| Line 68... |
Line 68... |
| 68 |
The d, p and q functions are calculated based on
|
68 |
The d, p and q functions are calculated based on
|
| 69 |
\code{cwilcox(k, m, n)},
|
69 |
\code{cwilcox(k, m, n)},
|
| 70 |
the number of choices with statistic \code{k} from samples of size
|
70 |
the number of choices with statistic \code{k} from samples of size
|
| 71 |
\code{m} and \code{n}.
|
71 |
\code{m} and \code{n}.
|
| 72 |
Since \R 4.6.0, \code{cwilcox} is calculated using an implementation
|
72 |
Since \R 4.6.0, \code{cwilcox} is calculated using an implementation
|
| 73 |
of the formula from \bibcitet{R:Löffler:1983} provided by
|
73 |
of the formula from \bibcitet{R:Loeffler:1983} provided by
|
| 74 |
\I{Andreas Löffler}, \I{Aidan Lakshman}, and \I{Ivan Krylov}
|
74 |
\I{Andreas Löffler}, \I{Aidan Lakshman}, and \I{Ivan Krylov}
|
| 75 |
(\PR{18655}), substantially reducing the memory complexity of the
|
75 |
(\PR{18655}), substantially reducing the memory complexity of the
|
| 76 |
older implementation based on the recursions in
|
76 |
older implementation based on the recursions in
|
| 77 |
\bibcitet{R:Mann+Whitney:1947}.
|
77 |
\bibcitet{R:Mann+Whitney:1947}.
|
| 78 |
Then \code{dwilcox} normalizes,
|
78 |
Then \code{dwilcox} normalizes,
|