SCM Repository

[matrix] Diff of /pkg/TODO
 [matrix] / pkg / TODO Diff of /pkg/TODO

revision 530, Mon Feb 7 10:58:27 2005 UTC revision 2115, Sat Feb 23 09:23:17 2008 UTC
# Line 1  Line 1
1    - Check for DimNames propagation in coercion and other operations.
2
3  - Report the problem in the Linux ldexp manual page.  The second and  - Report the problem in the Linux ldexp manual page.  The second and
4    third calls in the Synopsis should be to ldexpf and ldexpl.    third calls in the Synopsis should be to ldexpf and ldexpl.
5
6  - [,] indexing  - provide methods for "dspMatrix" and "dppMatrix"!
7
8    - implement (more) methods for supporting "packed" (symmetric / triangular)
9      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
12      (have some dtr* <-> dtp*)
13
14    - 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       [done with dgeMatrix_matrix_mm();  not yet for other classes;
18        and for _crossprod()]
19
20    -----
21
22    - "Math2" , "Math", "Arith":
23       keep triangular and symmetric Matrices when appropriate:
24       particularly desirable for  "Math2": round(), signif()
25
26    - "d" <-> "l" coercion for all "[TCR]" sparse matrices is really trivial:
27      "d" -> "l" : drops the 'x' slot
28      "l" -> "d" : construct an 'x' slot of all '1'
29      We currently have many of these conversions explicitly, e.g.
30       setAs("dsTMatrix", "lsTMatrix",
31          function(from) new("lsTMatrix", i = from@i, j = from@j, uplo = from@uplo,
32                             Dim = from@Dim, Dimnames = from@Dimnames))
33      but I would rather want to automatically construct all these coercion
34      methods at once by a ``method constructor'', i.e.,
35      for all  "dsparse*" -> "lsparse*" and vice versa.
36      How can one do this {in a documented way} ?
37
38    - Think of constructing  setAs(...) calls automatically in order to
39      basically enable all ``sensible'' as(fromMatrix, toMatrix)  calls,
40      possibly using canCoerce(.)
41
42    - setAs(<Mcl>,  "[dln]Matrix") for <Mcl> in {Matrix or denseMatrix + sparseMatrix}
43
44    - When we have a packed matrix, it's a waste to go through "full" to "sparse":
45      ==> implement
46            setAs("dspMatrix", "sparseMatrix")
47            setAs("dppMatrix", "sparseMatrix")
48            setAs("dtpMatrix", "sparseMatrix")
49      and the same for "lsp" , "ltp"  and  "nsp" , "ntp" !
50
51    - tcrossprod(x, y) : do provide methods for y != NULL
52      calling Lapack's DGEMM for "dense"
53      [2005-12-xx: done for dgeMatrix at least]
54
55    - BUGlet:  Shouldn't lose factorization here:
56      h6 <- Hilbert(6); chol(h6) ; str(h6) # has factor
57      str(H6 <- as(h6, "dspMatrix"))       # has lost factor
58      ## and the same in a similar situation involving  "dpo", "dpp"
59
60    - Factorizations: LU done; also Schur()  for  *sparse*  Matrices.
61
62    - is.na() method for all our matrices [ ==> which(*, arr.ind=TRUE) might work ]
63
64    - use  .Call(Csparse_drop, M, tol) in more places,
65      both with 'tol = 0.' to drop "values that happen to be 0" and for
66      zapsmall() methods for Csparse*
67
68    - implement .Call(Csparse_scale, ....) interfacing to cholmod_scale()
69      in src/CHOLMOD/Include/cholmod_matrixops.h : for another function
70      specifically for multiplying a cholmod_sparse object by a diagonal matrix.
71      Use it in %*% and [t]crossprod methods.
72
73    - chol() and determinant() should ``work'': proper result or "good" error
74      message.
75
76    - make sure *all* group methods have (maybe "bail-out") setMethod for "Matrix".
77      e.g. zapsmall(<pMatrix>) fails "badly"
78
79    - sum(): implement methods which work for *all* our matrices.
80
81    - Implement  expand(.) for the Cholesky() results
82      "dCHMsimpl" and  "dCHMsuper"  -- currently have no *decent* way to get at
83      the matrix factors of the corresponding matrix factorization !!
84
85    - rbind2(<sparse>, <dense>) does not work  (e.g. <dgC>, <dge>)
86
87    - <sparse> %*% <dense>  {also in crossprod/tcrossprod}  currently always
88      returns <dense>, since --> Csparse_dense_prod --> cholmod_sdmult
89      and that does only return dense.
90      When the sparse matrix is very sparse, i.e. has many rows with only zero
91      entries, it would make much sense to return sparse.
92
93    - sparse-symmetric + diagonal should stay sparse-symmetric
94      (only stays sparse): Matrix(0, 4, 4) + Diagonal(4, 1:4)
95      --> R/diagMatrix.R ('FIXME')
96      but also R/Ops.R  to ensure  sp-sym. + sp-sym. |-> sp-sym.  etc
97
98    - Diagonal(n) %*% A ---  too slow!! --> ~/R/MM/Pkg-ex/Matrix/diag-Tamas-ex.R
99
100    - ! <symmetricMatrix>  loses symmetry, both for dense and sparse matrices.
101      !M  where M is "sparseMatrix", currently always gives dense. This only
102      makes sense when M is ``really sparse''.
103
104    - msy <- as(matrix(c(2:1,1:2),2), "dsyMatrix"); str(msy)
105
106      shows that the Cholesky factorization is computed ``too quickly''.
107      Can be a big pain for largish matrices, when it is unneeded.
108
109    - example(Cholesky, echo=FALSE) ; cm <- chol(mtm); str(cm); str(mtm)
110
111      shows that chol() does not seems to make use of an already
112      present factorization and rather uses one with more '0' in x slot.
113
114    - diag(m) <- val    currently automatically works via  m[cbind(i,i)] <- val
115      This (`[<-` method) is now "smart" for diagonalMatrix, but needs also to
116      be for triangularMatrix, and probably also "dense*general*Matrix" since the
117      above currently goes via "matrix" and back instead of using the 'x' slot
118      directly; in particular, the triangular* "class property" is lost!
119
120    - image(M, ..): Think about an optional smart option which keeps
121       "0 |-> transparent" and allows colors to differentiate negative and
122       positive entries.
123
124    - examples for solve( Cholesky(.), b, system = c("A", "LDLt"....))
125      probably rather in man/CHMfactor-class.Rd than man/Cholesky.Rd
126
127  - group generics: "Arith", but also "Ops" and "Math";  see  - tr(A %*% B) {and even  tr(A %*% B %*% C) ...} are also needed
128    ?Math.data.frame  and the examples in    frequently in some computations {conditional normal distr. ...}.
129    ?SetGeneric    Since this can be done faster than by
130        sum(diag(A %*% B))  even for traditional matrices, e.g.
131                   sum(A * t(B)) or {even faster for "full" mat}
132                   crossprod(as.vector(A), as.vector(B))
133      and even more so for, e.g.  <sparse> %*% <dense>
134      {used in Soeren's 'gR' computations},
135      we should also provide a generic and methods.
136
137  - bCrosstab(): do we really want the diagonal "V:V" crosstabs?  - qr.R(qr(x)) may differ for the "same" matrix, depending on it being
138                 (if so or in any case: add to  man/bCrosstab.Rd )    sparse or dense:
139        "qr.R(<sparse>) may differ from qr.R(<dense>) because of permutations"
140
141  - src/Metis/ : one of the two Makefiles needs fixing, as changing    This is not really acceptable and currently influences  rcond() as well.
src/Metis/*.c  does not lead to recompilation.
--DB - it seems both Makefiles need fixing.  I think I have the
src/Metis/Makefile fixed but not src/Makefile
142
143  - man/Matrix.Rd :  has example with dimnames, but we just drop them!  - chol() and qr() generic:  currently have *two* arguments, and give the msg
MM thinks dimnames should be supported (but then ...)
144
145  - data/ : names 'mm' and even more 'y' are  ``too short''.    >  New generic for "chol" does not agree with implicit generic from package
146          If we really want to keep them, don't use "LazyData"    >  "base"; a new generic will be assigned with package "Matrix"
(such that one needs  data(*) explicitly);
But MM would rather want something like  ex.mm and ex.y
147
148  - "factors" maybe should move up to "Matrix"    (and ditto for "qr")
149
150  -------    It was mentioned by an R-core member that he thought it did not make
151      sense to also dispatch on 'tol' or 'pivot' ...  --> maybe change that..
152
153  We have a (at least one) basic problem :  - eigen() should become generic, and get a method at least for diagonal,
154     Currently the show() method fail sometime after coercion:    but also for symmetric -> dsyMatrix  [LAPACK dsyev() uses UPLO !],
155     e.g. 'sy' show()s wrongly, even though it "str()" fine :    but also simply for dgeMatrix (without going via tradition matrices).
156      What about Sparse?  There's fill-in, but it may still be sensible, e.g.
157      mlist <- list(1, 2:3, diag(x=5:3), 27, cbind(1,3:6), 100:101)
158      ee <- eigen(tcrossprod(bdiag(lapply(mlist, as.matrix))))
159      Matrix( signif(ee\$vectors, 3) )
160
161        (po <- crossprod(Matrix(0:3, 2))) # ok  - facmul() has no single method defined;  it looks like a good idea though
162        (ge <- as(po, "dgeMatrix")) # ok    (instead of the infamous qr.qy, qr.qty,.... functions)
(sy <- as(po, "dsyMatrix")) # BAD
str(sy) # looks fine
163
164    or  - symmpart() and skewpart()  for *sparse* matrices still use (x +/- t(x))/2
165        example(expand) # -> ex\$L and ex\$U look bad, however    and could be made more efficient.
166        as(ex\$L, "dgeMatrix") # `works'    Consider going via  asTuniq() or something very close to
167      .Arith.Csparse() in R/Ops.R
168
169    {Of course, we don't need a workaround but must understand  - many setAs(*, "[dl]..Matrix") are still needed, as long as e.g.
170     and solve the problem}    replCmat() uses as_CspClass() and drop0(.) which itself call
171      as_CspClass() quite a bit.  --> try to replace these by
172      as(*, "CsparseMatrix"); forceSymmetric, etc.

Legend:
 Removed from v.530 changed lines Added in v.2115

 root@r-forge.r-project.org ViewVC Help Powered by ViewVC 1.0.0
Thanks to:  