weird behaviour when executing python script

Question

I don't know if any of you have ever had such a behaviour, but after debugging my code I noticed a really weird thing happening to my code. Basically, I am filling a numpy array ('vols') with values contained in a pandas dataframe (with 2 hierarchical columns) at a certain index ('date'), but if I try to execute this piece of code in a script:

   for i in range(capEndDates_size):
      for j in range(strike_size):         
       vols[i,j] = float(df[capEndDates[i]][strikes_list[j]].ix[date])

I cannot properly fill all values of variable 'vols' with those included in the dataframe at index 'date', instead getting some 'nan' where I would expect values. The funny thing is that if I execute the piece of code in the interpreter, this goes the right way (ie by pressing f9 on the editor of Spyder, which is the IDE I'm using)! I found a workaround at this bug by repeating those lines twice (that is, forcing the script to execute the lines two times). My solution is like this:

   for i in range(capEndDates_size):
      for j in range(strike_size):         
       vols[i,j] = float(df[capEndDates[i]][strikes_list[j]].ix[date])

   for i in range(capEndDates_size):
      for j in range(strike_size):         
       vols[i,j] = float(df[capEndDates[i]][strikes_list[j]].ix[date])

Which is really unacceptable. Does any one have an idea of why this is happening?

Answer 1

如果vols[i,j]是NaN，为什么不尝试在分配给vols之后插入硬断点（ import pdb; pdb.set_trace() ） vols[i,j]然后进行调试？