您好,登錄后才能下訂單哦!
1. 引言
最近在將一個算法由matlab轉成python,初學python,很多地方還不熟悉,總體感覺就是上手容易,實際上很優雅地用python還是蠻難的。目前為止,覺得就算法仿真研究而言,還是matlab用得特別舒服,可能是比較熟悉的緣故吧。matlab直接集成了很多算法工具箱,函數查詢、調用、變量查詢等非常方便,或許以后用久了python也會感覺很好用。與python相比,最喜歡的莫過于可以直接選中某段代碼執行了,操作方便,python也可以實現,就是感覺不是很方便。
言歸正傳,做算法要用到很多的向量和矩陣運算操作,這些嘛在matlab里面已經很熟悉了,但用python的時候需要用一個查一個,挺煩的,所以在此稍作總結,后續使用過程中會根據使用體驗更新。
python的矩陣運算主要依賴numpy包,scipy包以numpy為基礎,大大擴展了后者的運算能力。
2. 創建一般的多維數組
import numpy as np a = np.array([1,2,3], dtype=int) # 創建1*3維數組 array([1,2,3]) type(a) # numpy.ndarray類型 a.shape # 維數信息(3L,) a.dtype.name # 'int32' a.size # 元素個數:3 a.itemsize #每個元素所占用的字節數目:4 b=np.array([[1,2,3],[4,5,6]],dtype=int) # 創建2*3維數組 array([[1,2,3],[4,5,6]]) b.shape # 維數信息(2L,3L) b.size # 元素個數:6 b.itemsize # 每個元素所占用的字節數目:4 c=np.array([[1,2,3],[4,5,6]],dtype='int16') # 創建2*3維數組 array([[1,2,3],[4,5,6]],dtype=int16) c.shape # 維數信息(2L,3L) c.size # 元素個數:6 c.itemsize # 每個元素所占用的字節數目:2 c.ndim # 維數 d=np.array([[1,2,3],[4,5,6]],dtype=complex) # 復數二維數組 d.itemsize # 每個元素所占用的字節數目:16 d.dtype.name # 元素類型:'complex128'
3. 創建特殊類型的多維數組
a1 = np.zeros((3,4)) # 創建3*4全零二維數組 輸出: array([[ 0., 0., 0., 0.], [ 0., 0., 0., 0.], [ 0., 0., 0., 0.]]) a1.dtype.name # 元素類型:'float64' a1.size # 元素個數:12 a1.itemsize # 每個元素所占用的字節個數:8 a2 = np.ones((2,3,4), dtype=np.int16) # 創建2*3*4全1三維數組 a2 = np.ones((2,3,4), dtype='int16') # 創建2*3*4全1三維數組 輸出: array([[[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]], [[1, 1, 1, 1], [1, 1, 1, 1], [1, 1, 1, 1]]], dtype=int16) a3 = np.empty((2,3)) # 創建2*3的未初始化二維數組 輸出:(may vary) array([[ 1., 2., 3.], [ 4., 5., 6.]]) a4 = np.arange(10,30,5) # 初始值10,結束值:30(不包含),步長:5 輸出:array([10, 15, 20, 25]) a5 = np.arange(0,2,0.3) # 初始值0,結束值:2(不包含),步長:0.2 輸出:array([ 0. , 0.3, 0.6, 0.9, 1.2, 1.5, 1.8]) from numpy import pi np.linspace(0, 2, 9) # 初始值0,結束值:2(包含),元素個數:9 輸出: array([ 0. , 0.25, 0.5 , 0.75, 1. , 1.25, 1.5 , 1.75, 2. ]) x = np.linspace(0, 2*pi, 9) 輸出: array([ 0. , 0.78539816, 1.57079633, 2.35619449, 3.14159265, 3.92699082, 4.71238898, 5.49778714, 6.28318531]) a = np.arange(6) 輸出: array([0, 1, 2, 3, 4, 5]) b = np.arange(12).reshape(4,3) 輸出: array([[ 0, 1, 2], [ 3, 4, 5], [ 6, 7, 8], [ 9, 10, 11]]) c = np.arange(24).reshape(2,3,4) 輸出: array([[[ 0, 1, 2, 3], [ 4, 5, 6, 7], [ 8, 9, 10, 11]], [[12, 13, 14, 15], [16, 17, 18, 19], [20, 21, 22, 23]]])
使用numpy.set_printoptions可以設置numpy變量的打印格式
在ipython環境下,使用help(numpy.set_printoptions)查詢使用幫助和示例
4. 多維數組的基本操作
加法和減法操作要求操作雙方的維數信息一致,均為M*N為數組方可正確執行操作。
a = np.arange(4) 輸出: array([0, 1, 2, 3]) b = a**2 輸出: array([0, 1, 4, 9]) c = 10*np.sin(a) 輸出: array([ 0. , 8.41470985, 9.09297427, 1.41120008]) n < 35 輸出: array([ True, True, True, True], dtype=bool) A = np.array([[1,1],[0,1]]) B = np.array([[2,0],[3,4]]) C = A * B # 元素點乘 輸出: array([[2, 0], [0, 4]]) D = A.dot(B) # 矩陣乘法 輸出: array([[5, 4], [3, 4]]) E = np.dot(A,B) # 矩陣乘法 輸出: array([[5, 4], [3, 4]])
多維數組操作過程中的類型轉換
When operating with arrays of different types, the type of the resulting array corresponds to the more general or precise one (a behavior known as upcasting)
即操作不同類型的多維數組時,結果自動轉換為精度更高類型的數組,即upcasting
a = np.ones((2,3),dtype=int) # int32 b = np.random.random((2,3)) # float64 b += a # 正確 a += b # 錯誤
a = np.ones(3,dtype=np.int32) b = np.linspace(0,pi,3) c = a + b d = np.exp(c*1j) 輸出: array([ 0.54030231+0.84147098j, -0.84147098+0.54030231j, -0.54030231-0.84147098j]) d.dtype.name 輸出: 'complex128'
多維數組的一元操作,如求和、求最小值、最大值等
a = np.random.random((2,3)) a.sum() a.min() a.max() b = np.arange(12).reshape(3,4) 輸出: array([[ 0, 1, 2, 3], [ 4, 5, 6, 7], [ 8, 9, 10, 11]]) b.sum(axis=0) # 按列求和 輸出: array([12, 15, 18, 21]) b.sum(axis=1) # 按行求和 輸出: array([ 6, 22, 38]) b.cumsum(axis=0) # 按列進行元素累加 輸出: array([[ 0, 1, 2, 3], [ 4, 6, 8, 10], [12, 15, 18, 21]]) b.cumsum(axis=1) # 按行進行元素累加 輸出: array([[ 0, 1, 3, 6], [ 4, 9, 15, 22], [ 8, 17, 27, 38]]) universal functions B = np.arange(3) np.exp(B) np.sqrt(B) C = np.array([2.,-1.,4.]) np.add(B,C)
其他的ufunc函數包括:
all, any, apply_along_axis, argmax, argmin, argsort, average, bincount, ceil, clip, conj, corrcoef, cov, cross, cumprod, cumsum, diff, dot, floor,inner, lexsort, max, maximum, mean, median, min, minimum, nonzero, outer, prod, re, round, sort, std, sum, trace, transpose, var,vdot, vectorize, where
5. 數組索引、切片和迭代
a = np.arange(10)**3 a[2] a[2:5] a[::-1] # 逆序輸出 for i in a: print (i**(1/3.))
def f(x,y): return 10*x+y b = np.fromfunction(f,(5,4),dtype=int) b[2,3] b[0:5,1] b[:,1] b[1:3,:] b[-1]
c = np.array([[[0,1,2],[10,11,12]],[[100,101,102],[110,111,112]]]) 輸出: array([[[ 0, 1, 2], [ 10, 11, 12]], [[100, 101, 102], [110, 111, 112]]]) c.shape 輸出: (2L, 2L, 3L) c[0,...] c[0,:,:] 輸出: array([[ 0, 1, 2], [10, 11, 12]]) c[:,:,2] c[...,2] 輸出: array([[ 2, 12], [102, 112]]) for row in c: print(row) for element in c.flat: print(element)
a = np.floor(10*np.random.random((3,4))) 輸出: array([[ 3., 9., 8., 4.], [ 2., 1., 4., 6.], [ 0., 6., 0., 2.]]) a.ravel() 輸出: array([ 3., 9., 8., ..., 6., 0., 2.]) a.reshape(6,2) 輸出: array([[ 3., 9.], [ 8., 4.], [ 2., 1.], [ 4., 6.], [ 0., 6.], [ 0., 2.]]) a.T 輸出: array([[ 3., 2., 0.], [ 9., 1., 6.], [ 8., 4., 0.], [ 4., 6., 2.]]) a.T.shape 輸出: (4L, 3L) a.resize((2,6)) 輸出: array([[ 3., 9., 8., 4., 2., 1.], [ 4., 6., 0., 6., 0., 2.]]) a.shape 輸出: (2L, 6L) a.reshape(3,-1) 輸出: array([[ 3., 9., 8., 4.], [ 2., 1., 4., 6.], [ 0., 6., 0., 2.]])
詳查以下函數:
ndarray.shape, reshape, resize, ravel
6. 組合不同的多維數組
a = np.floor(10*np.random.random((2,2))) 輸出: array([[ 5., 2.], [ 6., 2.]]) b = np.floor(10*np.random.random((2,2))) 輸出: array([[ 0., 2.], [ 4., 1.]]) np.vstack((a,b)) 輸出: array([[ 5., 2.], [ 6., 2.], [ 0., 2.], [ 4., 1.]]) np.hstack((a,b)) 輸出: array([[ 5., 2., 0., 2.], [ 6., 2., 4., 1.]]) from numpy import newaxis np.column_stack((a,b)) 輸出: array([[ 5., 2., 0., 2.], [ 6., 2., 4., 1.]]) a = np.array([4.,2.]) b = np.array([2.,8.]) a[:,newaxis] 輸出: array([[ 4.], [ 2.]]) b[:,newaxis] 輸出: array([[ 2.], [ 8.]]) np.column_stack((a[:,newaxis],b[:,newaxis])) 輸出: array([[ 4., 2.], [ 2., 8.]]) np.vstack((a[:,newaxis],b[:,newaxis])) 輸出: array([[ 4.], [ 2.], [ 2.], [ 8.]]) np.r_[1:4,0,4] 輸出: array([1, 2, 3, 0, 4]) np.c_[np.array([[1,2,3]]),0,0,0,np.array([[4,5,6]])] 輸出: array([[1, 2, 3, 0, 0, 0, 4, 5, 6]])
詳細使用請查詢以下函數:
hstack, vstack, column_stack, concatenate, c_, r_
7. 將較大的多維數組分割成較小的多維數組
a = np.floor(10*np.random.random((2,12))) 輸出: array([[ 9., 7., 9., ..., 3., 2., 4.], [ 5., 3., 3., ..., 9., 7., 7.]]) np.hsplit(a,3) 輸出: [array([[ 9., 7., 9., 6.], [ 5., 3., 3., 1.]]), array([[ 7., 2., 1., 6.], [ 7., 5., 0., 2.]]), array([[ 9., 3., 2., 4.], [ 3., 9., 7., 7.]])] np.hsplit(a,(3,4)) 輸出: [array([[ 9., 7., 9.], [ 5., 3., 3.]]), array([[ 6.], [ 1.]]), array([[ 7., 2., 1., ..., 3., 2., 4.], [ 7., 5., 0., ..., 9., 7., 7.]])]
實現類似功能的函數包括:
hsplit,vsplit,array_split
8. 多維數組的復制操作
a = np.arange(12) 輸出: array([ 0, 1, 2, ..., 9, 10, 11]) not copy at all b = a b is a # True b.shape = 3,4 a.shape # (3L,4L) def f(x) # Python passes mutable objects as references, so function calls make no copy. print(id(x)) # id是python對象的唯一標識符 id(a) # 111833936L id(b) # 111833936L f(a) # 111833936L 淺復制 c = a.view() c is a # False c.base is a # True c.flags.owndata # False c.shape = 2,6 a.shape # (3L,4L) c[0,4] = 1234 print(a) 輸出: array([[ 0, 1, 2, 3], [1234, 5, 6, 7], [ 8, 9, 10, 11]]) s = a[:,1:3] s[:] = 10 print(a) 輸出: array([[ 0, 10, 10, 3], [1234, 10, 10, 7], [ 8, 10, 10, 11]]) 深復制 d = a.copy() d is a # False d.base is a # False d[0,0] = 9999 print(a) 輸出: array([[ 0, 10, 10, 3], [1234, 10, 10, 7], [ 8, 10, 10, 11]])
numpy基本函數和方法一覽
arange, array, copy, empty, empty_like, eye, fromfile, fromfunction, identity, linspace, logspace, mgrid, ogrid, ones, ones_like, r, zeros,zeros_like
Conversions
ndarray.astype, atleast_1d, atleast_2d, atleast_3d, mat
Manipulations
array_split, column_stack, concatenate, diagonal, dsplit, dstack, hsplit, hstack, ndarray.item, newaxis, ravel, repeat, reshape, resize,squeeze, swapaxes, take, transpose, vsplit, vstack
Questionsall, any, nonzero, where
Ordering
argmax, argmin, argsort, max, min, ptp, searchsorted, sort
Operations
choose, compress, cumprod, cumsum, inner, ndarray.fill, imag, prod, put, putmask, real, sum
Basic Statistics
cov, mean, std, var
Basic Linear Algebra
cross, dot, outer, linalg.svd, vdot
完整的函數和方法一覽表鏈接:
https://docs.scipy.org/doc/numpy-dev/reference/routines.html#routines
9. 特殊的索引技巧
a = np.arange(12)**2 輸出: array([ 0, 1, 4, ..., 81, 100, 121]) i = np.array([1,1,3,8,5]) a[i] 輸出: array([ 1, 1, 9, 64, 25]) j = np.array([[3,4],[9,7]]) a[j] 輸出: array([[ 9, 16], [81, 49]]) palette = np.array([[0,0,0],[255,0,0],[0,255,0],[0,0,255],[255,255,255]]) image = np.array([[0,1,2,0],[0,3,4,0]]) palette[image] 輸出: array([[[ 0, 0, 0], [255, 0, 0], [ 0, 255, 0], [ 0, 0, 0]], [[ 0, 0, 0], [ 0, 0, 255], [255, 255, 255], [ 0, 0, 0]]]) i = np.array([[0,1],[1,2]]) j = np.array([[2,1],[3,3]]) a[i,j] 輸出: array([[ 2, 5], [ 7, 11]]) l = [i,j] a[l] 輸出: array([[ 2, 5], [ 7, 11]]) a[i,2] 輸出: array([[ 2, 6], [ 6, 10]]) a[:,j] 輸出: array([[[ 2, 1], [ 3, 3]], [[ 6, 5], [ 7, 7]], [[10, 9], [11, 11]]])
s = np.array([i,j]) print(s) array([[[0, 1], [1, 2]], [[2, 1], [3, 3]]]) a[tuple(s)] 輸出: array([[ 2, 5], [ 7, 11]]) print(tupe(s)) 輸出: (array([[0, 1], [1, 2]]), array([[2, 1], [3, 3]]))
10. 尋找最大值/最小值及其對應索引值
time = np.linspace(20, 145, 5) 輸出: array([ 20. , 51.25, 82.5 , 113.75, 145. ]) data = np.sin(np.arange(20)).reshape(5,4) 輸出: array([[ 0. , 0.84147098, 0.90929743, 0.14112001], [-0.7568025 , -0.95892427, -0.2794155 , 0.6569866 ], [ 0.98935825, 0.41211849, -0.54402111, -0.99999021], [-0.53657292, 0.42016704, 0.99060736, 0.65028784], [-0.28790332, -0.96139749, -0.75098725, 0.14987721]]) ind = data.argmax(axis=0) 輸出: array([2, 0, 3, 1], dtype=int64) time_max = time[ind] 輸出: array([ 82.5 , 20. , 113.75, 51.25]) data_max = data[ind, xrange(data.shape[1])] 輸出: array([ 0.98935825, 0.84147098, 0.99060736, 0.6569866 ]) np.all(data_max == data.max(axis=0)) 輸出: True a = np.arange(5) a[[1,3,4]] = 0 print(a) 輸出: array([0, 0, 2, 0, 0])
a = np.arange(5) a[[0,0,2]] = [1,2,3] print(a) 輸出: array([2, 1, 3, 3, 4]) a = np.arange(5) a[[0,0,2]] += 1 print(a) 輸出: array([1, 1, 3, 3, 4])
a = np.arange(12).reshape(3,4) b = a > 4 輸出: array([[False, False, False, False], [False, True, True, True], [ True, True, True, True]], dtype=bool) a[b] 輸出: array([ 5, 6, 7, 8, 9, 10, 11]) a[b] = 0 print(a) 輸出: array([[0, 1, 2, 3], [4, 0, 0, 0], [0, 0, 0, 0]])
a = np.arange(12).reshape(3,4) b1 = np.array([False,True,True]) b2 = n.array([True,False,True,False]) a[b1,:] 輸出: array([[ 4, 5, 6, 7], [ 8, 9, 10, 11]]) a[b1] 輸出: array([[ 4, 5, 6, 7], [ 8, 9, 10, 11]]) a[:,b2] 輸出: array([[ 0, 2], [ 4, 6], [ 8, 10]]) a[b1,b2] 輸出: array([ 4, 10])
11. ix_() function
a = np.array([2,3,4,5]) b = np.array([8,5,4]) c = np.array([5,4,6,8,3]) ax,bx,cx = np.ix_(a,b,c) print(ax) # (4L, 1L, 1L) 輸出: array([[[2]], [[3]], [[4]], [[5]]]) print(bx) # (1L, 3L, 1L) 輸出: array([[[8], [5], [4]]]) print(cx) # (1L, 1L, 5L) 輸出: array([[[5, 4, 6, 8, 3]]]) result = ax + bx*cx 輸出: array([[[42, 34, 50, 66, 26], [27, 22, 32, 42, 17], [22, 18, 26, 34, 14]], [[43, 35, 51, 67, 27], [28, 23, 33, 43, 18], [23, 19, 27, 35, 15]], [[44, 36, 52, 68, 28], [29, 24, 34, 44, 19], [24, 20, 28, 36, 16]], [[45, 37, 53, 69, 29], [30, 25, 35, 45, 20], [25, 21, 29, 37, 17]]]) result[3,2,4] 輸出:17
12. 線性代數運算
a = np.array([[1.,2.],[3.,4.]]) a.transpose() # 轉置 np.linalg.inv(a) # 求逆 u = np.eye(2) # 產生單位矩陣 np.dot(a,a) # 矩陣乘積 np.trace(a) # 求矩陣的跡 y = np.array([5.],[7.]]) np.linalg.solve(a,y) # 求解線性方程組 np.linalg.eig(a) # 特征分解
“Automatic” Reshaping
a = np.arange(30) a.shape = 2,-1,3 a.shape # (2L, 5L, 3L) print(a) array([[[ 0, 1, 2], [ 3, 4, 5], [ 6, 7, 8], [ 9, 10, 11], [12, 13, 14]], [[15, 16, 17], [18, 19, 20], [21, 22, 23], [24, 25, 26], [27, 28, 29]]])
x = np.arange(0,10,2) y = np.arange(5) m = np.vstack([x,y]) 輸出: array([[0, 2, 4, 6, 8], [0, 1, 2, 3, 4]]) n = np.hstack([x,y]) 輸出: array([0, 2, 4, 6, 8, 0, 1, 2, 3, 4])
13. 矩陣的創建
a = np.array([1,2,3]) a1 = np.mat(a) 輸出: matrix([[1, 2, 3]]) type(a1) 輸出: numpy.matrixlib.defmatrix.matrix a1.shape 輸出: (1L, 3L) a.shape 輸出: (3L,) b=np.matrix([1,2,3]) 輸出: matrix([[1, 2, 3]]) from numpy import * data1 = mat(zeros((3,3))) data2 = mat(ones((2,4))) data3 = mat(random.rand(2,2)) data4 = mat(random.randint(2,8,size=(2,5))) data5 = mat(eye(2,2,dtype=int))
14. 常見的矩陣運算
a1 = mat([1,2]) a2 = mat([[1],[2]]) a3 = a1 * a2 print(a3) 輸出: matrix([[5]]) print(a1*2) 輸出: matrix([[2, 4]]) a1 = mat(eye(2,2)*0.5) print(a1.I) 輸出: matrix([[ 2., 0.], [ 0., 2.]]) a1 = mat([[1,2],[2,3],[4,2]]) a1.sum(axis=0) 輸出: matrix([[7, 7]]) a1.sum(axis=1) 輸出: matrix([[3], [5], [6]]) a1.max() # 求矩陣元素最大值 輸出: 4 a1.min() # 求矩陣元素最小值 輸出: 1 np.max(a1,0) # 求矩陣每列元素最大值 輸出: matrix([[4, 3]]) np.max(a1,1) # 求矩陣每行元素最大值 輸出: matrix([[2], [3], [4]]) a = mat(ones((2,2))) b = mat(eye((2))) c = hstack((a,b)) 輸出: matrix([[ 1., 1., 1., 0.], [ 1., 1., 0., 1.]]) d = vstack((a,b)) 輸出: matrix([[ 1., 1.], [ 1., 1.], [ 1., 0.], [ 0., 1.]])
15. 矩陣、數組、列表之間的互相轉換
aa = [[1,2],[3,4],[5,6]] bb = array(aa) cc = mat(bb) cc.getA() # 矩陣轉換為數組 cc.tolist() # 矩陣轉換為列表 bb.tolist() # 數組轉換為列表 # 當列表為一維時,情況有點特殊 aa = [1,2,3,4] bb = array(aa) 輸出: array([1, 2, 3, 4]) cc = mat(bb) 輸出: matrix([[1, 2, 3, 4]]) cc.tolist() 輸出: [[1, 2, 3, 4]] bb.tolist() 輸出: [1, 2, 3, 4] cc.tolist()[0] 輸出: [1, 2, 3, 4]
內容整理參考鏈接如下:
https://docs.scipy.org/doc/numpy-dev/user/quickstart.html
http://python.usyiyi.cn/translate/NumPy_v111/reference/arrays.scalars.html#arrays-scalars-built-in
以上就是本文的全部內容,希望對大家的學習有所幫助,也希望大家多多支持億速云。
免責聲明:本站發布的內容(圖片、視頻和文字)以原創、轉載和分享為主,文章觀點不代表本網站立場,如果涉及侵權請聯系站長郵箱:is@yisu.com進行舉報,并提供相關證據,一經查實,將立刻刪除涉嫌侵權內容。