python 以实际代码学习numpy

import numpy as np

list=[(1,2,3),(4,5,6)]
tuple=(1,2,3,4)

def np_2d_array(list):
    np_array=np.array(list)
    print np_array

def the_2ed_argument_of_array(list):
    np_array=np.array(list,dtype=complex)
    print np_array

#ones() is same as this
def the_use_of_zeros():
    print np.zeros((3,4))

#the initial elements depends on the state of memory
def the_use_of_empty():
    print np.empty((3,4))

#arange reterns arrays but range returns lists
def the_difference_between_arange_and_range():
    print np.arange(10,20,5)
    print range(10,20,5)
    print np.arange(1,3,0.2)                                     # arange() accepts float arguments but not range()

def the_use_of_reshape():
    print np.arange(6)                                           #1 dimensional array
    print np.arange(12).reshape(2,6)                             #2 dimensional array
    print np.arange(24).reshape(2,3,4)                           #3 dimensional array


def the_use_of_asterisk_among_arrays():
    print [[1, 2][3, 4]] * [[5, 6][7, 8]]


def the_use_of_dot(nparray1, nparray2):
    print np.dot(nparray1, nparray2)
    print nparray.dot(nparray2)

# some operations such as *=, += will modify the initial array, others will create a new one
# in numpy, there are some diferences with python27, in former, int + float is illegal, the latter will create a new object
def operations_modify_arrays():
    try:
        a = np.ones((2, 3), dtype=int)
        b = np.random.random((2, 3))
        a *= 3
        b += a
        a += b
    except TypeError as e:
        print e
    try:
        c=int(1)
        d=float(0.1)
        c+=d
        print c
    except TypeError as e:
        print e

# in python, if a list is too large, the system will print the corner of this list
# in numpy, there is a function which can print totally this large array
# more details can read numpy

def axis_parament():
    a=np.random.random((2,3))
    print a.sum()
    print a.min()
    print a.max()

    b=np.arange(12).reshape(3,4)
    print b.sum(axis=0)                                                 # sum of each column
    print b.min(axis=1)                                                 # the min of each row

def universal_functions():
    a=np.array([0,1,2])
    print np.exp(a)                                                     # the n-th of e
    print np.sqrt(a)
    b=np.array([3,4,5])
    print np.add(a,b)

# in numpy, the ndarray is same as lists in python, which can be indexed, sliced and iterated over
# there are codes I'm not familiar to
def slice_ndarray():
    a=np.array([0,1,2,3,4,5,6])
    a[:6:2]=-1000                                                       # from start to position 6, the first element of each two elements is assigned -1000
    print a
    print a[::-1]                                                      # the step is '-'

def the_function_of_fromfunction():
    def f(x,y):
        return 5*x+y
    b=np.fromfunction(f,(5,4),dtype=int)                                # due to the result, the function of this function is to use abscissa axis to be x, the other is y.
    print b
    print b[2,3]                                                        # b[2,3] is same as b[2][3]
    print b[2][3]