The R Project SVN R

Rev

Rev 88706 | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 88706 Rev 90082
Line 50... Line 50...
50 
  \R platforms (see \code{\link{prcomp}}).
 50 
  \R platforms (see \code{\link{prcomp}}).
51 
 
 51 
 
52 
  When \code{add = TRUE}, a minimal additive constant \eqn{c*} is
 52 
  When \code{add = TRUE}, a minimal additive constant \eqn{c*} is
53 
  computed such that the dissimilarities \eqn{d_{ij} + c*}{d[i,j] +
 53 
  computed such that the dissimilarities \eqn{d_{ij} + c*}{d[i,j] +
54 
  c*} are Euclidean and hence can be represented in \code{n - 1}
 54 
  c*} are Euclidean and hence can be represented in \code{n - 1}
55 
  dimensions.  Whereas S (Becker \abbr{et al.}, 1988) computes this
 55 
  dimensions.  Whereas S \bibcitep{R:Becker+Chambers+Wilks:1988} computes this
56 
  constant using an approximation suggested by \I{Torgerson}, \R uses the
 56 
  constant using an approximation suggested by \I{Torgerson}, \R uses the
57 
  analytical solution of \bibcitet{R:Cailliez:1983}, see also
 57 
  analytical solution of \bibcitet{R:Cailliez:1983}, see also
58 
  \bibcitet{R:Cox+Cox:2001}.
 58 
  \bibcitet{R:Cox+Cox:2001}.
59 
  Note that because of numerical errors the computed eigenvalues need
 59 
  Note that because of numerical errors the computed eigenvalues need
60 
  not all be non-negative, and even theoretically the representation
 60 
  not all be non-negative, and even theoretically the representation