| blob | a52b5959ff87268f3062d22489d253f91c259301 |
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 2.2.2-beta4 to 2.2.2
15 Features
17 - CWRES can now be calculated automatically if you have R and Xpose
18 installed. Remember to add a definition of IPRED in your NONMEM model.
20 - This version handles non-PsN standard (or absent) suffixes for
21 model files better than earlier versions.
23 Bug Fixes
25 - MSFO files are now copied back to the same directory of the model
26 file.
28 - MSFI files are now automatically copied to the PsN NONMEM run
29 directories.
31 - Output files from runs where the $MSFI record was used were not
32 parsed correctly. This is now fixed.
34 - Output files from runs where no $OMEGA _or_ $SIGMA were defined
35 were not parsed correctly.
37 - The header and the results from problematic runs, i.e. failed
38 covariance step, warnings etc, were not handled correctly when the
39 raw_results files were compiled. This is now fixed.
41 - In 2.2.1, we added a check for a '1' to find the end of the omega
42 and sigma estimate areas. this was a mistake since problems with
43 more than 13 omegas or sigmas get a '1'-line inserted before the
44 14:th estimate (for some reason). This is now removed.
46 Changes from 2.2.2-beta3 to 2.2.2-beta4
48 Features
50 - No new features
52 Bug Fixes
54 - Adaptive number of threads is now working in the new asynchronuous
55 job handling
57 - The CWRES computation had some fixes
59 - $INFN is now printed before $MODEL
61 - $OMEGA is now printed befor $PK
63 - PRIOR is now handled in the output module. It is still a bit
64 problematic to parse PRIOR output files correctly. Use with
65 causion.
67 - The initial estimates of large omega and sigma matrices is now
68 better handled
70 - The order of parsing events in the output module is now working
71 better.
73 - The handling of retries (or the results of retries) is now working
74 as it should.
76 Changes from 2.2.2-beta2 to 2.2.2-beta3
78 Features
80 - CWRES: Conditional weighted residuals. The necessary verbatim code
81 for the computation of CWRES are now automatically added if the
82 option -compute_cwres is used.
84 - PRIOR: The use of priors in NONMEM VI is now supported in the
85 computer intensive methods
87 - The behavior of the parallel execution has been revised, moving
88 from a synchronuous to an asynchronuous solution. Previously, the
89 memory foot-print of a parallel execution of say 100 NONMEM jobs
90 was huge (counted in several GB) and did not scale well. Now, it
91 will very seldom exceed 100 MB and more or less keep that size,
92 regardless of the number of parallel jobs.
94 Bug Fixes
96 - PsN couldn't parse IGNORE=(ITEM=VALUE) options in $DATA. This is
97 (again) fixed. It was previously corrected in the development
98 branch of PsN but never transferred to the stable branch.
100 - Fixed SAME block correlation computation.
102 Changes from 2.2.2-beta1 to 2.2.2-beta2
104 Features
106 - The parsing of the NONMEM output files has been revised and a
107 strict check is now performed after each section has been parsed. A
108 parsing error is raised if any non-compliant structures are
109 found. By doing this we get a better idea of which runs that really
110 finishes and which runs that are pre-maturely terminated
111 (e.g. through a operating system error). Remember that this is a
112 beta release and that there could still be valid text structures
113 that are not yet recognised correctly by PsN.
115 - Support for United Devices grip MP has been added and is available using
116 "-run_on_ud" command line option.
118 Bug Fixes
120 - the options directory and backwards_dir can now be specified in the
121 scm configuration file
123 - It is now possible to skip the definition of either continuous and
124 categorical covariates in the scm configuration file
126 - The exponential relation is now added correctly in the
127 scm. Previously it was incorrectly added as (for weight on volume)
128 TVV = THETA(1)*(1+EXP(THETA(2)*(WGT-median(WGT))))
129 Now it is added as
130 TVV = THETA(1)*EXP(THETA(2)*(WGT-median(WGT)))
132 - The output parse now computes correlations for omegas and sigmas correctly
133 when you have SAME blocks.
135 Changes from 2.2.1 to 2.2.2-beta1
137 Features
139 - The number of significant digits per parameter after a terminated
140 minimization are now parsed.
142 Bug Fixes
144 - The sample size option of the bootstrap did not work correctly with
145 stratified resampling. This is now fixed.
148 Changes from 2.2.0 to 2.2.1
150 Features
152 - Added support for the Intel Fortran compiler (mostly for those
153 using the new Intel Macs).
155 - We have added two new model fit quality terms: 'terminated' and
156 'really_bad' to the previous 'normal' and 'picky'.
158 picky - Minimization and covariance steps successful, no
159 parameters near boundaries and no warnings from the
160 covariance step.
162 normal - Minimization successful
164 terminated - Estimates of the objective function value and the
165 number of significant digits are available.
167 really_bad - An estimate of the OFV is available (but usually not
168 much more...).
170 Given that more than one run from a run has been performed (with
171 different initial estimates but otherwise identical) the following
172 selection procedure is done:
174 The lowest OFV of all tries is collected. Then, all runs within
175 5 units of the lowest OFV is tried according to
177 1. Of the runs passing the picky criteria, choose the run with the
178 highest number of significant digits.
180 2. If no run has passed the criteria above: of the runs passing the
181 normal criteria, choose the one with the highest number of
182 sginificant digits.
184 3. If no run has passed the criteria above: of the runs passing the
185 terminated criteria, choose the one with the highest number of
186 sginificant digits.
188 4. If still no run has passed the criteria above: of all runs that
189 at least have an estimate of the OFV, choose the run with the
190 lowest OFV.
192 5. As a last resort, choose the first run.
194 Comments: This is of course a very pragmatic approach but it is
195 needed to make the automatic procedures of the PsN tools to
196 work. We cannot force the user to manually scrutinise the results
197 of every single run of for example a bootstrap. The user has the
198 opportunity to do so if he or she wishes. If you do not feel
199 comfortable with this selection procedure or if you have comments
200 on how to improve it, please contact me (Lars Lindbom) at
201 lars.lindbom (at) farmbio.uu.se.
203 Bugfixes
205 - Data files with spaces around commas used for columns separation
206 could be problematic to parse. This is now fixed.
208 - There was a serious error in the seed handling in version
209 2.2.0. Under the following circumstances, the seed number was
210 ignored and a default seed number used: If the output file from the
211 original run was not present in the same directory as the original
212 model file when a PsN tool was started. This affected all PsN
213 tools, except the execute tool. This is now fixed.
215 - Up to 20 omegas can now be parsed in the output file. This is still
216 a limitation and it will be addressed in later releases.
218 - On fast distributed systems with a shared file system, the massive
219 parallelisation of the PsN tools could cause synchronization
220 problems (files could sometimes not be found, read or written). The
221 root of this problem is somewhat outside of the scope of PsN but we
222 have tried to find a workaround for this for the LSF grid system.
224 Changes from 2.1.10 to 2.2.0 ( released for PsN course )
226 Features
228 - PsN Output module use a proper Matrix library for handling some of
229 the matrices of the outputfile.
231 - The bootstrap can now be resumed with a lower number of
232 samples. Meaning that if you are in a hurry and want to stop the
233 bootstrap and get results after 1000 out of 2000 runs, you can. At
234 a later time you can then resume the 2000 runs.
236 - Nonmem warnings are now printed if "--verbose" is givven.
238 - Automatic incrementation of directory names are a bit smarter (no
239 more "directory modelfit_dirX is in the way" messages).
241 - Clean is much more stringent and intelligent and now manages to
242 remove entire tool directories if desired.
244 - option list printed from "-h|-?" only shows utility specific options.
246 - Default values configurable on a per utility basis.
248 - Online documentation improved
250 - Parallel run of NONMEM is not aborted as before if a outputfile is
251 not correctly parsed. It is just ignored, as if it had not
252 minimized successfully. (actually it is the output parser that
253 doesn't die, but the effect is the same)
255 - Parallel run is a bit nicer to the system when compiling many
256 NONMEMS. By adding a configurable delay between starts.
258 - Matlab and R scripts for various plots from cdd and bootstrap
259 results are distributed with PsN.
261 - Vast improvments in results presentations from all tools.
263 - Improvements in rerun algoritm with new "min_retries" option.
265 - Output from running utilties are much nicer.
267 - new "-quick_summary" prints minimization message and ofv value for
268 each finished run. As does "-verbose" option which also prints
269 "R:X" for each retrie (X is the retrie number).
271 - new "-summary" options prints diagnostics about the tool run when
272 it is finished. For "execute" the "-summary" option will have the
273 same function as the "sumo" utiltiy as well as some diagnostics.
275 - PsN now recognizes that control files may require fortran files and
276 copies them to the run directory. No need to add fotran files to
277 the --extra_files option any more.
279 - Added --extra_output which allows you to copy more than the output
280 file from the run directory.
282 Bugs
284 - PsN couldn't parse IGNORE=(ITEM=VALUE) options in $DATA. Fixed :)
286 - If star notation was used in the scm config file section [inits],
287 [upper_bounds] or [lower_bounds] there was a risk of defining
288 relations that was not specified in [test_relations]. Which in turn
289 would cause a lot of diffrent errors. Fixed :)
291 - Using low_cutoff and hi_cutoff in data::sd resulted in PsN trying to use
292 the new compact data strings as the old arrays. fixed.
294 - Removed unecessary regular expression in output file parser.
296 - Not really a bug. But i tried to add "FIX" as a record option to a theta
297 records. Then when "store_init" was called on the "FIX" option PsN crashed
298 because the option was not a "theta_option". So i added a check for
299 "store_init" abilties... if FIX would become a record option in the future.
301 - Member accessor "datafile" replaced with "datafiles" in model. Affected
302 only the bootstrap which have been updated.
304 - An assumptions about outputfiles crashed sumo occationally. Some more
305 rigorous contentchecking has been added and fixed the problem.
307 Internal changes
308 - Cleanup of dubious utility options.
309 - Option help sorted alphabetically.
310 - "flush", "target" removed from model.
311 - "remove_temp_files" removed from modelfit.
312 - Changed the Makefile considerably, much simpler now.
313 - added "column_to_array" function in data module.
315 Changes from 2.1.9 to 2.1.10
317 Features
318 - The 'picky' option of all utilities now triggers on
319 COVARIANCE STEP ABORTED and
320 PROGRAM TERMINATED BY OBJ
322 Changes from 2.1.8 to 2.1.9
324 Features
325 - Nordugrid support reintroduced (but will probably not be developed further)
326 - Sumo now supports multiple problems and subproblems.
327 - Sumo has new option "precision" which control the number of digits
328 to present estimates with.
329 - An extra 'short' logfile is output from the scm with the most important
330 information.
332 Bugs
333 - Bootstrap option "stratify_on" doesn't work with columns defined in
334 extra data files given with "extra_data" option. Not Fixed.
335 - Resumes of scm's with extra_data_files is now possible. Still some
336 problems with scm's where search_direction=both.
337 - LSF options was not communicated properly between tools and subtools
338 so submission of bootstraps, llp, cdd and scm did not work. Fixed.
339 - Default values for LSF can now be configured in psn.conf.
340 - Print order of modelfiles places $MIX before $PRED. Fixed.
341 - extra_files option was broken. Fixed.
343 Internal Changes
344 - Tool will always fork subtools, this will save memory in the scm (and
345 other recursive tools).
346 - Common options need only be specified once. In one of three categories:
347 model_options, tool_options, script_options.
348 - models can now be created with "reference" objects. This solved the
349 scm resume problem.
350 - Added function to model that cleans extra data code.
352 Changes from 2.1.7 to 2.1.8
354 Features
355 - Rudimentary support for execution on Platform LSF grid systems is now
356 included. The following extra attributes are available to all tools
357 for controlling the execution:
359 run_on_lsf - Use LSF
360 lsf_queue - Use this queue on the grid
361 lsf_options - General options for the grid queuing system.
362 lsf_project_name - Optional. May be used by the grid system
363 administrator to register control project
364 cpu usage.
365 lsf_job_name - Optional. May be used by a grid system
366 administrator to control multiple runs on a
367 grid system.
369 - All tools now have a raw_results file holding the diagnotics and
370 parameter estimates for each NONMEM run.
371 - All tools now handles model files containing one $PROBLEM and no
372 SUBPROBLEMS only.
373 - The Bootstrap and CDD routines now create separate files holding
374 information about which individuals that were included/excluded
375 in/from each created data set.
376 - The Bootstrap results file now has warnings section to better
377 display which runs that were used for result calculations and which
378 were not
379 - PsN now tries to read psn.conf from the users home directory. (Will
380 require some extra documentation).
381 - Optimized modelfit thread operation by removing Storable::Store.
382 - The clean options has four levels of operation.
384 0 - No cleaning at all.
385 1 - Clean more, but make sure resumes work (this is default).
386 2 - Clean more, but make sure summarize work.
387 3 - Clean everything, keeping only results and logs.
389 - Some attributes and accessors of the PsN-Core class 'output' have
390 had their names changed to make it more clear what their purposes
391 are:
392 termination is now called minimization_successful
393 termination_message is now called minimization_message
394 covstep_termination is now called covariance_step_successful
396 these attributes are new:
398 covariance_step_warnings (boolean; 0 or 1)
399 estimate_near_boundary (boolean; 0 or 1)
401 The attribute which was previously called termination could have the
402 values 0, 0.5 and 1 for 'minimization terminates', 'minimization
403 successful but warnings or errors were printed from the covariance
404 step' and finally 'minimization successful'. The new attribute
405 'minimization_successful' can only take the values 0 and 1 and the
406 new attribute covariance_step_warnings now takes the value 1 when
407 covariance step warnings were found. Checks are also perfomed to
408 test whether the estimates are close to any boundary.
409 - Html documents is now generated from command line help text.
410 which ensures that they are consistent and up to date.
411 - Personal configuration file now available under Unix like OS:s
413 Bugs
414 - Options specific to modelfiles were ignored in some Utilties.
415 - Gradients now output in windows. Just like with nmfe.
416 - Bootstrap now consumes considerably less memory.
417 - Removed unused/obsolete "resume" function from modelfit.
418 - Sorted method entries in modelfit diagram.
419 - Fixed bugs in parsing of OFV and eigenvalues.
421 Changes from 2.1.6 to 2.1.7
423 Bugs
424 - "clean" and "compress" options now work under Windows.
425 - SCM now handles crashed runs appropriately.
426 - The ofv_forward, ofv_backward, p_forward and p_backward options of
427 scm now works as they should.
429 Features
430 - SCM improvements. The final model is now available in the
431 first-level scm directory of each step.
432 - The Utilties -help option is improved. You can now give a list of
433 options to -help for specific documentation of those options, like
434 so:
436 $ scm -help do_not_drop
438 - Minor changes in how the [nm_version] section in psn.conf i
439 handled. Look in the psn.conf file that is distributed with PsN for
440 more information.
441 - SDE modelfiles are now supported. (see the -sde_records option)
442 - The update_inits method of the model class now accepts an output
443 file name as input.
444 - Some attributes and accessors of the PsN-Core class 'output' have
445 had their names changed to make it more clear what their purposes
446 are:
447 finalgrad is now called final_gradient
448 parampath is now called parameter_path
449 gradientpath is now called gradient_path
450 covterm is now called covstep_termination
451 - The Utilties now saves the command line that started it in a file
452 "command.txt" under the corresponding directory.
453 - Added a "silent" option to the Utilties that quenches all output.
454 - "nice" option added to utilties that allows you to set priority in
455 unix.
456 - Rudimentary support for NorduGrid reinstated.
457 - Installer script rewritten. It now supports Perl installations in
458 nonstandard directories. Discrepancies between *nix and MSWin32 is
459 smaller.
461 Internal changes
462 - The code that is autogenerated with "dia2code" is optimized a
463 bit and saves a hash "__valid_parameters" in each object, which
464 allows the constructors to be used as a generic (shallow) copy.
466 Misc
467 - continuous spelled correctly throughout SCM.
470 Changes from 2.1.5 to 2.1.6
472 Bugs
473 - Stratify_on bug in bootstrap method fixed.
474 - Data files in directories different from model file are now
475 possible.
476 - Path name handling reimplemented. Relative paths are now handled
477 better.
478 - Models that generate table files would make PsN complain about
479 missing files.
480 - Its now possible to specify multiple output names as a
481 comma-separated list to execute.
482 - The scm can now go backwards.
483 - Handling of table files now responds to the
484 "ignore_missing_output_files" flag.
485 - nmtran error messages was lost in Windows 2K. Execution of nmtran
486 is more plattform independent now. It uses a temporary file, but
487 it is never bigger than the nmtran messages.
488 - Fixed a bug that limited choice of compiler under unix.
489 - Typo in scm utility fixed. (included_relations in backwards step).
491 Features
492 - Added support for "$NONPARAMETRIC" record in model files.
493 - A few improved error messages related to parsing of model files.
494 - The SCM config file now features sections for parameter bounds and
495 user-defined code
497 Internal changes
498 - psn.conf settings are now available as global variables. psn.conf
499 is only parsed once.