1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
|
#### Methods for the sparse QR decomposition
## TODO: qr.R() generic that allows optional args ['backPermute']
## --- so we can add it to our qr.R() method, *instead* of this :
qrR <- function(qr, complete = FALSE, backPermute = TRUE, row.names = TRUE) {
ir <- seq_len(qr@Dim[if(complete) 1L else 2L])
r <- if(backPermute <- backPermute && (n <- length(qr@q)) && !isSeq(qr@q, n-1L))
qr@R[ir, order(qr@q), drop = FALSE] else
qr@R[ir, , drop = FALSE]
if(row.names && !is.null(rn <- qr@V@Dimnames[[1]])) # qr.R() in 'base' gives rownames
r@Dimnames[[1]] <- rn[seq_len(r@Dim[1L])]
if(complete || backPermute) r else as(r, "triangularMatrix")
}
setMethod("qr.R", signature(qr = "sparseQR"),
function(qr, complete = FALSE) {
if(nonTRUEoption("Matrix.quiet.qr.R") && nonTRUEoption("Matrix.quiet"))
warning("qr.R(<sparse>) may differ from qr.R(<dense>) because of permutations. Possibly use our qrR() instead")
qrR(qr, complete=complete, backPermute=FALSE)
})
## if(identical("", as.character(formals(qr.Q)$Dvec))) { # "new"
setMethod("qr.Q", "sparseQR",
function(qr, complete=FALSE, Dvec)
{
d <- qr@Dim
## ir <- seq_len(d[k <- if(complete) 1L else 2L])
k <- if(complete) 1L else 2L
if(missing(Dvec)) Dvec <- rep.int(1, if(complete) d[1] else min(d))
D <- .sparseDiagonal(d[1], x=Dvec, cols=0L:(d[k] -1L))
qr.qy(qr, D)
})
## } else {
## setMethod("qr.Q", "sparseQR",
## function(qr, complete=FALSE, Dvec = rep.int(1, if(complete) d[1] else min(d)))
## {
## d <- qr@Dim
## ir <- seq_len(d[k <- if(complete) 1L else 2L])
## D <- .sparseDiagonal(d[1], x=Dvec, cols=0L:(d[k] -1L))
## qr.qy(qr, D)
## })
## }
## NB: Here, the .Call()s to sparseQR_qty all set keep_names = TRUE
## --- instead of allowing it to become an argument,
## mainly because the base functions qr.qy() / qr.qty() have no '...' formal argument
## To change, would make these *implicit* generics in 'methods' - as qr.R
## Also, qr() itself has keep.names = TRUE/FALSE -- should be enough
setMethod("qr.qy", signature(qr = "sparseQR", y = "ddenseMatrix"),
function(qr, y) .Call(sparseQR_qty, qr, y, FALSE, TRUE),
valueClass = "dgeMatrix")
setMethod("qr.qy", signature(qr = "sparseQR", y = "matrix"),
function(qr, y) .Call(sparseQR_qty, qr, y, FALSE, TRUE),
valueClass = "dgeMatrix")
setMethod("qr.qy", signature(qr = "sparseQR", y = "numeric"),
## drop to vector {to be 100% standard-R-matrix compatible} :
function(qr, y) .Call(sparseQR_qty, qr, y, FALSE, TRUE)@x)
setMethod("qr.qy", signature(qr = "sparseQR", y = "Matrix"),
function(qr, y) .Call(sparseQR_qty, qr,
as(as(y, "denseMatrix"),"dgeMatrix"), FALSE, TRUE),
valueClass = "dgeMatrix")
setMethod("qr.qty", signature(qr = "sparseQR", y = "ddenseMatrix"),
function(qr, y) .Call(sparseQR_qty, qr, y, TRUE, TRUE),
valueClass = "dgeMatrix")
setMethod("qr.qty", signature(qr = "sparseQR", y = "matrix"),
function(qr, y) .Call(sparseQR_qty, qr, y, TRUE, TRUE),
valueClass = "dgeMatrix")
setMethod("qr.qty", signature(qr = "sparseQR", y = "numeric"),
function(qr, y) .Call(sparseQR_qty, qr, y, TRUE, TRUE)@x)
setMethod("qr.qty", signature(qr = "sparseQR", y = "Matrix"),
function(qr, y) .Call(sparseQR_qty, qr,
as(as(y, "denseMatrix"),"dgeMatrix"), TRUE, TRUE),
valueClass = "dgeMatrix")
## FIXME: really should happen in C, i.e sparseQR_coef() in ../src/sparseQR.c :
.coef.trunc <- function(qr, res, drop=FALSE) {
if(!all((d <- lengths(res@Dimnames)) == 0L) && !identical(d, D <- res@Dim)) {
## Fix dimnames from dim (when not NULL !) :
if(d[[1]]) length(res@Dimnames[[1]]) <- D[[1]]
if(d[[2]]) length(res@Dimnames[[2]]) <- D[[2]]
}
res[seq_len(ncol(qr@R)),,drop=drop]
}
setMethod("qr.coef", signature(qr = "sparseQR", y = "ddenseMatrix"),
function(qr, y)
.coef.trunc(qr, .Call(sparseQR_coef, qr, y)),
valueClass = "dgeMatrix")
setMethod("qr.coef", signature(qr = "sparseQR", y = "matrix"),
function(qr, y)
.coef.trunc(qr, .Call(sparseQR_coef, qr, y)),
valueClass = "dgeMatrix")
setMethod("qr.coef", signature(qr = "sparseQR", y = "numeric"),
function(qr, y)
.coef.trunc(qr, .Call(sparseQR_coef, qr, y), drop=TRUE))
setMethod("qr.coef", signature(qr = "sparseQR", y = "Matrix"),
function(qr, y)
.coef.trunc(qr, .Call(sparseQR_coef, qr,
as(as(y, "denseMatrix"),"dgeMatrix"))),
valueClass = "dgeMatrix")
## qr.resid() & qr.fitted() : ---------------------------
setMethod("qr.resid", signature(qr = "sparseQR", y = "ddenseMatrix"),
function(qr, y)
.Call(sparseQR_resid_fitted, qr, y, TRUE),
valueClass = "dgeMatrix")
setMethod("qr.resid", signature(qr = "sparseQR", y = "matrix"),
function(qr, y)
.Call(sparseQR_resid_fitted, qr, y, TRUE),
valueClass = "dgeMatrix")
setMethod("qr.resid", signature(qr = "sparseQR", y = "numeric"),
function(qr, y) drop(.Call(sparseQR_resid_fitted, qr, y, TRUE)))
setMethod("qr.resid", signature(qr = "sparseQR", y = "Matrix"),
function(qr, y)
.Call(sparseQR_resid_fitted, qr,
as(as(y, "denseMatrix"),"dgeMatrix"), TRUE),
valueClass = "dgeMatrix")
setMethod("qr.fitted", signature(qr = "sparseQR", y = "ddenseMatrix"),
function(qr, y, k)
.Call(sparseQR_resid_fitted, qr, y, FALSE),
valueClass = "dgeMatrix")
setMethod("qr.fitted", signature(qr = "sparseQR", y = "matrix"),
function(qr, y, k)
.Call(sparseQR_resid_fitted, qr, y, FALSE),
valueClass = "dgeMatrix")
setMethod("qr.fitted", signature(qr = "sparseQR", y = "numeric"),
function(qr, y, k) drop(.Call(sparseQR_resid_fitted, qr, y, FALSE)))
setMethod("qr.fitted", signature(qr = "sparseQR", y = "Matrix"),
function(qr, y, k)
.Call(sparseQR_resid_fitted, qr,
as(as(y, "denseMatrix"),"dgeMatrix"), FALSE),
valueClass = "dgeMatrix")
##
setMethod("solve", signature(a = "sparseQR", b = "ANY"),
function(a, b, ...) qr.coef(a, b))
|
