numpyProblem3.py

#!/usr/bin/python

from numpy import *

x = random.random([10,3])
x[range(10), fabs(x-0.5).argsort()[:,0]]


#Break it down niao
x1 = fabs(x-0.5)     # numbers closest to 0.5 will be closest to 0 now.
x2 = x1.argsort()    # returns the indexes of each sorted *row*, numbers closest to 0 are first
x3 = x2[:,0]         # give me just the index of the smallest number
x4 = x[range(10),x3] # index back into the original array using the row information you've gathered