Una pequeña comparación de velocidad:
Python 2.7.11 |Anaconda 2.4.1 (64-bit)| (default, Dec 7 2015, 14:10:42) [MSC v.1500 64 bit (AMD64)] on win32
In[1]: l = [0,1,2,3,2,3,1,2,0]
In[2]: m = {0:10, 1:11, 2:12, 3:13}
In[3]: %timeit [m[_] for _ in l] # list comprehension
1000000 loops, best of 3: 762 ns per loop
In[4]: %timeit map(lambda _: m[_], l) # using 'map'
1000000 loops, best of 3: 1.66 µs per loop
In[5]: %timeit list(m[_] for _ in l) # a generator expression passed to a list constructor.
1000000 loops, best of 3: 1.65 µs per loop
In[6]: %timeit map(m.__getitem__, l)
The slowest run took 4.01 times longer than the fastest. This could mean that an intermediate result is being cached
1000000 loops, best of 3: 853 ns per loop
In[7]: %timeit map(m.get, l)
1000000 loops, best of 3: 908 ns per loop
In[33]: from operator import itemgetter
In[34]: %timeit list(itemgetter(*l)(m))
The slowest run took 9.26 times longer than the fastest. This could mean that an intermediate result is being cached
1000000 loops, best of 3: 739 ns per loop
Entonces, la comprensión de la lista y la selección de elementos son las formas más rápidas de hacer esto.
ACTUALIZACIÓN: Para grandes listas aleatorias y mapas tuve resultados un poco diferentes:
Python 2.7.11 |Anaconda 2.4.1 (64-bit)| (default, Dec 7 2015, 14:10:42) [MSC v.1500 64 bit (AMD64)] on win32
In[2]: import numpy.random as nprnd
l = nprnd.randint(1000, size=10000)
m = dict([(_, nprnd.rand()) for _ in range(1000)])
from operator import itemgetter
import operator
f = operator.itemgetter(*l)
%timeit f(m)
%timeit list(itemgetter(*l)(m))
%timeit [m[_] for _ in l] # list comprehension
%timeit map(m.__getitem__, l)
%timeit list(m[_] for _ in l) # a generator expression passed to a list constructor.
%timeit map(m.get, l)
%timeit map(lambda _: m[_], l)
1000 loops, best of 3: 1.14 ms per loop
1000 loops, best of 3: 1.68 ms per loop
100 loops, best of 3: 2 ms per loop
100 loops, best of 3: 2.05 ms per loop
100 loops, best of 3: 2.19 ms per loop
100 loops, best of 3: 2.53 ms per loop
100 loops, best of 3: 2.9 ms per loop
Así que en este caso el ganador es clara f = operator.itemgetter(*l); f(m)
, y forastero claro: map(lambda _: m[_], l)
.
ACTUALIZACIÓN para Python 3.6.4:
import numpy.random as nprnd
l = nprnd.randint(1000, size=10000)
m = dict([(_, nprnd.rand()) for _ in range(1000)])
from operator import itemgetter
import operator
f = operator.itemgetter(*l)
%timeit f(m)
%timeit list(itemgetter(*l)(m))
%timeit [m[_] for _ in l] # list comprehension
%timeit list(map(m.__getitem__, l))
%timeit list(m[_] for _ in l) # a generator expression passed to a list constructor.
%timeit list(map(m.get, l))
%timeit list(map(lambda _: m[_], l)
1.66 ms ± 74.2 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)
2.1 ms ± 93.2 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
2.58 ms ± 88.8 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
2.36 ms ± 60.7 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
2.98 ms ± 142 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
2.7 ms ± 284 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
3.14 ms ± 62.6 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
Entonces, los resultados para Python 3.6.4 son casi los mismos.
mydict
es una llamada de función (que devuelve un dict), entonces llama a la función varias veces, ¿verdad?