Originally posted by: JPRyan

I have a very simple test Pandas DataFrame

    from docplex.mp.advmodel import AdvModel
    from pandas.core.series import Series
    from pandas.core.frame import DataFrame

    mdl = AdvModel(name='Test')
    target = Series(mdl.continuous_var_list(['Asset'], name='Target', lb=0, ub=1), index=['Asset'])
    frame = DataFrame(1, ['Asset'], ['Column'])
    print(frame)
    print(frame * target)

but since I updated to pandas 0.24.1 from 0.23.4 this multiplication no longer works.  I get the error:

DOcplexException: Multiply expects variable, expr or number, array([1.]) was passed (type is <class 'numpy.ndarray'>)

When I revert to 0.23.4 it works again.  Any idea what is going on here?

Originally posted by: Viu-Long Kong

Hi,

Continous variables in docplex override the __rmul__ operator, and it this case, that operator gets a parameter it does not now what to do with. When evaluating frame * target,

it is called with a [NaN] value (while we were expecting [1] or 1, from you frame.

It looks like multiplication of a DataFrame by a Series is not well defined in pandas, leading to this.

Ruling out docplex, if I execute:

f1 = DataFrame(data=[2], index=['Asset'], columns=['Column'])
s1 = Series([7], index=['Asset'])

print(f1*s1)

the result is:

       Asset  Column
Asset    NaN     NaN

I don't know what was your initial intent, but maybe you meant f1.apply(lambda col: col *s1) (or frame.apply(lambda col: col * target) in your example ?)

print(f1.apply(lambda col: col* s1))

       Column
Asset      14

Originally posted by: JPRyan

Thanks for the response.  You are correct, I had the wrong multiplication in my toy problem

The correct call in Pandas would be 

  frame.mul(target, axis=0)

which solved the issue.  Many thanks.