| blob | 25833969e8efc68cbd351123f258267a8c97be37 |
1 This version of PsN is tested with perl-5.8.8
2 Depending on your distribution you may have to install the following
3 perl packages:
5 Math::Random
6 Storable::Store
8 All available from CPAN ( www.cpan.org )
10 Changelog
11 ----------------------------------------------
13 Changes from 2.2.4 to 2.2.5
15 Bugs
17 - Memory leak in output parser plugged. Output files from simulations
18 could use vast amounts of memory.
20 - Bootstrap options "skip_minimization_terminated" can now be disabled
21 by typing "-no-skip-minimization_terminated" (or -noskip-...)
23 - cdd xv can no be disabled with -no-xv or -noxv
25 - mc tool no uses considerably less memory. It is a lot slower
26 because it reads a lot of unecessary stuff from disk, it should be
27 reviewed.
29 - Memory leak in modelfit due to intermediate "raw_results.csv" in
30 NM_runX directory fixed by removing the feature, should be removed.
32 - A initial estimate of form: $OMEGA BLOCK SAME would not be
33 recognized by PsN model file parser. This is now handled
34 correctly. In 2.2.X it can be avoided by typing BLOCK(X) SAME,
35 where X is the size of the block.
37 Changes from 2.2.3 to 2.2.4
39 Features
41 - "--compute_cwres" option renamed to simply "--cwres"
43 - Simulated data suitable for creating mirror plots can be generated
44 by using a new options called --mirror_plots. The option takes a
45 value which defines the number of simulations that will be
46 geneated.
48 - A new option, --iofv, enables the computation of individual
49 objective function values. These values are printed to a table file
50 called iotab*, where * is the same number as is defined for any
51 sdtab or patab tables. These names correspond to the normal Xpose
52 table file format. If no sdtab or patab table is found in the
53 NONMEM control stream, the iotab file is given number 1.
55 - msfo resumes. Basic support for automatically rerunning a model
56 that has the MSFO=msffile option on $ESTIMATION. If the option is
57 set, the "msffile" exists and the option --msfo_resume is given,
58 PsN will remove initial estimates, and add the $MSFI record. This
59 is very untested and if you have OMEGA blocks, the model will run
60 but PsN will be unable to read the output file properly.
62 - "no_remote_compile" and "no_remote_execution" options has been
63 removed, as they were quite seldomly used and added a lot of
64 complexity to parallization code.
66 - Added --prepend_model_file_name option which adds the model file
67 name to output files to prevent them from being overwritten.
69 - A new option called "crash_restarts" which controls the number of
70 times PsN will restart a crashed run. (Simply put, run is
71 classified as crashed if the output file looks like it is cut off)
73 - PsN will now automatically run R-scripts generated for the
74 bootstrap, cdd and llp if R is installed and configured.
76 - When the CONT data item is used - either defined by the user or
77 automatically through the -wrap_data option - the tables get a
78 weird format where the real table content is printed on CONT=0 rows
79 and nonsense data printed on CONT=1 rows. A new option called
80 -unwrap_table_files reformats the table files and removes the CONT
81 column as well as all CONT=1 rows.
83 - update_inits script that will put final estimates from a NONMEM
84 output file and put into a given model file. This is quite
85 experimental. PsN will reformat the model file quite lot and
86 comments in the file might get lost. Your model file will be copied
87 with the addition of a ".org" extension, so you wont loose your
88 model.
90 Bug Fixes
92 - SGE queue and resource options were ignored. Fixed thanks to Jeroen
93 Elassaiss and Stefan Verhoeven.
95 - LSF Monitor fixed to work with the new serialized structure of PsN,
96 again thanks to Jeroen and Stefan.
98 - CWRES table file number now defaults to 1 if there is no patab or
99 sdtab to take the number from.
101 - Fixed run bugs under Windows 2000.
103 - Better support for G77 under windows.
105 - Parser for output files improved
107 Changes from 2.2.2 to 2.2.3
109 Features
111 - There has been a lot of requests for a raw_results file for the
112 runs in the NM_run catalogs, i.e. a summary of the runs with
113 perturbed initial estimates. This is now implemented. One
114 raw_results file is now created in each NM_run* catalog.
116 - The R-scripts for graphical visualization of bootstrap and cdds
117 have been updated. The format for the graphs is now pdf instead of
118 postscript. The bootstrap script can now be adjusted to exclude
119 bootstrap samples that contain copies of a given set of
120 individuals.
122 - When the wrap_data option is used, the secondary columns, i.e. the
123 data columns that are kept on rows with CONT=1, are now available
124 in $ERROR.
126 Bug Fixes
128 - The calculation of CWRES was previously done on the _last_ run of a
129 set of retries, not on the _best_ run. This has now been corrected.
131 - The restarting of crashed runs was broken. This is now fixed.
133 - The asynchronuous job scheduling broke the handling of seed numbers
134 for random sequences (e.g. perturbation of initial estimates). This
135 is now fixed. The present version is however not compatible with
136 older versions and old runs should therefore not be resumed using
137 this version.
139 - wrap_data should now work as it should. The functionality was
140 broken as a consequence of the move to the asynchronuous job
141 handling.
143 - Labels for omegas were not handled correctly for SAME blocks. Among
144 other things, this had the effect that the updating of initial
145 estimates from the original model to the method created models
146 (e.g. bootstrap samples) was corrupted. This is now fixed.
148 - The transformation of the estimate of the omega matrix from
149 variance and covariances to standard deviations and correlations
150 was rewritten to address problems with a combination of BLOCK(X)
151 ans SAME where X was larger than 1.
153 - The selection of a 'best' model fit after a series of perturbations
154 of the initial estimates will be changed. In 2.2.2, the lowest OFV
155 among fits of the perturbed models was collected across all models
156 (still only comparing identical models with perturbed initial
157 estimates). In 2.2.3 the lowest OFV is collected in three
158 categories of model fits: those that minimized succesfully, those
159 that did not minimize successfully but returned an OFV plus an
160 estimate of the number of significant digits and finally failed
161 runs that managed to return an OFV but not much more. This will
162 give PsN a more fine-grained control over the results. For example,
163 it would be desirable to be able to return the run with the lowest
164 OFV within the selection of runs with a successful minimization
165 even though the lowest OFV of all perturbed runs was returned from
166 a run that terminated.
168 - The sigmas were not parsed correctly when tables were printed to
169 the output file. This is not fixed.
171 - Terminated covariance steps were sometimes still being parsed (and
172 failed). Fixed.
174 - Setting the clean level to 3 did not remove the
175 NM_run-directories. This is now fixed.
177 - The parsing of options for the initial values of omegas and sigmas
178 sometimes wrongly recognized letters behind a semi-colon (i.e. a
179 comment) as keywords for a block (FIX, SAME, etc). This is now
180 corrected.
182 - If you "fix" more than one omega or sigma in a block PsN would put
183 a FIX after each value, which NONMEM doesn't allow. Now PsN will
184 only put one FIX inside a block. However, if a any part of the
185 block has been fixed, its likely so that PsN can not(!) unfix
186 it.(so, one bug fixed, another introduced)
188 - If you an initial estimate was added with "_init_attr", the option
189 "add_if_absent" was true and the last initial estimate was "SAME"
190 PsN would try to add the estimate after "SAME" which is wrong, now
191 PsN adds a new record.
193 - Fixed a parsing error of the lst file. If simulation without a
194 SIGMA was performed, the parser failed to read OMEGA initial
195 estiamtes.
197 - The internal method "indexes" would not properly handle "SAME" in
198 initial estimates. The result was that "update_inits" would skip
199 initial values, and get them out of order. Fixed.
202 Changes from 2.2.2-beta4 to 2.2.2
204 Features
206 - CWRES can now be calculated automatically if you have R and Xpose
207 installed. Remember to add a definition of IPRED in your NONMEM model.
209 - This version handles non-PsN standard (or absent) suffixes for
210 model files better than earlier versions.
212 Bug Fixes
214 - MSFO files are now copied back to the same directory of the model
215 file.
217 - MSFI files are now automatically copied to the PsN NONMEM run
218 directories.
220 - Output files from runs where the $MSFI record was used were not
221 parsed correctly. This is now fixed.
223 - Output files from runs where no $OMEGA _or_ $SIGMA were defined
224 were not parsed correctly.
226 - The header and the results from problematic runs, i.e. failed
227 covariance step, warnings etc, were not handled correctly when the
228 raw_results files were compiled. This is now fixed.
230 - In 2.2.1, we added a check for a '1' to find the end of the omega
231 and sigma estimate areas. this was a mistake since problems with
232 more than 13 omegas or sigmas get a '1'-line inserted before the
233 14:th estimate (for some reason). This is now removed.
235 Changes from 2.2.2-beta3 to 2.2.2-beta4
237 Features
239 - No new features
241 Bug Fixes
243 - Adaptive number of threads is now working in the new asynchronuous
244 job handling
246 - The CWRES computation had some fixes
248 - $INFN is now printed before $MODEL
250 - $OMEGA is now printed befor $PK
252 - PRIOR is now handled in the output module. It is still a bit
253 problematic to parse PRIOR output files correctly. Use with
254 causion.
256 - The initial estimates of large omega and sigma matrices is now
257 better handled
259 - The order of parsing events in the output module is now working
260 better.
262 - The handling of retries (or the results of retries) is now working
263 as it should.
265 Changes from 2.2.2-beta2 to 2.2.2-beta3
267 Features
269 - CWRES: Conditional weighted residuals. The necessary verbatim code
270 for the computation of CWRES are now automatically added if the
271 option -compute_cwres is used.
273 - PRIOR: The use of priors in NONMEM VI is now supported in the
274 computer intensive methods
276 - The behavior of the parallel execution has been revised, moving
277 from a synchronuous to an asynchronuous solution. Previously, the
278 memory foot-print of a parallel execution of say 100 NONMEM jobs
279 was huge (counted in several GB) and did not scale well. Now, it
280 will very seldom exceed 100 MB and more or less keep that size,
281 regardless of the number of parallel jobs.
283 Bug Fixes
285 - PsN couldn't parse IGNORE=(ITEM=VALUE) options in $DATA. This is
286 (again) fixed. It was previously corrected in the development
287 branch of PsN but never transferred to the stable branch.
289 - Fixed SAME block correlation computation.
291 Changes from 2.2.2-beta1 to 2.2.2-beta2
293 Features
295 - The parsing of the NONMEM output files has been revised and a
296 strict check is now performed after each section has been parsed. A
297 parsing error is raised if any non-compliant structures are
298 found. By doing this we get a better idea of which runs that really
299 finishes and which runs that are pre-maturely terminated
300 (e.g. through a operating system error). Remember that this is a
301 beta release and that there could still be valid text structures
302 that are not yet recognised correctly by PsN.
304 - Support for United Devices grip MP has been added and is available using
305 "-run_on_ud" command line option.
307 Bug Fixes
309 - the options directory and backwards_dir can now be specified in the
310 scm configuration file
312 - It is now possible to skip the definition of either continuous and
313 categorical covariates in the scm configuration file
315 - The exponential relation is now added correctly in the
316 scm. Previously it was incorrectly added as (for weight on volume)
317 TVV = THETA(1)*(1+EXP(THETA(2)*(WGT-median(WGT))))
318 Now it is added as
319 TVV = THETA(1)*EXP(THETA(2)*(WGT-median(WGT)))
321 - The output parse now computes correlations for omegas and sigmas correctly
322 when you have SAME blocks.
324 Changes from 2.2.1 to 2.2.2-beta1
326 Features
328 - The number of significant digits per parameter after a terminated
329 minimization are now parsed.
331 Bug Fixes
333 - The sample size option of the bootstrap did not work correctly with
334 stratified resampling. This is now fixed.
337 Changes from 2.2.0 to 2.2.1
339 Features
341 - Added support for the Intel Fortran compiler (mostly for those
342 using the new Intel Macs).
344 - We have added two new model fit quality terms: 'terminated' and
345 'really_bad' to the previous 'normal' and 'picky'.
347 picky - Minimization and covariance steps successful, no
348 parameters near boundaries and no warnings from the
349 covariance step.
351 normal - Minimization successful
353 terminated - Estimates of the objective function value and the
354 number of significant digits are available.
356 really_bad - An estimate of the OFV is available (but usually not
357 much more...).
359 Given that more than one run from a run has been performed (with
360 different initial estimates but otherwise identical) the following
361 selection procedure is done:
363 The lowest OFV of all tries is collected. Then, all runs within
364 5 units of the lowest OFV is tried according to
366 1. Of the runs passing the picky criteria, choose the run with the
367 highest number of significant digits.
369 2. If no run has passed the criteria above: of the runs passing the
370 normal criteria, choose the one with the highest number of
371 sginificant digits.
373 3. If no run has passed the criteria above: of the runs passing the
374 terminated criteria, choose the one with the highest number of
375 sginificant digits.
377 4. If still no run has passed the criteria above: of all runs that
378 at least have an estimate of the OFV, choose the run with the
379 lowest OFV.
381 5. As a last resort, choose the first run.
383 Comments: This is of course a very pragmatic approach but it is
384 needed to make the automatic procedures of the PsN tools to
385 work. We cannot force the user to manually scrutinise the results
386 of every single run of for example a bootstrap. The user has the
387 opportunity to do so if he or she wishes. If you do not feel
388 comfortable with this selection procedure or if you have comments
389 on how to improve it, please contact me (Lars Lindbom) at
390 lars.lindbom (at) farmbio.uu.se.
392 Bugfixes
394 - Data files with spaces around commas used for columns separation
395 could be problematic to parse. This is now fixed.
397 - There was a serious error in the seed handling in version
398 2.2.0. Under the following circumstances, the seed number was
399 ignored and a default seed number used: If the output file from the
400 original run was not present in the same directory as the original
401 model file when a PsN tool was started. This affected all PsN
402 tools, except the execute tool. This is now fixed.
404 - Up to 20 omegas can now be parsed in the output file. This is still
405 a limitation and it will be addressed in later releases.
407 - On fast distributed systems with a shared file system, the massive
408 parallelisation of the PsN tools could cause synchronization
409 problems (files could sometimes not be found, read or written). The
410 root of this problem is somewhat outside of the scope of PsN but we
411 have tried to find a workaround for this for the LSF grid system.
413 Changes from 2.1.10 to 2.2.0 ( released for PsN course )
415 Features
417 - PsN Output module use a proper Matrix library for handling some of
418 the matrices of the outputfile.
420 - The bootstrap can now be resumed with a lower number of
421 samples. Meaning that if you are in a hurry and want to stop the
422 bootstrap and get results after 1000 out of 2000 runs, you can. At
423 a later time you can then resume the 2000 runs.
425 - Nonmem warnings are now printed if "--verbose" is givven.
427 - Automatic incrementation of directory names are a bit smarter (no
428 more "directory modelfit_dirX is in the way" messages).
430 - Clean is much more stringent and intelligent and now manages to
431 remove entire tool directories if desired.
433 - option list printed from "-h|-?" only shows utility specific options.
435 - Default values configurable on a per utility basis.
437 - Online documentation improved
439 - Parallel run of NONMEM is not aborted as before if a outputfile is
440 not correctly parsed. It is just ignored, as if it had not
441 minimized successfully. (actually it is the output parser that
442 doesn't die, but the effect is the same)
444 - Parallel run is a bit nicer to the system when compiling many
445 NONMEMS. By adding a configurable delay between starts.
447 - Matlab and R scripts for various plots from cdd and bootstrap
448 results are distributed with PsN.
450 - Vast improvments in results presentations from all tools.
452 - Improvements in rerun algoritm with new "min_retries" option.
454 - Output from running utilties are much nicer.
456 - new "-quick_summary" prints minimization message and ofv value for
457 each finished run. As does "-verbose" option which also prints
458 "R:X" for each retrie (X is the retrie number).
460 - new "-summary" options prints diagnostics about the tool run when
461 it is finished. For "execute" the "-summary" option will have the
462 same function as the "sumo" utiltiy as well as some diagnostics.
464 - PsN now recognizes that control files may require fortran files and
465 copies them to the run directory. No need to add fotran files to
466 the --extra_files option any more.
468 - Added --extra_output which allows you to copy more than the output
469 file from the run directory.
471 Bugs
473 - PsN couldn't parse IGNORE=(ITEM=VALUE) options in $DATA. Fixed :)
475 - If star notation was used in the scm config file section [inits],
476 [upper_bounds] or [lower_bounds] there was a risk of defining
477 relations that was not specified in [test_relations]. Which in turn
478 would cause a lot of diffrent errors. Fixed :)
480 - Using low_cutoff and hi_cutoff in data::sd resulted in PsN trying to use
481 the new compact data strings as the old arrays. fixed.
483 - Removed unecessary regular expression in output file parser.
485 - Not really a bug. But i tried to add "FIX" as a record option to a theta
486 records. Then when "store_init" was called on the "FIX" option PsN crashed
487 because the option was not a "theta_option". So i added a check for
488 "store_init" abilties... if FIX would become a record option in the future.
490 - Member accessor "datafile" replaced with "datafiles" in model. Affected
491 only the bootstrap which have been updated.
493 - An assumptions about outputfiles crashed sumo occationally. Some more
494 rigorous contentchecking has been added and fixed the problem.
496 Internal changes
497 - Cleanup of dubious utility options.
498 - Option help sorted alphabetically.
499 - "flush", "target" removed from model.
500 - "remove_temp_files" removed from modelfit.
501 - Changed the Makefile considerably, much simpler now.
502 - added "column_to_array" function in data module.
504 Changes from 2.1.9 to 2.1.10
506 Features
507 - The 'picky' option of all utilities now triggers on
508 COVARIANCE STEP ABORTED and
509 PROGRAM TERMINATED BY OBJ
511 Changes from 2.1.8 to 2.1.9
513 Features
514 - Nordugrid support reintroduced (but will probably not be developed further)
515 - Sumo now supports multiple problems and subproblems.
516 - Sumo has new option "precision" which control the number of digits
517 to present estimates with.
518 - An extra 'short' logfile is output from the scm with the most important
519 information.
521 Bugs
522 - Bootstrap option "stratify_on" doesn't work with columns defined in
523 extra data files given with "extra_data" option. Not Fixed.
524 - Resumes of scm's with extra_data_files is now possible. Still some
525 problems with scm's where search_direction=both.
526 - LSF options was not communicated properly between tools and subtools
527 so submission of bootstraps, llp, cdd and scm did not work. Fixed.
528 - Default values for LSF can now be configured in psn.conf.
529 - Print order of modelfiles places $MIX before $PRED. Fixed.
530 - extra_files option was broken. Fixed.
532 Internal Changes
533 - Tool will always fork subtools, this will save memory in the scm (and
534 other recursive tools).
535 - Common options need only be specified once. In one of three categories:
536 model_options, tool_options, script_options.
537 - models can now be created with "reference" objects. This solved the
538 scm resume problem.
539 - Added function to model that cleans extra data code.
541 Changes from 2.1.7 to 2.1.8
543 Features
544 - Rudimentary support for execution on Platform LSF grid systems is now
545 included. The following extra attributes are available to all tools
546 for controlling the execution:
548 run_on_lsf - Use LSF
549 lsf_queue - Use this queue on the grid
550 lsf_options - General options for the grid queuing system.
551 lsf_project_name - Optional. May be used by the grid system
552 administrator to register control project
553 cpu usage.
554 lsf_job_name - Optional. May be used by a grid system
555 administrator to control multiple runs on a
556 grid system.
558 - All tools now have a raw_results file holding the diagnotics and
559 parameter estimates for each NONMEM run.
560 - All tools now handles model files containing one $PROBLEM and no
561 SUBPROBLEMS only.
562 - The Bootstrap and CDD routines now create separate files holding
563 information about which individuals that were included/excluded
564 in/from each created data set.
565 - The Bootstrap results file now has warnings section to better
566 display which runs that were used for result calculations and which
567 were not
568 - PsN now tries to read psn.conf from the users home directory. (Will
569 require some extra documentation).
570 - Optimized modelfit thread operation by removing Storable::Store.
571 - The clean options has four levels of operation.
573 0 - No cleaning at all.
574 1 - Clean more, but make sure resumes work (this is default).
575 2 - Clean more, but make sure summarize work.
576 3 - Clean everything, keeping only results and logs.
578 - Some attributes and accessors of the PsN-Core class 'output' have
579 had their names changed to make it more clear what their purposes
580 are:
581 termination is now called minimization_successful
582 termination_message is now called minimization_message
583 covstep_termination is now called covariance_step_successful
585 these attributes are new:
587 covariance_step_warnings (boolean; 0 or 1)
588 estimate_near_boundary (boolean; 0 or 1)
590 The attribute which was previously called termination could have the
591 values 0, 0.5 and 1 for 'minimization terminates', 'minimization
592 successful but warnings or errors were printed from the covariance
593 step' and finally 'minimization successful'. The new attribute
594 'minimization_successful' can only take the values 0 and 1 and the
595 new attribute covariance_step_warnings now takes the value 1 when
596 covariance step warnings were found. Checks are also perfomed to
597 test whether the estimates are close to any boundary.
598 - Html documents is now generated from command line help text.
599 which ensures that they are consistent and up to date.
600 - Personal configuration file now available under Unix like OS:s
602 Bugs
603 - Options specific to modelfiles were ignored in some Utilties.
604 - Gradients now output in windows. Just like with nmfe.
605 - Bootstrap now consumes considerably less memory.
606 - Removed unused/obsolete "resume" function from modelfit.
607 - Sorted method entries in modelfit diagram.
608 - Fixed bugs in parsing of OFV and eigenvalues.
610 Changes from 2.1.6 to 2.1.7
612 Bugs
613 - "clean" and "compress" options now work under Windows.
614 - SCM now handles crashed runs appropriately.
615 - The ofv_forward, ofv_backward, p_forward and p_backward options of
616 scm now works as they should.
618 Features
619 - SCM improvements. The final model is now available in the
620 first-level scm directory of each step.
621 - The Utilties -help option is improved. You can now give a list of
622 options to -help for specific documentation of those options, like
623 so:
625 $ scm -help do_not_drop
627 - Minor changes in how the [nm_version] section in psn.conf i
628 handled. Look in the psn.conf file that is distributed with PsN for
629 more information.
630 - SDE modelfiles are now supported. (see the -sde_records option)
631 - The update_inits method of the model class now accepts an output
632 file name as input.
633 - Some attributes and accessors of the PsN-Core class 'output' have
634 had their names changed to make it more clear what their purposes
635 are:
636 finalgrad is now called final_gradient
637 parampath is now called parameter_path
638 gradientpath is now called gradient_path
639 covterm is now called covstep_termination
640 - The Utilties now saves the command line that started it in a file
641 "command.txt" under the corresponding directory.
642 - Added a "silent" option to the Utilties that quenches all output.
643 - "nice" option added to utilties that allows you to set priority in
644 unix.
645 - Rudimentary support for NorduGrid reinstated.
646 - Installer script rewritten. It now supports Perl installations in
647 nonstandard directories. Discrepancies between *nix and MSWin32 is
648 smaller.
650 Internal changes
651 - The code that is autogenerated with "dia2code" is optimized a
652 bit and saves a hash "__valid_parameters" in each object, which
653 allows the constructors to be used as a generic (shallow) copy.
655 Misc
656 - continuous spelled correctly throughout SCM.
659 Changes from 2.1.5 to 2.1.6
661 Bugs
662 - Stratify_on bug in bootstrap method fixed.
663 - Data files in directories different from model file are now
664 possible.
665 - Path name handling reimplemented. Relative paths are now handled
666 better.
667 - Models that generate table files would make PsN complain about
668 missing files.
669 - Its now possible to specify multiple output names as a
670 comma-separated list to execute.
671 - The scm can now go backwards.
672 - Handling of table files now responds to the
673 "ignore_missing_output_files" flag.
674 - nmtran error messages was lost in Windows 2K. Execution of nmtran
675 is more plattform independent now. It uses a temporary file, but
676 it is never bigger than the nmtran messages.
677 - Fixed a bug that limited choice of compiler under unix.
678 - Typo in scm utility fixed. (included_relations in backwards step).
680 Features
681 - Added support for "$NONPARAMETRIC" record in model files.
682 - A few improved error messages related to parsing of model files.
683 - The SCM config file now features sections for parameter bounds and
684 user-defined code
686 Internal changes
687 - psn.conf settings are now available as global variables. psn.conf
688 is only parsed once.