python - Fast array/list type for holding numpy ndarrays -

i have lot of list of numpy ndarrays, of different sizes, in code. (i have lists of lists of numpy ndarrays, less issue lists small (<5 elements).)

i have learn ndarrays of ndarrays of different dimention, doesn't workout..

operations tend on it:

• accessing/writing in reverse. (currently writing done reversing list twice)
• appending it
• applying functions on contained ndarrays, eg summing 2 lists of ndarrrays elementwise, multipliplying -1 etc.

i manipulating them following functions (as more direct operations)

def uniop_nested(func,o_list):     def inner(i_list):     ¦   if isinstance(i_list[0],np.ndarray):     ¦   ¦   ¦return map(func, i_list)     ¦   else:     ¦   ¦   ¦return map(inner, i_list)     return inner(o_list)    def binop_nested(func, o1, o2):     if not isinstance(o1,np.ndarray):     ¦   return [binop_nested(func, i1, i2)  (i1,i2) in zip(o1,o2)]     else:     ¦   return func(o1,o2)    def add_nested(s1,s2):     return binop_nested(np.add,s1,s2) 

i'm finding profiling, these lists issue.