Errata -- July 2004 -- Your contributions are welcome!

Chapter 2

p 20 Table 2.1 
  Z Format for 3.4e+38 should be '7f7fffff' rather than '7fefffff'

p 20, line 1 should read
  V = Z'BF800000'   not Z'C1100000'

p 20, line -14 parenthetical remark should begin with
  '3 more than the ...'   3, not 4 as written

p 23, line 5, equation (2.5.1)
  sign should be minus, not plus:
  SUM (xi-xbar)**2 = SUM(xi-x1)**2 - n(x1-xbar)**2

p 31, line -7, equation (2.7.7)
  sign of b should be negative, to read
   2c/[ -b + sqrt(b^2-4ac) ]

p 34, Exercise 2.12, fourth digit on Euler's gamma is wrong:
    .577215664
 
p 34, Exercise 2.16 (c), (d) variable should be x
  (c) f(x) = log(1+x)
  (d) f(x) = PHI(x)


Chapter 3
  
p 41, third display of code, second line should read
  i=n+1-ii  ! count down from n to 1
        ^^ two i's

p 42, line -5, entries in M(2) which are neither 0 nor 1 are
   -A(1)sub32/A(1)sub22
   -A(1)sub42/A(1)sub22
(numerator subscripts should be 22 and 32 instead of 31 and 41)

p 47, line -7, superscript of last matrix (Lbold) is [k]T (not k-1)

p 48, line -11, equation (3.5.4), superscript for Dbold is [k-1], not [k]

p 49, Example 3.3, line for k=2 should read 'for the unknown L21'
      (not A21)

Chapter 3

p 59, Exercise 3.23, Sherman-Morrison-Woodbury formula is incorrect,
      although correctly given in (5.9.4) on p 99
        inv(A+uv') = inv(A) - 1/(1+v'inv(A)u)*inv(A)*u*v'*inv(A)

Chapter 4

p 67, line -9, denominator should be quadratic form in a column of ones
       (not a bilinear form in 1 and z)(replace z by 1)

p 79, line -7, denominator in general case should be (j-d) to read:
       r(j) = (j-1+d)*r(j-1)/(j-d)

Chapter 5

p 97, lots of wrong signs (7 of them)
  U12 X = ... = matrix  <-- sign of 2,1 element should be +
  U13 U12 X = matrix1 * matrix <- sign of 2,1 element should be +
            = matrix    <-- sign of 2,1 element should be +

  'use [vector]' <-- sign of first element should be +
   U23 U13 U12 X = matrix1 * matrix2 
      matrix1 <-- signs of 2,2 and 3,3 elements should be +
      matrix2 <-- sign of 2,1 element should be +

p 108, line -3, file name is 'chex57', not 'chkx57'

p 110, Exercise 5.11 sign is wrong in expression for Si:
     Si = Si-1 + (i yi - Ti )**2/[i(i-1)] 


Chapter 7

p 144, equation (7.3.9), expression for d1 should be
     d1 = 6(s2 - z1')/h2     (subscript for s is 2, not 1)

p 145, line 8, below equation (7.3.10), condition is
     | z^(4) | <= M          (replace y with z)

p 149, line 12, the sentence beginning with 'Completeness'
     should read
      Completeness of a set of orthonormal functions means
      that a function orthogonal to every member is a function
      with zero norm.

p 166, line 3, Exercise 7.9, display should read
     Fhat( y + log m ) = exp(-exp(-y))
                   ^ should be m, not y

p 166, Exercise 7.15 should refer to Exercise 7.14
p 166, Exercise 7.16 should refer to Exercise 7.14


Chapter 8

p 188, Example 8.4, first display should read
     f(xbold - u dbold) = f(vector) = 2(1-4u)**2 + (2-4u)**2
         sign^                            ^           ^
     in vector '2u' should be '4u' (twice)
       
p 188, Example 8.4, line after display should read
     'which has a minimum at u=1/3'  (not 2/3)
  
Chapter 9

p 224, line 13 (display), expression for C4 is missing 
      sigma-hat-squared


Chapter 10

p 243, line -9, (10.3.5) delete second summation 

p 252, lines -8 and -7, missing parentheses and extra t1 to read
     ln(tbold) = n1 log[2 t1 t2] + n2 log[t1 (2-t1-2t2)]
                 + n3 log[t2 (2-t2-2t1)] + 2n4 log[1-t1-t2]

p 271, bottom
 *** Most of these results depend on Y having a multivariate normal
     distribution, and some of these results are just plain wrong. ***
     Look for future corrections.

p 275, Exercise 10.24 -- depends of getting correct results on swindles


Chapter 11

p 293, Algorithm D1 (gchirv), 
  step (3) (b) should read
    (b) If Z < 0 then r = r + Z**2 / [3(Z + eta)]
  step (5) -- delete minus sign:
    If 2 log U < log .....

p 293, line -5, wrong expression for gamma
    (i)  to generate gamma(alpha, beta), call D1 with gamma = 2 alpha