Differences between revisions 1 and 21 (spanning 20 versions)

Sage Interactions - Linear Algebra

Contents

Sage Interactions - Linear Algebra

Numerical instability of the classical Gram-Schmidt algorithm FIXME

by Marshall Hampton (tested by William Stein, who thinks this is really nice!)

xxxxxxxxxx
 
def GS_classic(a_list):
    '''
    Given a list of vectors or a matrix, returns the QR factorization using the classical (and numerically unstable) Gram-Schmidt algorithm.    
    '''
    if type(a_list) != list:
        cols = a_list.cols()
        a_list = [x for x in cols]
    indices = range(len(a_list))
    q = []
    r = [[0 for i in indices] for j in indices]
    v = [a_list[i].copy() for i in indices]
    for i in indices:
        for j in range(0,i):
            r[j][i] = q[j].inner_product(a_list[i])
            v[i] = v[i] - r[j][i]*q[j]
        r[i][i] = (v[i]*v[i])^(1/2)
        q.append(v[i]/r[i][i])
    q = matrix([q[i] for i in indices]).transpose()
    return q, matrix(r)
def GS_modern(a_list):
    '''
    Given a list of vectors or a matrix, returns the QR factorization using the 'modern' Gram-Schmidt algorithm.    
    '''
    if type(a_list) != list:
        cols = a_list.cols()
        a_list = [x for x in cols]
    indices = range(len(a_list))
    q = []
    r = [[0 for i in indices] for j in indices]
    v = [a_list[i].copy() for i in indices]
    for i in indices:
        r[i][i] = v[i].norm(2)
        q.append(v[i]/r[i][i])
        for j in range(i+1, len(indices)):
            r[i][j] = q[i].inner_product(v[j])
            v[j] = v[j] - r[i][j]*q[i]
    q = matrix([q[i] for i in indices]).transpose()
    return q, matrix(r)
html('<h2>Numerical instability of the classical Gram-Schmidt algorithm</h2>')
@interact
def gstest(precision = slider(range(3,53), default = 10), a1 = input_box([1,1/1000,1/1000]), a2 = input_box([1,1/1000,0]), a3 = input_box([1,0,1/1000])):
    myR = RealField(precision)
    displayR = RealField(5)
    html('precision in bits: ' + str(precision) + '<br>')
    A = matrix([a1,a2,a3])
    A = [vector(myR,x) for x in A]
    qn, rn = GS_classic(A)
    qb, rb = GS_modern(A)
    html('Classical Gram-Schmidt:')
    show(matrix(displayR,qn))
    html('Stable Gram-Schmidt:')
    show(matrix(displayR,qb))

Equality of det(A) and det(A.tranpose())

srg = srange(-4,4,1/10,include_endpoint=True) @interact def dualv(a1=slider(srg,default=1),a2=slider(srg,default=2), a3=slider(srg,default=-1),a4=slider(srg,default=3)):

A1 = arrow2d([0,0],[a1,a2],rgbcolor='black') A2 = arrow2d([0,0],[a3,a4],rgbcolor='black') A3 = arrow2d([0,0],[a1,a3],rgbcolor='black') A4 = arrow2d([0,0],[a2,a4],rgbcolor='black')
p1 = polygon(0,0],[a1,a2],[a1+a3,a2+a4],[a3,a4],[0,0, alpha=.5) p2 = polygon(0,0],[a1,a3],[a1+a2,a3+a4],[a2,a4],[0,0,rgbcolor='red', alpha=.5) A = matrix(a1,a2],[a3,a4) html('<h3>The determinant of a matrix is equal to the determinant of the transpose</h3>') html("det("%(latex(A),latex(A.transpose()))) show(A1+A2+A3+A4+p1+p2)

Linear transformations

by Jason Grout

A square matrix defines a linear transformation which rotates and/or scales vectors. In the interact command below, the red vector represents the original vector (v) and the blue vector represents the image w under the linear transformation. You can change the angle and length of v by changing theta and r.

xxxxxxxxxx
 
@interact
def linear_transformation(A=matrix([[1,-1],[-1,1/2]]),theta=slider(0, 2*pi, .1), r=slider(0.1, 2, .1, default=1)):
    v=vector([r*cos(theta), r*sin(theta)])
    w = A*v
    circles = sum([circle((0,0), radius=i, color='black') for i in [1..2]])
    html("$%s %s=%s$"%tuple(map(latex, [A, v.column().n(4), w.column().n(4)])))
    show(v.plot(color='red')+w.plot(color='blue')+circles,aspect_ratio=1)

Gerschgorin Circle Theorem

by Marshall Hampton. This animated version requires convert (imagemagick) to be installed, but it can easily be modified to a static version. The animation illustrates the idea behind the stronger version of Gerschgorin's theorem, which says that if the disks around the eigenvalues are disjoint then there is one eigenvalue per disk. The proof is by continuity of the eigenvalues under a homotopy to a diagonal matrix.

xxxxxxxxxx
 
from scipy import linalg
html('<h2>The Gerschgorin circle theorem</h2>')
@interact
def Gerschgorin(Ain = input_box(default='[[10,1,1/10,0],[-1,9,0,1],[1,0,2,3/10],[-.5,0,-.3,1]]', type = str, label = 'A = '), an_size = slider(1,100,1,1.0)):
    A = sage_eval(Ain)
    size = len(A)
    html('$A = ' + latex(matrix(RealField(10),A))+'$')
    A = matrix(RealField(10),A)
    B = [[0 for i in range(size)] for j in range(size)]
    for i in range(size):
        B[i][i] = A[i][i]
    B = matrix(B)
    frames = []
​
    centers = [(real(q),imag(q)) for q in [A[i][i] for i in range(size)]]
    radii_row = [sum([abs(A[i][j]) for j in range(i)+range(i+1,size)]) for i in range(size)]
    radii_col = [sum([abs(A[j][i]) for j in range(i)+range(i+1,size)]) for i in range(size)]
    x_min = min([centers[i][0]-radii_row[i] for i in range(size)]+[centers[i][0]-radii_col[i] for i in range(size)])
    x_max = max([centers[i][0]+radii_row[i] for i in range(size)]+[centers[i][0]+radii_col[i] for i in range(size)])
    y_min = min([centers[i][1]-radii_row[i] for i in range(size)]+[centers[i][1]-radii_col[i] for i in range(size)])
    y_max = max([centers[i][1]+radii_row[i] for i in range(size)]+[centers[i][1]+radii_col[i] for i in range(size)])
​
    if an_size > 1: 
        t_range= srange(0,1+1/an_size,1/an_size)
    else:
        t_range = [1]
    for t in t_range:
        C = t*A + (1-t)*B
        eigs = [CDF(x) for x in linalg.eigvals(C.numpy())]
        eigpoints = points([(real(q),imag(q)) for q in eigs],pointsize = 10, rgbcolor = (0,0,0))
        centers = [(real(q),imag(q)) for q in [A[i][i] for i in range(size)]]
        radii_row = [sum([abs(C[i][j]) for j in range(i)+range(i+1,size)]) for i in range(size)]
        radii_col = [sum([abs(C[j][i]) for j in range(i)+range(i+1,size)]) for i in range(size)]
        scale = max([(x_max-x_min),(y_max-y_min)])
        scale = 7/scale
        row_circles = sum([circle(centers[i],radii_row[i],fill=True, alpha = .3) for i in range(size)])
        col_circles = sum([circle(centers[i],radii_col[i],fill=True, rgbcolor = (1,0,0), alpha = .3) for i in range(size)])
        ft = eigpoints+row_circles+col_circles
        frames.append(ft)
    show(animate(frames,figsize = [(x_max-x_min)*scale,(y_max-y_min)*scale], xmin = x_min, xmax=x_max, ymin = y_min, ymax = y_max))

Singular value decomposition

by Marshall Hampton

xxxxxxxxxx
 
import scipy.linalg as lin
var('t')
def rotell(sig,umat,t,offset=0):
    temp = matrix(umat)*matrix(2,1,[sig[0]*cos(t),sig[1]*sin(t)])
    return [offset+temp[0][0],temp[1][0]]
@interact
def svd_vis(a11=slider(-1,1,.05,1),a12=slider(-1,1,.05,1),a21=slider(-1,1,.05,0),a22=slider(-1,1,.05,1),ofs= ('offset image from domain',False)):
    rf_low = RealField(12)
    my_mat = matrix(rf_low,2,2,[a11,a12,a21,a22])
    u,s,vh = lin.svd(my_mat.numpy())
    if ofs: 
        offset = 3
        fsize = 6
        colors = [(1,0,0),(0,0,1),(1,0,0),(0,0,1)]
    else: 
        offset = 0
        fsize = 5
        colors = [(1,0,0),(0,0,1),(.7,.2,0),(0,.3,.7)]
    vvects = sum([arrow([0,0],matrix(vh).row(i),rgbcolor = colors[i]) for i in (0,1)])    
    uvects = Graphics()
    for i in (0,1):
        if s[i] != 0: uvects += arrow([offset,0],vector([offset,0])+matrix(s*u).column(i),rgbcolor = colors[i+2])
    html('<h3>Singular value decomposition: image of the unit circle and the singular vectors</h3>')
    html("$A = %s  = %s %s %s$"%(latex(my_mat), latex(matrix(rf_low,u.tolist())), latex(matrix(rf_low,2,2,[s[0],0,0,s[1]])), latex(matrix(rf_low,vh.tolist())))) 
    image_ell = parametric_plot(rotell(s,u,t, offset),(0,2*pi))
    graph_stuff=circle((0,0),1)+image_ell+vvects+uvects
    graph_stuff.set_aspect_ratio(1)
    show(graph_stuff,frame = False,axes=False,figsize=[fsize,fsize])

Discrete Fourier Transform

by Marshall Hampton

xxxxxxxxxx
 
import scipy.fftpack as Fourier
@interact
def discrete_fourier(f = input_box(default=sum([sin(k*x) for k in range(1,5,2)])), scale = slider(.1,20,.1,5)):
    pbegin = -float(pi)*scale
    pend = float(pi)*scale
    html("<h3>Function plot and its discrete Fourier transform</h3>")
    show(plot(f, (x,pbegin, pend), plot_points = 512), figsize = [4,3])
    f_vals = [f(x=ind) for ind in srange(pbegin, pend,(pend-pbegin)/512.0)]
    my_fft = Fourier.fft(f_vals)
    show(list_plot([abs(i) for i in my_fft], plotjoined=True), figsize = [4,3])

The Gauss-Jordan method for inverting a matrix

by Hristo Inouzhe

xxxxxxxxxx
 
#Choose the size D of the square matrix:
D = 3
​
example = [[1 if k==j else 0 for k in range(D)] for j in range(D)]
example[0][-1] = 2
example[-1][0] = 3
​
@interact
def _(M=input_grid(D,D, default = example,
                   label='Matrix to invert', to_value=matrix),
      tt = text_control('Enter the bits of precision used'
                        ' (only if you entered floating point numbers)'),  
      precision = slider(5,100,5,20),
      auto_update=False):
    if det(M)==0:
        print 'Failure: Matrix is not invertible'
        return
    if M.base_ring() == RR:
        M = M.apply_map(RealField(precision))
    N=M
    M=M.augment(identity_matrix(D))
    print 'We construct the augmented matrix'
    show(M)
    for m in range(0,D-1):
        if M[m,m] == 0:
            lista = [(M[j,m],j) for j in range(m,D)]
            maxi, c = max(lista)
            M[c,:],M[m,:]=M[m,:],M[c,:]
            print 'We permute rows %d and %d'%(m+1,c+1)
            show(M)
        for n in range(m+1,D):
            a=M[m,m]
            if M[n,m]!=0:
                print "We add %s times row %d to row %d"%(-M[n,m]/a, m+1, n+1)
                M=M.with_added_multiple_of_row(n,m,-M[n,m]/a)
                show(M)
    for m in range(D-1,-1,-1):
        for n in range(m-1,-1,-1):
            a=M[m,m]
            if M[n,m]!=0:
                print "We add %s times row %d to the row %d"%(-M[n,m]/a, m+1, n+1)
                M=M.with_added_multiple_of_row(n,m,-M[n,m]/a)
                show(M)       
    for m in range(0,D):
        if M[m,m]!=1:
            print 'We divide row %d by %s'%(m+1,M[m,m])
            M = M.with_row_set_to_multiple_of_row(m,m,1/M[m,m])
            show(M)   
    M=M.submatrix(0,D,D)
    print 'We keep the right submatrix, which contains the inverse'        
    html('$$M^{-1}=%s$$'%latex(M))
    print 'We check it actually is the inverse'
    html('$$M^{-1}*M=%s*%s=%s$$'%(latex(M),latex(N),latex(M*N)))

...(goes all the way to invert the matrix)

-  ⇤ ← Revision 1 as of 2008-05-07 13:28:12 → 
  Size: 5314
  Editor: schilly
  Comment:
+   ← Revision 21 as of 2012-11-15 17:49:32 → ⇥
  Size: 11255
  Editor: mhampton
  Comment:
-Deletions are marked like this.
+Additions are marked like this.
 Line 2:
-goto [:interact:interact main page]

== Numerical instability of the classical Gram-Schmidt algorithm ==
+goto [[interact|interact main page]]

<<TableOfContents>>

== Numerical instability of the classical Gram-Schmidt algorithm FIXME ==
-Line 6:
+Line 8:
-{{{
+{{{#!sagecell
-Line 60:
+Line 62:
-attachment:GramSchmidt.png
+{{attachment:GramSchmidt.png}}

== Equality of det(A) and det(A.tranpose()) ==
srg = srange(-4,4,1/10,include_endpoint=True)
@interact
def dualv(a1=slider(srg,default=1),a2=slider(srg,default=2), a3=slider(srg,default=-1),a4=slider(srg,default=3)):
    A1 = arrow2d([0,0],[a1,a2],rgbcolor='black')
    A2 = arrow2d([0,0],[a3,a4],rgbcolor='black')
    A3 = arrow2d([0,0],[a1,a3],rgbcolor='black')
    A4 = arrow2d([0,0],[a2,a4],rgbcolor='black')
    p1 = polygon([[0,0],[a1,a2],[a1+a3,a2+a4],[a3,a4],[0,0]], alpha=.5)
    p2 = polygon([[0,0],[a1,a3],[a1+a2,a3+a4],[a2,a4],[0,0]],rgbcolor='red', alpha=.5)
    A = matrix([[a1,a2],[a3,a4]])
    html('<h3>The determinant of a matrix is equal to the determinant of the transpose</h3>')
    html(" $d e t ($ "%(latex(A),latex(A.transpose())))
    show(A1+A2+A3+A4+p1+p2)
-Line 68:
+Line 85:
-{{{
@interact
def linear_transformation(theta=slider(0, 2*pi, .1), r=slider(0.1, 2, .1, default=1)):
    A=matrix([[1,-1],[-1,1/2]])
+{{{#!sagecell
@interact
def linear_transformation(A=matrix([[1,-1],[-1,1/2]]),theta=slider(0, 2*pi, .1), r=slider(0.1, 2, .1, default=1)):
-Line 74:
+Line 90:
-    circles = sum([circle((0,0), radius=i, rgbcolor=(0,0,0)) for i in [1..2]])
    print jsmath("v = %s,\; %s v=%s"%(v.n(4),latex(A),w.n(4)))
    show(v.plot(rgbcolor=(1,0,0))+w.plot(rgbcolor=(0,0,1))+circles,aspect_ratio=1)
}}}
attachment:Linear-Transformations.png
+    circles = sum([circle((0,0), radius=i, color='black') for i in [1..2]])
    html(""%tuple(map(latex, [A, v.column().n(4), w.column().n(4)])))
    show(v.plot(color='red')+w.plot(color='blue')+circles,aspect_ratio=1)
}}}
{{attachment:Linear-Transformations.png}}

== Gerschgorin Circle Theorem ==
by Marshall Hampton.  This animated version requires convert (imagemagick) to be installed, but it can easily be modified to a static version.
The animation illustrates the idea behind the stronger version of Gerschgorin's theorem, which says that if the disks around the eigenvalues are disjoint then there is one eigenvalue per disk.  The proof is by continuity of the eigenvalues under a homotopy to a diagonal matrix. 
{{{#!sagecell
from scipy import linalg
html('<h2>The Gerschgorin circle theorem</h2>')
@interact
def Gerschgorin(Ain = input_box(default='[[10,1,1/10,0],[-1,9,0,1],[1,0,2,3/10],[-.5,0,-.3,1]]', type = str, label = 'A = '), an_size = slider(1,100,1,1.0)):
    A = sage_eval(Ain)
    size = len(A)
    html(' $A =^{'} + l a t e x (m a t r i x (R e a l F i e l d (10), A)) +^{'}$ ')
    A = matrix(RealField(10),A)
    B = [[0 for i in range(size)] for j in range(size)]
    for i in range(size):
        B[i][i] = A[i][i]
    B = matrix(B)
    frames = []

    centers = [(real(q),imag(q)) for q in [A[i][i] for i in range(size)]]
    radii_row = [sum([abs(A[i][j]) for j in range(i)+range(i+1,size)]) for i in range(size)]
    radii_col = [sum([abs(A[j][i]) for j in range(i)+range(i+1,size)]) for i in range(size)]
    x_min = min([centers[i][0]-radii_row[i] for i in range(size)]+[centers[i][0]-radii_col[i] for i in range(size)])
    x_max = max([centers[i][0]+radii_row[i] for i in range(size)]+[centers[i][0]+radii_col[i] for i in range(size)])
    y_min = min([centers[i][1]-radii_row[i] for i in range(size)]+[centers[i][1]-radii_col[i] for i in range(size)])
    y_max = max([centers[i][1]+radii_row[i] for i in range(size)]+[centers[i][1]+radii_col[i] for i in range(size)])

    if an_size > 1: 
        t_range= srange(0,1+1/an_size,1/an_size)
    else:
        t_range = [1]
    for t in t_range:
        C = t*A + (1-t)*B
        eigs = [CDF(x) for x in linalg.eigvals(C.numpy())]
        eigpoints = points([(real(q),imag(q)) for q in eigs],pointsize = 10, rgbcolor = (0,0,0))
        centers = [(real(q),imag(q)) for q in [A[i][i] for i in range(size)]]
        radii_row = [sum([abs(C[i][j]) for j in range(i)+range(i+1,size)]) for i in range(size)]
        radii_col = [sum([abs(C[j][i]) for j in range(i)+range(i+1,size)]) for i in range(size)]
        scale = max([(x_max-x_min),(y_max-y_min)])
        scale = 7/scale
        row_circles = sum([circle(centers[i],radii_row[i],fill=True, alpha = .3) for i in range(size)])
        col_circles = sum([circle(centers[i],radii_col[i],fill=True, rgbcolor = (1,0,0), alpha = .3) for i in range(size)])
        ft = eigpoints+row_circles+col_circles
        frames.append(ft)
    show(animate(frames,figsize = [(x_max-x_min)*scale,(y_max-y_min)*scale], xmin = x_min, xmax=x_max, ymin = y_min, ymax = y_max))
}}}
{{attachment:Gerschanimate.png}}

{{attachment:Gersch.gif}}
-Line 82:
+Line 147:
-{{{
+{{{#!sagecell
-Line 89:
+Line 154:
-def svd_vis(a11=slider(-1,1,.05,1),a12=slider(-1,1,.05,1),a21=slider(-1,1,.05,0),a22=slider(-1,1,.05,1),ofs= selector(['Off','On'],label='offset image from domain')):
+def svd_vis(a11=slider(-1,1,.05,1),a12=slider(-1,1,.05,1),a21=slider(-1,1,.05,0),a22=slider(-1,1,.05,1),ofs= ('offset image from domain',False)):
-Line 93:
+Line 158:
-    if ofs == 'On':
+    if ofs:
-Line 106:
+Line 171:
-    print jsmath("A = %s  = %s %s %s"%(latex(my_mat), latex(matrix(rf_low,u.tolist())), latex(matrix(rf_low,2,2,[s[0],0,0,s[1]])), latex(matrix(rf_low,vh.tolist())))) 
    image_ell = parametric_plot(rotell(s,u,t, offset),0,2*pi)
+    html("$A = %s  = %s %s %s$"%(latex(my_mat), latex(matrix(rf_low,u.tolist())), latex(matrix(rf_low,2,2,[s[0],0,0,s[1]])), latex(matrix(rf_low,vh.tolist())))) 
    image_ell = parametric_plot(rotell(s,u,t, offset),(0,2*pi))
-Line 110:
+Line 175:
-    show(graph_stuff,frame = False,axes=False,figsize=[fsize,fsize])
}}}
attachment:svd1.png
+    show(graph_stuff,frame = False,axes=False,figsize=[fsize,fsize])}}}
{{attachment:svd1.png}}
-Line 116:
+Line 180:
-{{{
+{{{#!sagecell
-Line 120:
+Line 184:
-    var('x')
-Line 124:
+Line 187:
-    show(plot(f, pbegin, pend, plot_points = 512), figsize = [4,3])
    f_vals = [f(ind) for ind in srange(pbegin, pend,(pend-pbegin)/512.0)]
+    show(plot(f, (x,pbegin, pend), plot_points = 512), figsize = [4,3])
    f_vals = [f(x=ind) for ind in srange(pbegin, pend,(pend-pbegin)/512.0)]
-Line 127:
+Line 190:
-    show(list_plot([abs(x) for x in my_fft], plotjoined=True), figsize = [4,3])
}}}
attachment:dfft1.png
+    show(list_plot([abs(i) for i in my_fft], plotjoined=True), figsize = [4,3])
}}}
{{attachment:dfft1.png}}

== The Gauss-Jordan method for inverting a matrix ==
by Hristo Inouzhe
{{{#!sagecell
#Choose the size D of the square matrix:
D = 3

example = [[1 if k==j else 0 for k in range(D)] for j in range(D)]
example[0][-1] = 2
example[-1][0] = 3

@interact
def _(M=input_grid(D,D, default = example,
                   label='Matrix to invert', to_value=matrix),
      tt = text_control('Enter the bits of precision used'
                        ' (only if you entered floating point numbers)'),  
      precision = slider(5,100,5,20),
      auto_update=False):
    if det(M)==0:
        print 'Failure: Matrix is not invertible'
        return
    if M.base_ring() == RR:
        M = M.apply_map(RealField(precision))
    N=M
    M=M.augment(identity_matrix(D))
    print 'We construct the augmented matrix'
    show(M)
    for m in range(0,D-1):
        if M[m,m] == 0:
            lista = [(M[j,m],j) for j in range(m,D)]
            maxi, c = max(lista)
            M[c,:],M[m,:]=M[m,:],M[c,:]
            print 'We permute rows %d and %d'%(m+1,c+1)
            show(M)
        for n in range(m+1,D):
            a=M[m,m]
            if M[n,m]!=0:
                print "We add %s times row %d to row %d"%(-M[n,m]/a, m+1, n+1)
                M=M.with_added_multiple_of_row(n,m,-M[n,m]/a)
                show(M)
    for m in range(D-1,-1,-1):
        for n in range(m-1,-1,-1):
            a=M[m,m]
            if M[n,m]!=0:
                print "We add %s times row %d to the row %d"%(-M[n,m]/a, m+1, n+1)
                M=M.with_added_multiple_of_row(n,m,-M[n,m]/a)
                show(M)       
    for m in range(0,D):
        if M[m,m]!=1:
            print 'We divide row %d by %s'%(m+1,M[m,m])
            M = M.with_row_set_to_multiple_of_row(m,m,1/M[m,m])
            show(M)   
    M=M.submatrix(0,D,D)
    print 'We keep the right submatrix, which contains the inverse'        
    html(' $M^{- 1} =$ '%latex(M))
    print 'We check it actually is the inverse'
    html(' $M^{- 1} * M =$ '%(latex(M),latex(N),latex(M*N)))
}}}
{{attachment:gauss-jordan.png}}

...(goes all the way to invert the matrix)

Diff for "interact/linear_algebra"