Умножение разреженная матрица

__author__ = 'kardashevskiy'
import numpy as np
from scipy import sparse
 
def sg1(B,b,tol=1.0e-8):
    """
    Solve the linear system Ax=b by conjugate gradient method
    """
    n=len(b)
    x=np.zeros((n),'float')
    r=np.copy(b)
    for i in range(n):
        r[i]=np.dot(B[i,0:n],x[0:n])-b[i]
    s=np.copy(r)
    As=np.zeros((n),'float')
    for k in range(n):
        for i in range(n):
            As[i]=np.dot(B[i,0:n],s[0:n])
        alpha=np.dot(r,r)/np.dot(s,As)
        x=x-alpha*s
        for i in range(n):
            r[i]=np.dot(B[i,0:n],x[0:n])-b[i]
        if np.dot(r,r)<tol**2:
            break
        else:
            beta=-np.dot(r,As)/np.dot(s,As)
            s=r+beta*s
    return x,k

__author__ = 'kardashevskiy'
import numpy as np
from scipy import sparse
from scipy.sparse import linalg
from sg1 import sg1
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import time
def f(x, y, r):
    return 2*c*(y-y**2+r*(x**2-x))
def poisson(n, h, r):
    V = []
    I = []
    J = []
    b = []
    for j in range(n):
        for i in range(n):
            k = i+n*j
            b.append(h**2*f(i*h+h, j*h+h, r))
            V.append(2.-2*r)
            I.append(k)
            J.append(k)
            if i>0:
                V.append(-1.)
                I.append(k)
                J.append(k-1)
            if i<n-1:
                V.append(-1.)
                I.append(k)
                J.append(k+1)
            if j>0:
                V.append(r)
                I.append(k)
                J.append(k-n)
            if j<n-1:
                V.append(r)
                I.append(k)
                J.append(k+n)
    return sparse.coo_matrix((V, (I,J))), b
N =100
h = 1./N
c = 4.
r = 0.4
tst = time.clock()
A, b = poisson(N-1, h, r)
B = A.tocsr()
v, info = sg1(B, b,tol=1.e-08)
dt = time.clock() - tst
 
if iter == 0:
    print "N = %i, time solution = %1.3e" % (N, dt)
 
fig = plt.figure()
ax = Axes3D(fig)
x = np.linspace(h/2, 1.-h/2, N-1)
y = np.linspace(h/2, 1.-h/2, N-1)
X,Y = np.meshgrid(x, y)
Z = np.reshape(v, (N-1, N-1))
ax.plot_surface(X, Y, Z, rstride=2, cstride=2, cmap='spectral')
plt.show()

/usr/bin/python /home/kardashevskiy/PycharmProjects/tests/qwe1.py
Traceback (most recent call last):
  File "/home/kardashevskiy/PycharmProjects/tests/qwe1.py", line 47, in <module>
    v, info = sg1(B, b,tol=1.e-08)
  File "/home/kardashevskiy/PycharmProjects/tests/sg1.py", line 13, in sg1
    r[i]=np.dot(B[i,0:n],x[0:n])-b[i]
  File "/usr/lib/python2.7/dist-packages/scipy/sparse/compressed.py", line 193, in __sub__
    raise NotImplementedError('adding a scalar to a sparse '
NotImplementedError: adding a scalar to a sparse matrix is not supported
 
Process finished with exit code 1

__author__ = 'kardashevskiy'
#Hyperbolic equation
import numpy as np
from scipy import sparse
from scipy.sparse import linalg
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
import time
def f(x, y, r):
    return 2*c*(y-y**2+r*(x**2-x))
def poisson(n, h, r):
    V = []
    I = []
    J = []
    b = []
    for j in range(n):
        for i in range(n):
            k = i+n*j
            b.append(h**2*f(i*h+h, j*h+h, r))
            V.append(2.-2*r)
            I.append(k)
            J.append(k)
            if i>0:
                V.append(-1.)
                I.append(k)
                J.append(k-1)
            if i<n-1:
                V.append(-1.)
                I.append(k)
                J.append(k+1)
            if j>0:
                V.append(r)
                I.append(k)
                J.append(k-n)
            if j<n-1:
                V.append(r)
                I.append(k)
                J.append(k+n)
    return sparse.coo_matrix((V, (I,J))), b
 
N =100
h = 1./N
c = 4.
r = 0.4
tst = time.clock()
A, b = poisson(N-1, h, r)
B = A.tocsr()
v, info = linalg.cg(B, b, tol=1.e-08)
dt = time.clock() - tst
 
if iter == 0:
    print "N = %i, time solution = %1.3e" % (N, dt)
 
fig = plt.figure()
ax = Axes3D(fig)
x = np.linspace(h/2, 1.-h/2, N-1)
y = np.linspace(h/2, 1.-h/2, N-1)
X,Y = np.meshgrid(x, y)
Z = np.reshape(v, (N-1, N-1))
ax.plot_surface(X, Y, Z, rstride=2, cstride=2, cmap='spectral')
plt.show()