28/1/2021 matrices parte 2 - Jupyter Notebook

Aplicaciones de las matrices con python

1.- Busque una imagen en google de 200 x 200 pixeles

imagesize:200x200 perro

In [1]:

import PIL
print('Pillow Version:', PIL.__version__)

Pillow Version: 7.1.2

In [3]:

import numpy
print('numpy Version:', numpy.__version__)

numpy Version: 1.18.1

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 1/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [4]:

from PIL import Image

image = Image.open('perro.jpg')

print(image.format)
print(image.size)
print(image.mode)

JPEG
(200, 200)
RGB

In [5]:

image.show()

Convertimos el objeto imagen en un array

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 2/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [6]:

import numpy as np

img_data = np.asarray(image)
print(type(img_data))
print(img_data.shape)

<class 'numpy.ndarray'>
(200, 200, 3)

In [9]:

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 3/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [7]:

img_data[0:10,0:10,0]

Out[7]:

array([[255, 255, 255, 255, 255, 255, 255, 255, 255, 255],
[255, 255, 255, 255, 255, 255, 255, 255, 255, 255],
[255, 255, 255, 255, 255, 255, 255, 255, 255, 255],
[255, 255, 255, 255, 255, 255, 255, 255, 255, 255],
[255, 255, 255, 255, 255, 255, 255, 255, 255, 255],
[254, 254, 254, 254, 254, 254, 254, 254, 254, 254],
[254, 254, 254, 254, 254, 254, 254, 254, 254, 254],
[254, 254, 254, 254, 254, 254, 254, 254, 254, 254],
[255, 255, 255, 255, 255, 255, 255, 255, 254, 254],
[255, 255, 255, 255, 255, 255, 255, 255, 254, 254]], dtype=u
int8)

In [18]:

red_im = Image.fromarray(img_data[:,:,0])

In [19]:

red_im.show()

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 4/20

28/1/2021 matrices parte 2 - Jupyter Notebook

Ejemplo de efectos en la copia de arreglos

matA = np.array([[1,2],[3,4]]); matB=matA; matA[0,0]=10

In [20]:

matA = np.array([[1,2],[3,4]]); matB=matA; matA[0,0]=10

In [21]:

matB

Out[21]:

array([[10, 2],
[ 3, 4]])

In [22]:

matA = np.array([[1,2],[3,4]]); matB=np.array(matA); matA[0,0]=10; matB

Out[22]:

array([[1, 2],
[3, 4]])

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 5/20

28/1/2021 matrices parte 2 - Jupyter Notebook

Mostramos la imagen en tonos de rojo

In [29]:

red_data = np.array(img_data)
red_data[:,:,0]=img_data[:,:,0]
red_data[:,:,1:3]=0
Image.fromarray(red_data).show()

In [ ]:

In [30]:

gr_im = Image.fromarray(img_data)
gr_im.show()

In [12]:

In [ ]:

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 6/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [18]:

gr_im = Image.fromarray(img_data).save('otro_perro.png')

In [ ]:

Ortogonalizacion de matrices

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 7/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [70]:

def gs(A):
(m,n) = A.shape
Q = np.zeros_like(A)
R = np.zeros((n,n))
for j in range(n):
print(f"Columna {j+1}")
v = A[:,j]
for i in range(j):
R[i,j] = np.dot(Q[:,i],A[:,j])
v = v - R[i,j]*Q[:,i]
R[j,j] = np.linalg.norm(v)
Q[:,j] = v/R[j,j]
#print(v/R[j,j],Q[:,j] )
return Q,R

In [71]:

A= np.array([[1,2,3],[3,4,5],[5,6,7]],dtype='f4')

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 8/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [72]:

Q,R = gs(A); print(Q.dtype)

Columna 1
Columna 2
Columna 3
float32

In [73]:

Q@R

Out[73]:

array([[1. , 1.99999996, 2.99999995],

[3.00000001, 4.00000009, 4.99999992],
[5.00000001, 5.99999998, 6.99999987]])

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 9/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [75]:

Q.T@Q

Out[75]:

array([[ 9.9999994e-01, 8.9406967e-07, -7.7666223e-02],

[ 8.9406967e-07, 9.9999994e-01, -9.9671614e-01],
[-7.7666223e-02, -9.9671614e-01, 9.9999994e-01]], dtype=flo
at32)

In [76]:

Out[76]:

array([[5.91608000e+00, 7.43735695e+00, 8.95863438e+00],

[0.00000000e+00, 8.28078687e-01, 1.65616584e+00],
[0.00000000e+00, 0.00000000e+00, 8.49679873e-06]])

In [ ]:

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 10/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [34]:

A=np.array( [ [1,2,1],[2,2,-1],[1,1,2],[1,-2,1] ])
gs(A)

Columna 1
Columna 2
Columna 3

Out[34]:

array([[0, 0, 0],
[0, 0, 0],
[0, 0, 0],
[0, 0, 0]])

In [ ]:

In [ ]:

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 11/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [65]:

def qr(A):
m, n = A.shape
Q = np.zeros((m, n))
R = np.zeros((n, n))

for j in range(n):
v = A[:, j]

for i in range(j - 1):

q = Q[:, i]
R[i, j] = q.dot(v)
v = v - R[i, j] * q

norm = np.linalg.norm(v)
Q[:, j] = v / norm
R[j, j] = norm
return Q, R
#A = np.random.rand(4,4) * 1000
A= np.array([[1,2,3],[3,4,5],[5,6,7]])
Q, R = qr(A)

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 12/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [66]:

np.linalg.norm(A-Q@R)

Out[66]:

0.0

Solucion de minimos cuadrados

Ejemplo
127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 13/20
28/1/2021 matrices parte 2 - Jupyter Notebook

Resolver

𝑥 + 2𝑦 + 𝑧 = 1
2𝑥 + 𝑦 − 𝑧 = 4
𝑥 + 𝑦 + 2𝑧 = 2
𝑥 − 2𝑦 + 𝑧 = 0
In [67]:

A=np.array( [ [1,2,1],[2,2,-1],[1,1,2],[1,-2,1] ])

In [68]:

b = np.array( [1,4,2,0])

In [69]:

Q, R = qr(A)

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 14/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [70]:

Out[70]:

array([[ 0.37796447, 0.5547002 , 0.28171808],

[ 0.75592895, 0.5547002 , -0.61977979],
[ 0.37796447, 0.2773501 , 0.6761234 ],
[ 0.37796447, -0.5547002 , 0.28171808]])

In [71]:

Out[71]:

array([[2.64575131, 0. , 0.75592895],
[0. , 3.60555128, 0. ],
[0. , 0. , 2.53546276]])

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 15/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [73]:

Q.T@Q

Out[73]:

array([[ 1.00000000e+00, 5.24142418e-01, 1.66533454e-16],

[ 5.24142418e-01, 1.00000000e+00, -1.56269077e-01],
[ 1.66533454e-16, -1.56269077e-01, 1.00000000e+00]])

In [76]:

sol=np.linalg.solve(R,Q.T@b)

In [77]:

print(sol)

[ 1.66666667 0.92307692 -0.33333333]

Factorizacion Cholesky

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 16/20

28/1/2021 matrices parte 2 - Jupyter Notebook

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 17/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [88]:

def chol(A):
m,n = A.shape
L = np.zeros_like(A)
L[0,0]=np.sqrt(A[0,0])
for k in range(n-1):
L[k+1,0] = A[k+1,0]/L[0,0]

for i in range(1,k-1):
L[k+1,i] = ( A[k+1,i] - np.sum( L[k+1,1:i-1]*L[i,1:i-1] ) )/L[i,i]

L[k+1,1:k] = np.sqrt(A[k+1,k+1]-np.sum( L[k+1,1:k]**2))

return L

In [92]:

A = 2*np.eye(4)+np.diag([-1,-1,-1],-1)+np.diag([-1,-1,-1],1)

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 18/20

28/1/2021 matrices parte 2 - Jupyter Notebook

In [90]:

Out[90]:

array([[ 2., -1., 0., 0.],

[-1., 2., -1., 0.],
[ 0., -1., 2., -1.],
[ 0., 0., -1., 2.]])

In [91]:

chol(A)

Out[91]:

array([[ 1.41421356, 0. , 0. , 0. ],
[-0.70710678, 0. , 0. , 0. ],
[ 0. , 0. , 0. , 0. ],
[ 0. , 1.41421356, 0. , 0. ]])

In [ ]:

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 19/20

28/1/2021 matrices parte 2 - Jupyter Notebook

127.0.0.1:8888/notebooks/Documentos/2020-1/matrices parte 2.ipynb 20/20