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
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
|
optimr <- function(par, fn, gr=NULL, hess=NULL, lower=-Inf, upper=Inf,
method=NULL, hessian=FALSE, control=list(), ...) {
## 180706: Problem with maxit. Likely issue is that opm gets ctrldefault and
## updates with actual control list. But here in nlm, maxit defaults to 100.
## We need to be sure user has not tried to set both controls "maximize" and "fnscale"
# First case: User set's nothing
if (is.null(control$maximize) && is.null(control$fnscale)){control$fnscale = 1.0}
# Otherwise, user has set one or both. If both, there may be conflict.
else if (! is.null(control$maximize) ) { # user has set "maximize" control
if (control$maximize) { # user wants maximize. Check fnscale
if (is.null(control$fnscale)) {
control$fnscale <- -1.0 # set maximization
} else if (control$fnscale < 0.0) {
warning("User has set control$maximize = TRUE and admissible control$fnscale")
} else { stop("control$fnscale and control$maximize conflict") }
} else # control$maximize set to FALSE by user
if (is.null(control$fnscale)) {
control$fnscale <- 1.0 # set maximization
} else if (control$fnscale > 0.0) {
warning("User has set control$maximize = FALSE and admissible control$fnscale")
} else { stop("control$fnscale and control$maximize conflict") }
} # end is.null(control$maximize)
npar <- length(par)
ctrl <- ctrldefault(npar)
ncontrol <- names(control)
nctrl <- names(ctrl)
for (onename in ncontrol) {
if (onename %in% nctrl) {
if (! is.null(control[onename]) || ! is.na(control[onename]) )
ctrl[onename]<-control[onename]
}
}
control <- ctrl # note the copy back! control now has a FULL set of values
# with user input over-writing defaults as appropriate
## 180706: Should we try to streamline?
if (is.null(method)) method <- control$defmethod
outmethod <- checksolver(method, control$allmeth, control$allpkg) # there will only be one!
if (is.null(outmethod)) {
if (control$trace > 0) cat("Solver ",method," missing\n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 8888 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- paste("Missing method ",method)
ans$hessian <- NULL
return(ans) # can't proceed without solver
}
# Check if bounded
bdmsk <- bmchk(par, lower=lower, upper=upper, shift2bound=TRUE)
if (bdmsk$parchanged) warning("Parameter(s) changed to nearest bounds\n")
control$have.bounds <- bdmsk$bounds # and set a control value
orig.method <- method
orig.gr <- gr
orig.fn <- fn
if (is.null(hessian) ){
savehess <- FALSE
} else { savehess <- hessian } # logical -- whether to save hessian
if (is.null(control$trace)) control$trace <- control$trace
if (is.null(control$parscale)) {
pscale <- rep(1,npar)
if(control$trace > 0) { cat("Unit parameter scaling\n") }
} else {
pscale <- control$parscale
if(control$trace > 0) {
cat("Parameter scaling:")
print(pscale)
}
}
spar <- par/pscale # scaled parameters
slower <- -Inf
supper <- Inf # to ensure defined
if (control$have.bounds) {
slower <- lower/pscale
supper <- upper/pscale
}
# 160615 -- decided to postpone adding nloptr
efn <- function(spar, ...) {
# rely on pscale being defined in this enclosing environment
par <- spar*pscale
val <- fn(par, ...) * control$fnscale
}
appgr<-FALSE # so far assuming analytic gradient
# DO NOT PROVIDE A DEFAULT -- LET METHOD DO THIS
# if (is.null(gr)) gr <- control$defgrapprox
# if (is.null(gr)) cat("gr is NULL\n")
if (is.character(gr)) {
appgr <- TRUE # to inform us that we are using approximation
egr <- function(spar, ...){
if (control$trace > 1) {
cat("control$fnscale =",control$fnscale," pscale=")
print(pscale)
cat("gr:")
print(gr)
cat("par:")
print(par)
}
par <- spar*pscale
result <- do.call(gr, list(par, userfn=fn, ...)) * control$fnscale
}
} else {
if (is.null(gr)) {egr <- NULL}
else {
egr <- function(spar, ...) {
par <- spar*pscale
result <- gr(par, ...) * pscale * control$fnscale
}
}
} # end egr definition
if (is.null(hess)) {
ehess <- NULL
} else { ehess <- function(spar, ...) {
par <- spar*pscale
result <- hess(par, ...) * pscale * pscale * control$fnscale
result
}
}
if (appgr && (control$trace>0)) cat("Using numerical approximation '",gr,"' to gradient in optimru()\n")
nlmfn <- function(spar, ...){
f <- efn(spar, ...)
if (is.null(egr)) { g <- NULL} else {g <- egr(spar, ...) }
attr(f,"gradient") <- g
if (is.null(ehess)) { h <- NULL } else {h <- ehess(spar, ...) }
attr(f,"hessian") <- h
f
}
## cat("Check nlmfn at spar\n")
## print(nlmfn(spar, ...))
## Masks
maskmeth <- control$maskmeth
msk <- bdmsk$bdmsk # Only need the masks bit from here on
if (any(msk == 0) ) {
if ( !(method %in% maskmeth) ) {
stopmsg <- paste("Method ",method," cannot handle masked (fixed) parameters")
stop(stopmsg)
}
if (control$trace > 0) cat("Masks present\n")
}
# replacement for optim to minimize using a single method
# time is in opm(), but not here
# The structure has par, value, counts, convergence, message, hessian
# Run a single method
# expand bounds
if (length(lower) == 1 && is.finite(lower) ) lower<-rep(lower,npar)
if (length(upper) == 1 && is.finite(upper) ) upper<-rep(upper,npar)
mcontrol <- list() # define the control list
# Methods from optim()
if (method == "Nelder-Mead" ||
method == "BFGS" ||
method == "L-BFGS-B" ||
method == "CG" ||
method == "SANN") {
# Take care of methods from optim(): Nelder-Mead, BFGS, L-BFGS-B, CG
mcontrol$maxit <- control$maxit
if (! is.null(control$maxit)) {mcontrol$maxit <- control$maxit}
mcontrol$trace <- control$trace
mcontrol$parscale <- NULL # using user fn
mcontrol$fnscale <- NULL
## mcontrol$fnscale <- control$fnscale # 180313 Carlo Lapid ?? wrong, use efn, egr
# Note: hessian always FALSE in these calls. But savehess may recover it.
# cat("Before optim() call - control$have.bounds =",control$have.bounds,"\n")
if (control$have.bounds) {
if (method != "L-BFGS-B") {
errmsg <- "optim() can only handle bounds with L-BFGS-B\n"
if (control$trace > 0) cat(errmsg,"\n")
ans <- list()
class(ans)[1] <- "try-error"
warning("optimr: optim() with bounds ONLY uses L-BFGS-B")
} else {
ans <- try(optim(par=par, fn=efn, gr=egr,
lower=lower, upper=upper, method="L-BFGS-B", hessian=FALSE,
control=mcontrol, ...))
}
} else {
ans <- try(optim(par=par, fn=efn, gr=egr,
method=method, hessian=FALSE, control=mcontrol, ...))
}
if (inherits(ans,"try-error")) { # bad result -- What to do?
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
errmsg <- "optim method failure\n"
if (method != "L-BFGS-B")
errmsg <- paste("optim() with bounds ONLY uses L-BFGS-B: ", errmsg)
if (control$trace>0) cat(errmsg)
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- errmsg
} # otherwise ans is OK and we return it
## return(ans) # to ensure we return
} # end if using optim() methods
## --------------------------------------------
else if (method == "nlminb") {
# Here we use portLib routine nlminb rather than optim as our minimizer
mcontrol$iter.max<-mcontrol$maxit # different name for iteration limit in this routine
mcontrol$maxit<-NULL # and we null it out
mcontrol$abs.tol <- 0 # To fix issues when minimum is less than 0. 20100711
mcontrol$eval.max <- control$maxfeval
if ( is.null(control$trace) || is.na(control$trace) || control$trace == 0) {
mcontrol$trace = 0
} else {
mcontrol$trace = 1 # this is EVERY iteration. nlminb trace is freq of reporting.
}
ans <- try(nlminb(start=spar, objective=efn, gradient=egr, hessian=ehess, lower=slower,
upper=supper, control=mcontrol, ...))
if (! inherits(ans, "try-error")) {
# Translate output to common format and names
ans$value<-ans$objective
ans$par <- ans$par*pscale
ans$objective<-NULL
ans$counts[1] <- ans$evaluations[1]
ans$counts[2] <- ans$evaluations[2]
ans$evaluations<-NULL # cleanup
ans$iterations<-NULL
ans$hessian <- NULL
} else { # bad result -- What to do?
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
if (control$trace>0) cat("nlminb failure\n")
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- "nlminb failed" # 180318 change from NULL
ans$hessian <- NULL
}
## return(ans)
} ## end if using nlminb
## --------------------------------------------
else if (method == "nlm") { # Use stats package nlm routine
# if (is.null(gr)) { stop("optimr -- nlm -- we do not allow gr = NULL") }
if (! is.null(control$maxit) ) {iterlim <- control$maxit }
else { iterlim <- 100 }
print.level <- 0
errmsg <- NULL
if (control$have.bounds) {
if(control$trace > 0) cat("nlm cannot handle bounds\n")
errmsg <- "nlm cannot handle bounds\n"
## stop("nlm tried with bounds")
ans <- list()
class(ans)[1] <- "try-error"
} else {
if (! is.null(control$trace) && (control$trace > 0) ) {print.level <- 2 }
ans <- try(nlm(f=nlmfn, p=spar, iterlim=iterlim, print.level=print.level, ...))
}
if (! inherits(ans, "try-error")) {
if (ans$code == 1 || ans$code == 2 || ans$code == 3) ans$convergence <- 0
if (ans$code == 4) ans$convergence <- 1
if (ans$code == 5) ans$convergence <- 5
# Translate output to common format
ans$value <- ans$minimum
ans$minimum <- NULL
ans$par <- ans$estimate*pscale
ans$estimate <- NULL
ans$counts[2] <- ans$iterations
ans$counts[1] <- NA
ans$iterations <- NULL
ans$hessian <- NULL
ans$gradient <- NULL # We lose information here
ans$message <- paste("nlm: Convergence indicator (code) = ",ans$code)
ans$code <- NULL
} else {
if (control$trace > 0) cat("nlm failed for this problem\n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- "nlm failed" # 180318 change from NULL
ans$hessian <- NULL
}
print.level <- NULL # clean up
## return(ans)
} # end if using nlm
## --------------------------------------------
else if (method == "Rcgmin") { # Use Rcgmin routine (ignoring masks)
mcontrol$trace <- control$trace
mcontrol$maxit <- control$maxit # 151217 JN
if (! is.null(egr)) {
if (control$have.bounds) { # 151220 -- this was not defined
# 170919 -- explicit reference to package
ans <- try(Rcgminb(par=spar, fn=efn, gr=egr, lower=slower,
upper=supper, bdmsk=msk, control=mcontrol, ...))
} else {
ans <- try(Rcgminu(par=spar, fn=efn, gr=egr, control=mcontrol, ...))
}
}
if (!is.null(egr) && !inherits(ans, "try-error")) {
ans$par <- ans$par*pscale
ans$message <- NA
ans$hessian <- NULL
ans$bdmsk <- NULL # clear this
} else {
if (control$trace>0) {
cat("Rcgmin failed for current problem \n")
if(is.null(egr)) cat("Note: Rcgmin needs gradient function specified\n")
}
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
if(is.null(egr)) {
ans$message <- "Must specify gradient function for Rcgmin"
ans$convergence <- 9998 # for no gradient where needed
}
ans$hessian <- NULL
}
## return(ans)
} ## end if using Rcgmin
## --------------------------------------------
else if (method == "Rvmmin") { # Use Rvmmin routine (ignoring masks??)
mcontrol$maxit <- control$maxit
mcontrol$maxfeval <- control$maxfeval
mcontrol$trace <- control$trace # 140902 Note no check on validity of values
if (! is.null(egr)) {
ans <- try(Rvmmin(par=spar, fn=efn, gr=egr, lower=slower,
upper=supper, bdmsk=msk, control=mcontrol, ...))
}
if (control$trace > 2) {
cat("Rvmmin ans:")
print(ans)
}
if (! is.null(egr) && !inherits(ans, "try-error")) {
ans$par <- ans$par*pscale
ans$bdmsk <- NULL
} else {
if (control$trace>0) {
cat("Rvmmin failed for current problem \n")
if(is.null(egr)) cat("Note: Rvmmin needs gradient function specified\n")
}
ans<-list() # ans not yet defined, so set as list
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
if(is.null(egr)) {
ans$message <- "Must specify gradient function for Rvmmin"
ans$convergence <- 9998 # for no gradient where needed
}
ans$hessian <- NULL
}
## return(ans)
} ## end if using Rvmmin
## --------------------------------------------
else if (method == "snewton") { # Use snewton routine (no bounds or masks??)
mcontrol$maxit <- control$maxit
mcontrol$maxfeval <- control$maxfeval # changed from maxfevals 180321
mcontrol$trace <- control$trace # 140902 Note no check on validity of values
ans<-list(par=NA, value=NA, counts=NA, convergence=-1,
message=NA, hessian=NULL) # ans not yet defined, so set as list
if ( (! is.null(egr)) && (! is.null(ehess)) ) {
if (control$have.bounds) { # 170919 make package explicit
stop("snewton does not handle bounds") # ?? error -- doesn't handle bounds
}
tans <- try( snewton(par=spar, fn=efn, gr=egr, hess=ehess, control=mcontrol,...))
if (control$trace>1) {
cat("snewton returns tans:")
print(tans)
}
if (inherits(tans, "try-error")) {
ans$message <- "snewton failed"
ans$convergence <- 9999
if (control$trace>0) {
cat(ans$message,"\n")
}
}
} else {
if(is.null(egr)) {
ans$message <- "Must specify gradient function for snewton"
ans$convergence <- 9998 # for no gradient where needed
warning("Note: snewton needs gradient function specified")
}
if(is.null(ehess)) {
ans$message <- "Must specify Hessian function (hess) for snewton"
ans$convergence <- 9997 # for no gradient where needed
warning("Note: snewton needs Hessian function (hess) specified")
}
}
if (ans$convergence > 9996){
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
# Note: in optim() no provision for hessian count
ans$hessian <- NULL
if (control$trace>1) {
cat("snewton falure ans:")
print(ans)
}
} else { # have an answer
ans$par <- tans$par*pscale
ans$value <- tans$value
attr(ans, "gradient") <- tans$grad
if(hessian) ans$hessian <- tans$Hess
ans$counts[1] <- tans$counts$nfn
ans$counts[2] <- tans$counts$ngr
ans$message <- tans$message
ans$convergence <- tans$convcode
tans <- NULL # probably unnecessary, but for safety
if (control$trace>1) {
cat("rejigged ans:")
print(ans)
}
} # end have answer
## return(ans)
} ## end if using snewton
## --------------------------------------------
else if (method == "snewtonm") { # Use snewtonm routine (no bounds or masks??)
mcontrol$maxit <- control$maxit
mcontrol$maxfeval <- control$maxfeval # changed from maxfevals 180321
mcontrol$trace <- control$trace # 140902 Note no check on validity of values
ans<-list(par=NA, value=NA, counts=NA, convergence=-1,
message=NA, hessian=NULL) # ans not yet defined, so set as list
if ( (! is.null(egr)) && (! is.null(ehess)) ) {
if (control$have.bounds) { # 170919 make package explicit
stop("snewtonm does not handle bounds") # ?? error -- doesn't handle bounds
}
tans <- try( snewtonm(par=spar, fn=efn, gr=egr, hess=ehess, control=mcontrol,...))
if (control$trace>0) {
cat("snewtonm returns tans:")
print(tans)
}
if (inherits(tans, "try-error")) {
ans$message <- "snewtonm failed"
ans$convergence <- 9999
if (control$trace>0) {
cat(ans$message,"\n")
}
}
} else {
if(is.null(egr)) {
ans$message <- "Must specify gradient function for snewtonm"
ans$convergence <- 9998 # for no gradient where needed
warning("Note: snewtonm needs gradient function specified")
}
if(is.null(ehess)) {
ans$message <- "Must specify Hessian function (hess) for snewtonm"
ans$convergence <- 9997 # for no Hessian where needed
warning("Note: snewtonm needs Hessian function specified")
}
} # end of fails
if (ans$convergence > 9996){ # Bad solution
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
# Note: in optim() no provision for hessian count
ans$hessian <- NULL
if (control$trace>1) {
cat("snewton falure ans:")
print(ans)
}
} else { # have an answer
# cat("copy answer with tans$par:\n")
# print(tans$par)
ans$par <- tans$par*pscale
ans$value <- tans$value
attr(ans, "gradient") <- tans$grad
if(hessian) ans$hessian <- tans$Hess
ans$counts[1] <- tans$counts$nfn
ans$counts[2] <- tans$counts$ngr
ans$message <- tans$message
ans$convergence <- tans$convcode
tans <- NULL # probably unnecessary, but for safety
if (control$trace>1) {
cat("rejigged ans:")
print(ans)
}
} # end have answer
ans
} ## end if using snewtonm
## --------------------------------------------
else if (method == "hjn") {# Use JN Hooke and Jeeves
if (control$trace > 0) {
# this function is in optimr, so does not need explicit package
cat("hjn:control$have.bounds =",control$have.bounds,"\n")
cat("optimr - hjn - msk:")
print(msk)
}
# 180327 Cannot maximize with hjn itself.
mcontrol <- control # copy
mcontrol$maximize <- NULL # and null out maximize
ans <- try(hjn(spar, efn, lower=slower, upper=supper, bdmsk=msk,
control=control, ...))
if (! inherits(ans, "try-error")) {
## Need to check these carefully??
ans$par <- ans$par*pscale
ans$value <- ans$value*control$fnscale
ans$message <- NA # Should add a msg ??
} else {
if (control$trace > 0) cat("hjn failed for current problem \n")
ans<-list() # ans not yet defined, so set as list
ans$value <- control$badval
ans$par <- rep(NA,npar)
ans$convergence <- 9999 # failed in run
ans$counts[1] <- NA
ans$counts[1] <- NA
ans$hessian <- NULL
ans$message <- NA
}
## return(ans)
} ## end if using hjn
## --------------------------------------------
else if (method == "spg") { # Use BB package routine spg as minimizer
mcontrol$maximize <- NULL # Use external maximization approach
mcontrol$maxit <- control$maxit
mcontrol$maxfeval <- control$maxfeval
if (control$trace > 0) {
mcontrol$trace <- TRUE
if (control$trace > 1) mcontrol$triter <- 1 # default is 10
} else { mcontrol$trace <- FALSE }
ans <- try(BB::spg(par=spar, fn=efn, gr=egr, lower=slower, upper=supper,
control=mcontrol, ...))
if (! inherits(ans, "try-error")) {
ans$par <- ans$par*pscale
ans$counts[1] <- ans$feval
ans$feval<-NULL # to erase conflicting name
ans$counts[2] <- ans$iter
ans$fn.reduction <- NULL # so it does not interfere
ans$iter<-NULL
ans$gradient<-NULL # loss of information
} else { # spg failed
if (control$trace > 0) cat("spg failed for this problem\n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
ans$hessian <- NULL
}
## return(ans)
} # end if using spg
## --------------------------------------------
else if (method == "ucminf") {
## Use ucminf routine
if (is.null(control$maxit)) { mcontrol$maxeval <- 500 } # ensure there is a default value
else { mcontrol$maxeval <- control$maxit}
mcontrol$maxit <- NULL # 150427 ensure nulled for ucminf
errmsg <- NULL
if (control$have.bounds) {
if (control$trace > 0) cat("ucminf cannot handle bounds\n")
errmsg <- "ucminf cannot handle bounds\n"
stop(errmsg)
ans <- list()
class(ans)[1] <- "try-error"
} else {
uhessian <- 0 # Ensure hessian NOT computed
ans <- try(ucminf::ucminf(par=spar, fn=efn, gr=egr,
hessian = uhessian, control=mcontrol, ...))
}
if (! inherits(ans, "try-error")) {
# From ucminf documentation: convergence = 1 Stopped by small gradient (grtol).
# 2 Stopped by small step (xtol).
# 3 Stopped by function evaluation limit (maxeval).
# 4 Stopped by zero step from line search
# -2 Computation did not start: length(par) = 0.
# -4 Computation did not start: stepmax is too small.
# -5 Computation did not start: grtol or xtol <= 0.
# -6 Computation did not start: maxeval <= 0.
# -7 Computation did not start: given Hessian not pos. definite.
# message: String with reason of termination.
if (ans$convergence == 1
|| ans$convergence == 2
|| ans$convergence == 4) {
ans$convergence <- 0
}
ans$par <- ans$par*pscale
ans$counts[1] <- ans$info[4]
ans$counts[2] <- ans$info[4] # calls fn and gr together
ans$info <- NULL # to erase conflicting name
ans$nitns <- NULL
ans$hessian <- NULL
ans$invhessian.lt <- NULL
if (control$trace > 0) cat("ucminf message:",ans$message,"\n")
} else { # ucminf failed
if (control$trace > 0) cat("ucminf failed for this problem\n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- errmsg
ans$hessian <- NULL
}
uhessian <- NULL
## return(ans)
} ## end if using ucminf
## --------------------------------------------
else if (method == "Rtnmin") { # Use Rtnmin routines
if (control$trace>0) {mcontrol$trace <- TRUE } else {mcontrol$trace <- FALSE}
ans<-list() # ans not yet defined, so set as list
errmsg <- NA
class(ans)[1] <- "undefined" # initial setting
if (is.null(egr)) { ## fixed msg below (referred to lbfgs) 170214
if (control$trace > 0) cat("Rtnmin MUST have gradient provided\n")
errmsg <- "Rtnmin MUST have gradient provided"
class(ans)[1] <- "try-error"
} else {
if (control$have.bounds) {
ans <- try(tnbc(x=spar, fgfun=nlmfn, lower=slower,
upper=supper, trace=mcontrol$trace, ...))
} else {
ans <- try(tn(x=spar, fgfun=nlmfn, trace=mcontrol$trace, ...))
}
}
if (inherits(ans,"try-error")) {
if (control$trace>0) cat("Rtnmin failed for current problem \n")
ans$convergence <- 9999 # failed in run
ans$message <- "Rtnmin failed fo current problem"
if (is.null(egr)) {
ans$convergence <- 9998
ans$message <- errmsg
ans$value <- 1234567E20
}
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA
ans$hessian <- NULL
} else {
ans$par <- ans$xstar*pscale
ans$xstar <- NULL
ans$value <- as.numeric(ans$f)
ans$f <- NULL
ans$g <- NULL
ans$convergence <- ans$ierror
ans$ierror <- NULL
ans$counts[1] <- ans$nfngr
ans$counts[2] <- ans$nfngr
ans$nfngr <- NULL
ans$hessian <- NULL
ans$message <- NA
}
## return(ans)
} ## end if using Rtnmin
## --------------------------------------------
else if (method == "bobyqa") {# Use bobyqa routine from minqa package
mcontrol$maxfun <- control$maxfeval
mcontrol$iprint <- control$trace
myrhobeg <- min(supper - slower)/3 # JN 160107 (3), 160125 (5)
if ((myrhobeg < 1e-8) || ! is.finite(myrhobeg) ) myrhobeg <- 0.5
mcontrol$rhobeg <- myrhobeg # to avoid 0 when parameters 0
ans <- try(minqa::bobyqa(par=spar, fn=efn, lower=slower,
upper=supper, control=mcontrol,...))
if (! inherits(ans, "try-error")) {
ans$convergence <- 0
# if (ans$feval > mcontrol$maxfun) {
# ans$convergence <- 1 # too many evaluations
# }
ans$convergence <- ans$ierr
ans$ierr <- NULL
ans$message <- ans$msg
ans$msg <- NULL
ans$counts[1] <- ans$feval
ans$counts[2] <- NA
ans$feval <- NULL
ans$value<-ans$fval
ans$par <- ans$par*pscale
ans$fval <- NULL # not used
ans$hessian <- NULL
} else {
if (control$trace > 0) cat("bobyqa failed for current problem \n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
ans$hessian <- NULL
}
ans <- unclass(ans) # because minqa does strange things!
## return(ans)
} ## end if using bobyqa
## --------------------------------------------
else if (method == "uobyqa") {# Use uobyqa routine from minqa package
mcontrol$maxfun <- control$maxfeval
mcontrol$iprint <- control$trace
myrhobeg <- min(abs(spar)) # JN 160107 (3), 160125 (5)
if ((myrhobeg < 1e-8) || ! is.finite(myrhobeg) ) myrhobeg <- 0.5
mcontrol$rhobeg <- myrhobeg # to avoid 0 when parameters 0
if (control$have.bounds) {
warning("Cannot use uobyqa with bounds")
if (control$trace > 0) cat("Cannot use uobyqa with bounds\n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
ans$hessian <- NULL
## return(ans)
}
ans <- try(minqa::uobyqa(par=spar, fn=efn, control=mcontrol,...))
if (! inherits(ans, "try-error")) {
ans$convergence <- 0
# if (ans$feval > mcontrol$maxfun) {
# ans$convergence <- 1 # too many evaluations
# }
ans$convergence <- ans$ierr
ans$ierr <- NULL
ans$message <- ans$msg
ans$msg <- NULL
ans$counts[1] <- ans$feval
ans$counts[2] <- NA
ans$feval <- NULL
ans$value<-ans$fval
ans$par <- ans$par*pscale
ans$fval <- NULL # not used
ans$hessian <- NULL
} else {
if (control$trace > 0) cat("uobyqa failed for current problem \n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
ans$hessian <- NULL
}
ans <- unclass(ans) # because minqa does strange things!
## return(ans)
} ## end if using uobyqa
## --------------------------------------------
else if (method == "newuoa") {# Use newuoa routine from minqa package
if (control$trace > 1) cat("Trying newuoa\n")
mcontrol$maxfun <- control$maxfeval
mcontrol$iprint <- control$trace
myrhobeg <- min(abs(spar)) # JN 160107 (3), 160125 (5)
if ((myrhobeg < 1e-8) || ! is.finite(myrhobeg) ) myrhobeg <- 0.5
mcontrol$rhobeg <- myrhobeg # to avoid 0 when parameters 0
if (control$have.bounds) {
warning("Cannot use newuoa with bounds")
if (control$trace > 0) cat("Cannot use newuoa with bounds\n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
ans$hessian <- NULL
## return(ans)
}
ans <- try(minqa::newuoa(par=spar, fn=efn, control=mcontrol,...))
if (! inherits(ans, "try-error")) {
ans$convergence <- 0
# if (ans$feval > mcontrol$maxfun) {
# ans$convergence <- 1 # too many evaluations
# }
ans$convergence <- ans$ierr
ans$ierr <- NULL
ans$message <- ans$msg
ans$msg <- NULL
ans$counts[1] <- ans$feval
ans$feval <- NULL
ans$counts[2] <- NA
ans$value<-ans$fval
ans$par <- ans$par*pscale
ans$fval <- NULL # not used
ans$hessian <- NULL
} else {
if (control$trace > 0) cat("bobyqa failed for current problem \n")
ans<-list() # ans not yet defined, so set as list
ans$convergence <- 9999 # failed in run
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$counts[1] <- NA # save function and gradient count information
ans$counts[2] <- NA # save function and gradient count information
ans$message <- NULL
ans$hessian <- NULL
}
ans <- unclass(ans) # because minqa does strange things!
## return(ans)
} ## end if using newuoa
## --------------------------------------------
else if (method == "nmkb") {# Use nmkb routine from dfoptim package
if (any(par == lower) || any(par==upper)) {
if (control$trace>0) cat("nmkb cannot start if on any bound \n")
warning("nmkb() cannot be started if any parameter on a bound")
ans <- list() # ans not yet defined, so set as list
ans$value <- control$badval
ans$par <- rep(NA,npar)
ans$convergence <- 9999 # failed in run - ?? consider special code for nmkb on bounds
ans$fevals <- NA
ans$gevals <- NA
ans$nitns <- NA
ans$hessian <- NULL
} else { # ok to proceed with nmkb()
if (! is.null(control$maxit)) {
mcontrol$maxfeval <- control$maxit
} else {
mcontrol$maxfeval <- 5000*round(sqrt(npar+1)) # ?? default at 100215, but should it be changed?
}
if (control$trace > 0) { mcontrol$trace <- TRUE } # logical needed, not integer
else { mcontrol$trace<-FALSE }
if (control$have.bounds) {
ans <- try(dfoptim::nmkb(par=spar, fn=efn, lower = slower,
upper = supper, control=mcontrol, ...))
} else {# 170919 explicit package in call
ans <- try(dfoptim::nmk(par=spar, fn=efn, control=mcontrol, ...))
}
if (control$trace > 1) {
cat("Outputting ans for nmkb:\n")
print(ans)
}
if (! inherits(ans, "try-error")) {
ans$value <- as.numeric(ans$value)
ans$par <- ans$par*pscale
ans$counts[1] <- ans$feval
ans$feval <- NULL
ans$counts[2] <- NA
ans$nitns <- NA # not used
# What about 'restarts' and 'message'??
warning(ans$message," Restarts for stagnation =",ans$restarts)
ans$restarts <- NULL
ans$hessian <- NULL
} else {
if (control$trace>0) cat("nmkb failed for current problem \n")
ans <- list(fevals=NA) # ans not yet defined, so set as list
ans$value <- control$badval
ans$par <- rep(NA,npar)
ans$counts[1] <- NA
ans$counts[2] <- NA
ans$convergence <- 9999 # failed in run
ans$message<-"Failed"
ans$hessian <- NULL
}
} # end of check for parameter on bound
## return(ans)
} ## end if using nmkb
## --------------------------------------------
else if (method == "hjkb") {# Use hjkb routine from dfoptim package
if (control$trace > 0) {
mcontrol$info <- TRUE # logical needed, not integer
} else { mcontrol$info <- FALSE }
mcontrol$maxfeval <- control$maxfeval
if (control$have.bounds) {
ans <- try(dfoptim::hjkb(par=spar, fn=efn, lower = slower,
upper = supper, control=mcontrol, ...))
} else {
ans <- try(dfoptim::hjk(par=spar, fn=efn, control=mcontrol, ...))
}
if (! inherits(ans, "try-error")) {
ans$value <- as.numeric(ans$value)
ans$par <- ans$par*pscale
ans$counts[1] <- ans$feval
ans$feval <- NULL
ans$counts[2] <- NA
ans$nitns <- NULL # not used
ans$restarts <- NULL
ans$hessian <- NULL
ans$nitns <- NULL # loss of information
} else {
if (control$trace>0) cat("hjkb failed for current problem \n")
ans <- list(value=control$badval, par=rep(NA,npar), message="Failed",
convergence=9999)
ans$counts[1]<- NA
ans$counts[2]<- NA
ans$hessian <- NULL
}
## return(ans)
} ## end if using hjkb
## --------------------------------------------
else if (method == "lbfgsb3c") {# Use 2011 L-BFGS-B wrapper
if (control$trace > 1) cat("lbfgsb3c\n")
mcontrol$trace <- control$trace
# 170924 no longer needed
## if (control$trace < 1) {mcontrol$iprint <- -1} else {mcontrol$iprint <- control$trace}
if (control$trace > 0) cat("lbfgsb3c:control$have.bounds =",control$have.bounds,"\n")
if (control$have.bounds) {
slower <- lower/pscale
supper <- upper/pscale
ans <- try(lbfgsb3c::lbfgsb3c(par=spar, fn=efn, gr=egr, lower = slower,
upper = supper, control=mcontrol, ...)) # explicit pkg in call 170919
} else {
ans <- try(lbfgsb3c::lbfgsb3c(par=spar, fn=efn, gr=egr, control=mcontrol, ...))
}
if (! inherits(ans, "try-error")) {
## Need to check these carefully?? -- changed 20191202 for lbfgsb3c
# ans$convergence <- 0
ans$par <- ans$par*pscale
# ans$prm <- NULL
# ans$value<-as.numeric(ans$f)
# ans$f <- NULL
# ans$counts[1] <- ans$info$isave[34]
# ans$counts[2] <- ans$counts[1]
# ans$info <- NULL ## Note -- throwing away a lot of information
# ans$g <- NULL ## perhaps keep -- but how??
# ans$hessian <- NULL
# ans$message <- NA
ans$niter <- NULL # loss of information
} else {
if (control$trace>0) cat("lbfgsb3c failed for current problem \n")
ans<-list(fevals=NA) # ans not yet defined, so set as list
ans$value <- control$badval
ans$par<-rep(NA,npar)
ans$convergence<-9999 # failed in run
ans$counts[1] <- NA
ans$counts[1] <- NA
# ans$hessian <- NULL
# ans$message <- NA
}
## return(ans)
} ## end if using lbfgsb3c
## --------------------------------------------
else if (method == "lbfgs") {# Use unconstrained method from lbfgs package
if (control$trace > 1) cat("lbfgs\n")
if (control$trace < 1) {invisible <- 1} else {invisible <- 0}
if (control$trace > 1) cat("lbfgs:control$have.bounds =",control$have.bounds,"\n")
ans <- list() # to define the answer object
errmsg <- NA
class(ans)[1] <- "undefined" # initial setting
## cat("in lbfgs section, control$have.bounds=",control$have.bounds,"\n")
if (control$have.bounds) {
cat("control$have.bounds seems TRUE\n")
if (control$trace > 0) cat("lbfgs::lbfgs cannot handle bounds\n")
errmsg <- "lbfgs::lbfgs cannot handle bounds\n"
## stop("lbfgs::lbfgs tried with bounds")
class(ans)[1] <- "try-error"
}
if (is.null(egr)) {
if (control$trace > 0) cat("lbfgs::lbfgs MUST have gradient provided\n")
errmsg <- "lbfgs::lbfgs MUST have gradient provided\n"
class(ans)[1] <- "try-error"
}
if (inherits(ans, "undefined")){
dotstuff <- list(...)
# cat("dotstuff:\n")
# print(dotstuff)
dotstuff$pscale <- pscale
dotstuff$fnscale <- control$fnscale
eopt <- list2env(dotstuff) # put it in an environment
# print(ls(eopt))
ans <- try(lbfgs::lbfgs(efn, egr, vars=spar,
environment=eopt, invisible=invisible))
}
# cat("interim answer:")
# print(ans)
if (! inherits(ans, "try-error")) {
## Need to check these carefully??
ans$par <- ans$par*pscale
ans$value <- ans$value*control$fnscale
ans$counts[1] <- NA # lbfgs seems to have no output like this
ans$counts[2] <- NA
} else {
if (control$trace>0) cat("lbfgs failed for current problem \n")
ans<-list() # ans not yet defined, so set as list
ans$value <- control$badval
ans$par <- rep(NA,npar)
ans$convergence <- 9999 # failed in run
if (is.null(egr)) ans$convergence <- 9998 # no gradient
ans$counts[1] <- NA
ans$counts[1] <- NA
ans$hessian <- NULL
if (! is.na(errmsg)) ans$message <- errmsg
}
## return(ans)
} ## end if using lbfgs
## --------------------------------------------
else if (method == "subplex") {# Use unconstrained method from subplex package
if (control$trace > 1) cat("subplex\n")
if (control$trace < 1) {invisible <- 1} else {invisible <- 0}
if (control$trace > 1) cat("subplex:control$have.bounds =",control$have.bounds,"\n")
ans <- list() # to define the answer object
if (control$trace > 0) warning("subplex has no trace mechanism")
class(ans)[1] <- "undefined" # initial setting
if (control$have.bounds) {
cat("control$have.bounds seems TRUE\n")
if (control$trace > 0) cat("subplex::subplex cannot handle bounds\n")
stop("subplex::subplex cannot handle bounds")
}
if (class(ans)[1] == "undefined"){
ans <- try(subplex::subplex(par=spar, fn=efn, control=list(maxit=control$maxfeval)))
}
if (!inherits(ans, "try-error") && (ans$convergence != -2)) {
## Need to check these carefully??
ans$par <- ans$par*pscale
ans$value <- ans$value*control$fnscale
ans$counts[1] <- ans$count
ans$counts[2] <- NA
ans$count <- NULL
ccode <- ans$convergence
ans$convergence <- 9999
ans$message <- paste("subplex:",ans$message)
if ((ccode == 0) || (ccode == 1)) {
ans$convergence <- 0
if (ccode == 0) { ans$message <- "subplex: success" }
else { ans$message <- "subplex: Limit of precision reached" }
} # converged OK
else {if (ccode == -1) {
ans$convergence <- 1
ans$message <- "subplex: function evaluation limit reached"
} # effort limit
}
} else {
if (ccode == -2) {
ans$convergence <- 20
}
else { ans$convergence <- 9999}
ans$value <- control$badval
ans$par <- rep(NA,npar)
ans$counts[1] <- NA
ans$counts[2] <- NA
ans$hessian <- NULL
}
} ## end if using subplex
## --------------------------------------------
## END OF optimrx extra methods
# --- UNDEFINED METHOD ---
else { errmsg<-paste("UNDEFINED METHOD:", method, sep='')
stop(errmsg, call.=FALSE)
}
# Exit from routine
ans$value <- ans$value * control$fnscale # reset for maximum
if (savehess) { # compute hessian
if (is.null(orig.gr)) {
hess <- hessian(orig.fn, ans$par, ...) # from numDeriv
} else {
hess <- jacobian(orig.gr, ans$par, ...) # use Jacobian of gradient
}
} else { hess <- NULL } # to ensure it is defined
ans$hessian <- hess
ans # last statement of routine
} ## end of optimrx
|
