Errata for GVL4

Last Updated: 10/19/2014

 

Page

 

The Mistake

xii

  In the Acknowledgements, Terkelsen not Terkelson, Balzano not Bolzano  and there are 2 "through" 's
xix   http://www.cs.cornell.edu/cv/GVL4/golubandvanloan.htm   is better than http://www.cs.cornell.edu/cv/GVL4/
11   At the beginning of section 1.1.14, should be Consider the kji variant...
20   4th line from bottom, v = [n 1:n-1].
28   In the middle, X is n2-by-n1 , not n1-by-m2
29   In line 3, should be   ... can be computed ...    instead of   ... can computed ...                               At the very end of section 1.3.8, should be "section 12.3" not "section 13.1"
41   In P1.4.2, the subscript ranges should go up to n, not n-1

41

  Algorithm 1.4.4, should have y(1:2:n) and y(2:2:n)
97    It should be -q in the numerator in the Method 2 recipe for rmin

120

  Algorithm 3.2.4, it should be stated that   n= Nr  is assumed.
161   3rd line in Cor 4.2.4 proof: An L^{-1} should be an L^{-T}

181

  Algorithm 4.3.6, 2nd loop, should be  b(k) = b(k) - beta(k-1) * b(k-1)
236   In Alg 5.1.1, beta should be set to +2 (not -2) in the elseif part.
237   Missing right paren in th erecipe for \beta in (5.14)
238   Should be 2pn(2m-n) in first line at the top of the page.
261   In equation (5.3.3), upper limit in the summation is "m", not  "2".
266   In (5.3.11) at the top, should be "x_{LS}" in the lhs denominator, not just "x"
399   In Algorithm 7.6.3, replace the k=j+1:q loop with                                                                       for k=j:q, T{i,k} = T{i,k} - Z*T{j,k}; end                                                                               for k=i:q, T{k,j} = T{k,j} + T{k,i}*Z; end                                                          
402   Trailing 3x3 in the displayed matrix at the top should be [lambda 0 a; 0 lambda b; 0 0 lambda].
455   Missing sqrt in last equation in proof. Should have \sqrt{1 - d_{0}^2}
458  
So that A is overwritten by Q'AQ = T, must set A(k+2:n,k) and A(k,k+2:n) to zero each step
493   Should be (4m^2)n + 8m(n^2) in line 3 of the table.
517   (9.1.8) should have    X^{-1}f(A) X     not    X f(A) X^{-1}
534   Top line: (7.9.8) not (7.8.8). And in  P9.3.4, solution should be F_{22}^TF_{12}.
537   Near the bottom: S^2 = I  not S^2 = S

538

  Missing right paren in (9.4.5)
542   In P9.4.1, should be X^{-1} = A  not  X^{-1} - A
558   In the recipe for R_{G} at the bottom, should be "2k" not "2n"
560    The recipe for sigma in step 3 should be multiplied by || u ||^2 . (because \int p_{i}(lambda)p_{i}(lambda) = || u ||^2 )
561   Last line of \S10.2.5: Should be \beta_{k}^{2} not \beta{k+1}^2.
575   In equation (10.4.18), should be beta_{n} and \beta_{n+1} in the last two rows.
583   In the displayed equation AQ_{+} = ... , add the term H_{+}(j+1,j)*Q_{+}(:,j+1) to the rhs. Modify "Replace r_{c}" line in the loop below accordingly.
685   In Alg 12.1.1, should be \alpha = a_{1} not \alpha = a_{11}
728   Table at the bottom. Line 3: I_{m1} for  I_n1. Line 4. I_m2 and I_m1 for I_n2 and I_n1. Line 5: I_m3, I_m2 and I_m1 for I_n3, I_n2, and I_n1.
735   in section 12.5.4, the "\dotcircle" operator is the Khatri-Rao product operator based on column-partitionings, e.g. [a1 a2 a3] \dotcircle [b1 b2 b3] = [kron(a1,b1)  kron(a2,b2)  kron(a3,b3)].
738   At the top, the multiple Hadamard product inside the norm should be transposed.
743   In (12.5.29), G_1 = U_1 just before the loop.