On 6/8/2013 12:16 AM, Tim Chase wrote:
On 2013-06-08 07:04, Carlos Nepomuceno wrote:
alpha, beta = (1 if some_calculation(params) else 0, 1 if
other_calculation(params) else 0)
This one sets them to absolute values, rather than the incrementing
functionality in question:
alpha += temp_a
beta += temp_b
The actual code in question does the initialization outside a loop:
alphas_updated = betas_updated = 0
for thing in bunch_of_things:
a, b = process(thing)
alphas_updated += a
betas_updated += b
I am pretty sure that this is the faster way to get the result.
and it just bugs me as being a little warty for having temp
variables when Python does things like tuple-unpacking so elegantly.
So use it ;-).
That said, as mentioned in a contemporaneous reply to Jason, I haven't
found anything better that is still readable.
The generalization of what you want is a mutable vector with an in-place
augmented assignment version of element by element addition. That
probably has been written. Nnmpy arrays do vector addition and maybe do
it in-place also (I just do ot know). But here is a custom class for the
problem you presented.
def process(val):
return val, 2*val
class Pair(list):
def __init__(self, a, b):
self.append(a)
self.append(b)
def inc(self, a, b):
self[0] += a
self[1] += b
pair = Pair(0, 0)
for val in [1,2,3]:
pair.inc(*process(val))
print(pair)
>>>
[6, 12]
This is cleaner but a bit slower than your in-lined version. I did not
use __iadd__ and += because unpacking 'other' (here the process return)
in the call does the error checking ('exactly two values') for 'free'.
--
Terry Jan Reedy
--
http://mail.python.org/mailman/listinfo/python-list
