solveCrossprod求 X'X的逆, ( X' == t(X), 即X的行列变换矩阵 )

很多时候，我们需要计算形如X’X的逆(即X' %*% X)。这很容易实现，例如：

> x=matrix(rnorm(9),ncol=3,nrow=3)

> x

           [,1]        [,2]        [,3]

[1,] -0.1806586 -0.76340512 0.002652331

[2,] -1.8018584  0.04467943 1.416332187

[3,]  1.2785359 -1.31653513 0.180653002

> solve(crossprod(x))

          [,1]      [,2]     [,3]

[1,] 1.2181837 0.9664576 1.470940

[2,] 0.9664576 1.2010110 1.204599

[3,] 1.4709402 1.2045986 2.269921

R中的strucchange包中的函数solveCrossprod()也可完成：

> install.packages("strucchange")

> library("strucchange")

参数

> args(solveCrossprod)

function (X, method = c("qr", "chol", "solve")) 

NULL

> solveCrossprod(x,method=”qr”)

          [,1]      [,2]     [,3]

[1,] 1.2181837 0.9664576 1.470940

[2,] 0.9664576 1.2010110 1.204599

[3,] 1.4709402 1.2045986 2.269921

> solveCrossprod(x,method=”chol”)

          [,1]      [,2]     [,3]

[1,] 1.2181837 0.9664576 1.470940

[2,] 0.9664576 1.2010110 1.204599

[3,] 1.4709402 1.2045986 2.269921

> solveCrossprod(x,method=”solve”)

          [,1]      [,2]     [,3]

[1,] 1.2181837 0.9664576 1.470940

[2,] 0.9664576 1.2010110 1.204599

[3,] 1.4709402 1.2045986 2.269921

[参考]

1. > help("solveCrossprod")

solveCrossprod           package:strucchange           R Documentation

Inversion of X'X

Description:

     Computes the inverse of the cross-product of a matrix X.

Usage:

     solveCrossprod(X, method = c("qr", "chol", "solve"))
     
Arguments:

       X: a matrix, typically a regressor matrix.

  method: a string indicating whether the QR decomposition, the
          Cholesky decomposition or ‘solve’ should be used.

Details:

     Using the Cholesky decomposition of X'X (as computed by
     ‘crossprod(X)’) is computationally faster and preferred to
     ‘solve(crossprod(X))’. Using the QR decomposition of X is slower
     but should be more accurate.

Value:

     a matrix containing the inverse of ‘crossprod(X)’.

Examples:

     X <- cbind(1, rnorm(100))
     solveCrossprod(X)
     solve(crossprod(X))