SCM

SCM Repository

[matrix] Annotation of /pkg/TODO
ViewVC logotype

Annotation of /pkg/TODO

Parent Directory Parent Directory | Revision Log Revision Log


Revision 2120 - (view) (download)

1 : maechler 2112 - Check for DimNames propagation in coercion and other operations.
2 :    
3 : bates 344 - Report the problem in the Linux ldexp manual page. The second and
4 :     third calls in the Synopsis should be to ldexpf and ldexpl.
5 : maechler 472
6 : maechler 634 - provide methods for "dspMatrix" and "dppMatrix"!
7 :    
8 : maechler 538 - implement (more) methods for supporting "packed" (symmetric / triangular)
9 : maechler 634 matrices; particularly something like pack() and unpack() [to/from our
10 :     classes from/to "numeric"] --- have already man/unpack.Rd but no method yet!
11 : maechler 538
12 : maechler 834 (have some dtr* <-> dtp*)
13 :    
14 : bates 645 - combine the C functions for multiplication by special forms and
15 :     solution wrt special forms by using a 'right' argument and a
16 :     'classed' argument.
17 : maechler 675 [done with dgeMatrix_matrix_mm(); not yet for other classes;
18 :     and for _crossprod()]
19 :    
20 : bates 1578 -----
21 : bates 712
22 : maechler 935 - "Math2" , "Math", "Arith":
23 :     keep triangular and symmetric Matrices when appropriate:
24 :     particularly desirable for "Math2": round(), signif()
25 : maechler 956
26 : maechler 2120 For triangular matrices, more specifically make sure the four rules of
27 :     "triangular matrix algebra" (Golub+Van Loan 1996, 3.1.8, p.93) are
28 :     fulfilled; currently <dtC> %*% <dtC> --> <dgC>
29 :    
30 : maechler 956 - "d" <-> "l" coercion for all "[TCR]" sparse matrices is really trivial:
31 :     "d" -> "l" : drops the 'x' slot
32 :     "l" -> "d" : construct an 'x' slot of all '1'
33 :     We currently have many of these conversions explicitly, e.g.
34 :     setAs("dsTMatrix", "lsTMatrix",
35 :     function(from) new("lsTMatrix", i = from@i, j = from@j, uplo = from@uplo,
36 :     Dim = from@Dim, Dimnames = from@Dimnames))
37 :     but I would rather want to automatically construct all these coercion
38 :     methods at once by a ``method constructor'', i.e.,
39 :     for all "dsparse*" -> "lsparse*" and vice versa.
40 :     How can one do this {in a documented way} ?
41 : maechler 1087
42 : maechler 2048 - Think of constructing setAs(...) calls automatically in order to
43 :     basically enable all ``sensible'' as(fromMatrix, toMatrix) calls,
44 :     possibly using canCoerce(.)
45 :    
46 :     - setAs(<Mcl>, "[dln]Matrix") for <Mcl> in {Matrix or denseMatrix + sparseMatrix}
47 :    
48 :     - When we have a packed matrix, it's a waste to go through "full" to "sparse":
49 :     ==> implement
50 :     setAs("dspMatrix", "sparseMatrix")
51 :     setAs("dppMatrix", "sparseMatrix")
52 :     setAs("dtpMatrix", "sparseMatrix")
53 :     and the same for "lsp" , "ltp" and "nsp" , "ntp" !
54 :    
55 : maechler 1097 - tcrossprod(x, y) : do provide methods for y != NULL
56 :     calling Lapack's DGEMM for "dense"
57 : maechler 1109 [2005-12-xx: done for dgeMatrix at least]
58 :    
59 : maechler 1201 - BUGlet: Shouldn't lose factorization here:
60 :     h6 <- Hilbert(6); chol(h6) ; str(h6) # has factor
61 :     str(H6 <- as(h6, "dspMatrix")) # has lost factor
62 :     ## and the same in a similar situation involving "dpo", "dpp"
63 : maechler 1226
64 : maechler 1592 - Factorizations: LU done; also Schur() for *sparse* Matrices.
65 : maechler 1253
66 : maechler 1575 - is.na() method for all our matrices [ ==> which(*, arr.ind=TRUE) might work ]
67 : maechler 1592
68 : maechler 1619 - use .Call(Csparse_drop, M, tol) in more places,
69 :     both with 'tol = 0.' to drop "values that happen to be 0" and for
70 :     zapsmall() methods for Csparse*
71 : maechler 1654
72 :     - implement .Call(Csparse_scale, ....) interfacing to cholmod_scale()
73 :     in src/CHOLMOD/Include/cholmod_matrixops.h : for another function
74 :     specifically for multiplying a cholmod_sparse object by a diagonal matrix.
75 :     Use it in %*% and [t]crossprod methods.
76 :    
77 :     - chol() and determinant() should ``work'': proper result or "good" error
78 :     message.
79 : maechler 1659
80 : maechler 1714 - make sure *all* group methods have (maybe "bail-out") setMethod for "Matrix".
81 :     e.g. zapsmall(<pMatrix>) fails "badly"
82 : maechler 1725
83 : maechler 1738 - sum(): implement methods which work for *all* our matrices.
84 :    
85 :     - Implement expand(.) for the Cholesky() results
86 :     "dCHMsimpl" and "dCHMsuper" -- currently have no *decent* way to get at
87 :     the matrix factors of the corresponding matrix factorization !!
88 :    
89 : maechler 2112 - rbind2(<sparse>, <dense>) does not work (e.g. <dgC>, <dge>)
90 : maechler 1799
91 : maechler 1833 - <sparse> %*% <dense> {also in crossprod/tcrossprod} currently always
92 :     returns <dense>, since --> Csparse_dense_prod --> cholmod_sdmult
93 :     and that does only return dense.
94 :     When the sparse matrix is very sparse, i.e. has many rows with only zero
95 :     entries, it would make much sense to return sparse.
96 :    
97 :     - sparse-symmetric + diagonal should stay sparse-symmetric
98 :     (only stays sparse): Matrix(0, 4, 4) + Diagonal(4, 1:4)
99 : maechler 1845 --> R/diagMatrix.R ('FIXME')
100 :     but also R/Ops.R to ensure sp-sym. + sp-sym. |-> sp-sym. etc
101 : maechler 1833
102 : maechler 2048 - Diagonal(n) %*% A --- too slow!! --> ~/R/MM/Pkg-ex/Matrix/diag-Tamas-ex.R
103 :    
104 : maechler 1833 - ! <symmetricMatrix> loses symmetry, both for dense and sparse matrices.
105 :     !M where M is "sparseMatrix", currently always gives dense. This only
106 :     makes sense when M is ``really sparse''.
107 : maechler 1855
108 : maechler 2005 - msy <- as(matrix(c(2:1,1:2),2), "dsyMatrix"); str(msy)
109 :    
110 :     shows that the Cholesky factorization is computed ``too quickly''.
111 :     Can be a big pain for largish matrices, when it is unneeded.
112 :    
113 : maechler 2018 - example(Cholesky, echo=FALSE) ; cm <- chol(mtm); str(cm); str(mtm)
114 :    
115 :     shows that chol() does not seems to make use of an already
116 :     present factorization and rather uses one with more '0' in x slot.
117 :    
118 : maechler 2005 - diag(m) <- val currently automatically works via m[cbind(i,i)] <- val
119 : maechler 2115 This (`[<-` method) is now "smart" for diagonalMatrix, but needs also to
120 :     be for triangularMatrix, and probably also "dense*general*Matrix" since the
121 : maechler 2005 above currently goes via "matrix" and back instead of using the 'x' slot
122 : maechler 2115 directly; in particular, the triangular* "class property" is lost!
123 : maechler 2043
124 : maechler 2048 - image(M, ..): Think about an optional smart option which keeps
125 :     "0 |-> transparent" and allows colors to differentiate negative and
126 :     positive entries.
127 : maechler 2043
128 :     - examples for solve( Cholesky(.), b, system = c("A", "LDLt"....))
129 :     probably rather in man/CHMfactor-class.Rd than man/Cholesky.Rd
130 : maechler 2072
131 :     - tr(A %*% B) {and even tr(A %*% B %*% C) ...} are also needed
132 :     frequently in some computations {conditional normal distr. ...}.
133 :     Since this can be done faster than by
134 :     sum(diag(A %*% B)) even for traditional matrices, e.g.
135 :     sum(A * t(B)) or {even faster for "full" mat}
136 :     crossprod(as.vector(A), as.vector(B))
137 :     and even more so for, e.g. <sparse> %*% <dense>
138 :     {used in Soeren's 'gR' computations},
139 :     we should also provide a generic and methods.
140 : maechler 2103
141 :     - qr.R(qr(x)) may differ for the "same" matrix, depending on it being
142 :     sparse or dense:
143 :     "qr.R(<sparse>) may differ from qr.R(<dense>) because of permutations"
144 :    
145 :     This is not really acceptable and currently influences rcond() as well.
146 : maechler 2105
147 : maechler 2112 - chol() and qr() generic: currently have *two* arguments, and give the msg
148 : maechler 2105
149 : maechler 2112 > New generic for "chol" does not agree with implicit generic from package
150 :     > "base"; a new generic will be assigned with package "Matrix"
151 :    
152 : maechler 2106 (and ditto for "qr")
153 : maechler 2105
154 :     It was mentioned by an R-core member that he thought it did not make
155 :     sense to also dispatch on 'tol' or 'pivot' ... --> maybe change that..
156 : maechler 2106
157 :     - eigen() should become generic, and get a method at least for diagonal,
158 :     but also for symmetric -> dsyMatrix [LAPACK dsyev() uses UPLO !],
159 :     but also simply for dgeMatrix (without going via tradition matrices).
160 :     What about Sparse? There's fill-in, but it may still be sensible, e.g.
161 :     mlist <- list(1, 2:3, diag(x=5:3), 27, cbind(1,3:6), 100:101)
162 :     ee <- eigen(tcrossprod(bdiag(lapply(mlist, as.matrix))))
163 :     Matrix( signif(ee$vectors, 3) )
164 :    
165 :     - facmul() has no single method defined; it looks like a good idea though
166 :     (instead of the infamous qr.qy, qr.qty,.... functions)
167 : maechler 2112
168 :     - symmpart() and skewpart() for *sparse* matrices still use (x +/- t(x))/2
169 :     and could be made more efficient.
170 :     Consider going via asTuniq() or something very close to
171 :     .Arith.Csparse() in R/Ops.R
172 :    
173 : maechler 2115 - many setAs(*, "[dl]..Matrix") are still needed, as long as e.g.
174 :     replCmat() uses as_CspClass() and drop0(.) which itself call
175 :     as_CspClass() quite a bit. --> try to replace these by
176 :     as(*, "CsparseMatrix"); forceSymmetric, etc.

root@r-forge.r-project.org
ViewVC Help
Powered by ViewVC 1.0.0  
Thanks to:
Vienna University of Economics and Business Powered By FusionForge