Subtract one Pandas Series from another without creating NaN values for missing data

Question

When you subtract one series from another, the result contains NaNs if there is no record in the second series to subtract.

In [1]: import pandas as pd

In [2]: a = pd.Series({1: 100, 2: 102, 3: 103, 4: 104})

In [3]: a
Out[3]:
1    100
2    102
3    103
4    104
dtype: int64

In [4]: b = pd.Series({1: 5, 3: 7})

In [5]: a - b
Out[5]:
1    95.0
2     NaN
3    96.0
4     NaN
dtype: float64

Is there an easier way to do the subtraction with the assumption that missing data is equivalent to zero? Other than by manually inserting zeros like this:

In [15]: b_dash = pd.Series(b, index=a.index).fillna(0.0)

In [16]: a - b_dash
Out[16]:
1     95.0
2    102.0
3     96.0
4    104.0
dtype: float64

Answer 1

Using sub

a.sub(b,fill_value=0)
Out[213]: 
1     95.0
2    102.0
3     96.0
4    104.0
dtype: float64

Answer 2

Alternatively, you can perform the subtraction and then fillna with your base Series (in this case a):

(a-b).fillna(a)

# 1     95.0
# 2    102.0
# 3     96.0
# 4    104.0

Or, you can use pd.Series.update (which will modify a in-place where there's a match with the indexes from a-b):

a.update(a-b)

# 1     95
# 2    102
# 3     96
# 4    104