def __init__(self,x):
       self.x = x
    
    
    def __mul__(self, rhs):
       print 'called __mul__'
       if isinstance(rhs, adouble):
           print 'case adouble'
           return adouble(self.x * rhs.x)
       elif numpy.isscalar(rhs):
           print 'case scalar'
           return adouble(self.x * rhs)
       elif isinstance(rhs, numpy.ndarray):
           print 'case ndarray'
           return rhs * self

    def __rmul__(self, lhs):
       print 'called __rmul__'
       return self * lhs
    
    def __str__(self):
       return str(self.x)+'a'
    
    def __repr__(self):
       return str(self)

print 'numpy.__version__=',numpy.__version__,'\n'

x  = numpy.ones(2,dtype=float)
z = numpy.ones(2,dtype=float)
y = numpy.ones(2,dtype=object)


a = adouble(2)

print 'executing a *= x'
a *= x
print a,' where expected [2.0a, 2.0a]\n'

print 'executing y *= a'
y *= a
print y,' where expected [2.0a, 2.0a]\n'

print 'executing z = z * a'
z = z * a
print z,' where expected [2.0a, 2.0a]\n'

print 'executing x *= a'
x *= a
print x,' where expected [2.0a, 2.0a]\n'