ODA-226: Fixed bug where option continueWithoutObservations in AbstractSequentialAlgorithm did not work anymore. ---- Removed serialVersionUID from serializable exchangeItems. ---- Fix stand-alone tests and clean up files. ---- Removed unused test files. ---- Removing duplicate test cases. ---- optional argument to specify the JRE-environment for oda_run.bat-scripts; bugfix na testen op machine zonder %JAVA_HOME% ---- ODA-197: RangeValidation: changed code so that constraintExchangeItems have different ids from the wrapped exchangeItems. ---- Update vfproj for AIX changes. Add windows dll. ---- 1) optional argument to specify the JRE-environment for oda_run.bat-scripts 2) run_openda-XX.bat scripts made obsolete 3) run_mpi_XX.bat scripts deleted ---- Merge AIX version and changes with PC version. Add makefiles for AIX. Remove patch files for gfortran compatiability which are no longer required. ---- Explicitly free all tree-vectors en kalmanFilter Algoritms made currentState private. The set method (with is now forced to be used) will free the old current state ---- Support Native handle overview from java -added Mac binaries -added java files ---- Added support in bridge to print overview of native handles ---- ODA-197: RangeValidation: Work In Progress: wrap multiple constraints on the same exchangeItem at the same time. Do not use constraintId as id for rangeValidationExchangeItems. ---- Reduce size of tgz-files. ---- One more step for native build on teamcity. ---- Update of SWAN tests ---- One step for native builds on teamcity. ---- Remove binaries for linux32_gnu from repos. ---- ODA-197: Changed rangeValidationConfig schema to make it more user friendly. ---- Mac OS-X (re)moved some shared libraries ---- Mac OS-X added gcc libraries ---- Mac OS-X added some more missing third party labraries ---- Sort noos filenames, so they are always in alphabetical order. On linux this is not automatically so. ---- Added tgz archives as targets to build.xml for ant. ---- Mac OSX Added external libraries for netcdf and mpi ---- Added generated Mac OSX libraries to repository: TODO: add third party libraries as well! ---- Workaround for unhandled exchangeItem id="allElelmentsFromIoObject". ---- Added/activated support for NETCDF in windows native libraries. ---- Comments only. ---- ODA-197: changed unit test EfdcDllEnkfOpenDaInFewsTest to use new range validation. Improved error messages in RangeValidationConfigReader. ---- Added missing finish() method. ---- Added finish() method to model factory. Called by application runner, who now also calls the IStochObserver.free (instead of all the individual algorithms) ---- Solved some compiler warnings in native VS2010.sln. ---- Fix in setting values in Axpy call ---- ODA-197: added option to apply range validation on model exchange items in BBStochModelInstance. This can be configured in a new separate rangeValidationConfig.xml file that can be referenced from blackBoxStochModelConfig.xml file. The constraints are applied to the relevant model exchange items (as configured) by wrapping the modelExchangeItems in a ConstantLimitsRangeValidationExchangeItem that applies the limits whenever the axpyOnValues, multiplyValues, setValues or setValuesAsDoubles method is called. For now only implemented constant limits (lower limit, upper limit or both). Other types of constraints can be added in the future. ---- Fixed order of copied noos data. Reference file was constructed from noos files in reverse alphabetical order. ---- Added docu on allElementsFromIoObject to black box model config ---- ODA-197: fixed unit tests. ---- ODA-220: added info about jna jars. ---- cta_model_factory.c: - Use tstart and tstop instead of ChildComputeCurTime and ChildComputeEndTime in TimeStepAllModels. - Do not set ChildComputeCurTime and ChildComputeEndTime in CTAI_ModelFac_SetBlock. - Remove redundant ChildComputeCurTime and ChildComputeEndTime variables. cta_model_factory.h - Adapt interface for CTAI_ModelFac_SetBlock and CTAI_ModelFac_TimeStepAllModels. cta_model.c - Adapt calls to CTAI_ModelFac_SetBlock and CTAI_ModelFac_TimeStepAllModels. ---- Add warning for partial implementation of Linux and AIX versions. ---- ODA-220: Updated all jna jar files to JNA version 3.4.0. Added extra jar file aix-ppc64.jar that is needed for IBM AIX computers. ---- ODA-127: Fixed method IWdmFortranNativeDLL.wdbsgc_ so that the attributeValueStringLength is passed as an extra argument. This is needed for the code to work for JNA versions higher than 3.2.3. ---- Moved timing files to same dir as .oda file. ---- ODA-217: Created DLL functionmapper for efdc DLL compiled with XLFortran compiler ---- Modification on NetcdfGridTimeSeriesExchangeItem to support writing output of Modflow. ---- Deactivate EFDC_ddl tests for linux and mac ---- Fixed a bug in SwivtParametersResultWriter for lognormal distributions ---- Added option to script to build native from inside public ---- ODA-176: Fixed unit tests EfdcDllEnkfOpenDaInFewsTest, EfdcDllSimulationOpenDaInFewsTest and EfdcDllBlackBoxRunOpenDaInFewsTest. Only compare the first part of large output files to avoid out of memory problems in comparison in TeamCity run. ---- Re-activate the call to uncertaintyRegexMapping.keySet().toArray, but define uncertaintyRegexMapping as LinkedHashMap instead of HashMap. ---- ODA-176: Fixed unit tests EfdcDllEnkfOpenDaInFewsTest, EfdcDllSimulationOpenDaInFewsTest: Do not compare the line of the output netcdf files that contains the time of export, because that changes every time. ---- Fixed very old programming error in announceObs of native code ---- Added the announce_obsvalues to model factory (native models) ---- ODA-119: fixed bug in SwivtParametersResultWriter with path of optimal parameters file. ---- Fixed problem on some platforms loading libcta to models ---- ODA-180: Fixed bug: changed org.openda.observers.SwanResults to org.openda.model_swan.SwanResults in config files in zip file public/model_swan/java/src/org/openda/model_swan/swivt/calibrationConfigTemplate.zip ---- ODA-119: fixed bug in SwivtParametersResultWriter (optimal parameters were not written to file). ---- Switched on model_efdc_dll tests in ant build script. ---- Added aboutMeerpeilcorrection.html, forgotten in rev 3806. ---- ODA-119: added version number to title of CaseExtractorGUI window. ---- Add new windows dll for model_efdc_dll ---- Added algorithm and observer class for "Meerpeilcorrection", as developed for the FEWS-Meren project. ---- - When pass string two fortran, pass the string length at the end. - Set SVN:ignore on original_efdc_files and openDA_wrapper. - Use parameter max_path_length. - Check if passed path are not exceeding the max_path_length. - openda_wrapper_init_ is now a function returning 0 on succes. - fix uninitialized variables in READWIMS1.for. ---- Fixed crash when model has more sub-treeVectors than max maxNoCols-fixedColumnCount. Now amount of output is truncated to maxNoCols. ---- removed extension ".zip" from the restart filename for CtaModelState: a CtaModelState is written to NetCDF file(s) that is(are) zipped after writing. ---- The number of steps was one too large for AbstractSequentialAlgorithm ---- Add tolerance for last run from last obs to end of simulation ---- Linux tests for teamcity. Test with startup of Xvfb ---- Fixed error messages at initialization of native library (related to optional user libraries) ---- Removed old COSTA example input ---- Added two examples native_advec1d and native_oscill based on old costa examples ---- Added test-linux64-teamcity target to ant ---- Correct the method getUncertaintyIDs() to make sure that the order of elements of the observation realization vector remain consistent with the original observation vector. ---- Native: fixed memory leak in treevector (not all locals vars were freed in case of an error) Java: EnKF improved freeing of memory: -improved scope when calling gc() -explicitly free all (large) local treevectors when content is no longer needed ---- ODA-176: NetcdfDataObject: Removed unnecessary warning message. ---- native: fixed dependencies ---- Deleted obsolete post-build event. ---- Native compilation in windows: Changed dependencies such that build order makes sure that alll libraries are build before post-build action ---- Added build before running tests ---- Native: Removed (non-working) visual studio 2008 project files ---- Native code: Added model project files to vs2010 solution And many changes in the project settings to make it all compile/run ---- Remove binaries for linux64_gnu ---- remove mpic++ after build for teamcity ---- attempt to fix linux64_gnu build on teamcity ---- Updated build script for native ---- One step in fixing build scripts for teamcity ---- Update of linux64_gnu libs ---- Update build script for native on teamcity ---- Move output to correct directory ---- Removed libcta.lib from native_bin. Restored LoadLibrary to a single statement. ---- Correction to build directories and post-build-event. ---- Addition to previous commit in order to please TeamCity. ---- - adding enum for output level in IResultWriter - correct for implementation on GUI ---- Added quiet method around attempt to load user observers. This solves large list of ERRORS/WARNINGS at beginning of run ---- Changed libopendabridge project to opendabridge. Added post-build event to copy dll's to native_bin. Added required loadLibrary statements to CtaArrayTest. ---- Removal of previously committed file. It does not solve the problem and there is a TeamCity configuration option to replace its intended function. ---- Introduction of a batch file for setting the java.library.path. This aims to solve the missing dll's error resulting from the System.loadLibrary("opendabridge") call in CtaObject.java. ---- Added DistributedCounter to initialize seed in StochVectors' random generator. Parallel runs did not produce correct random numbers. (same on all processes). ---- Fix export of CTA_MESSAGE_QUIET. Syntax seems to be case dependent: #define CTA_MESSAGE_QUIET_F77 F77_CALL(cta_message_quiet,CTA_MESSAGE_QUIET) Note the lowercase cta_message_quiet. ---- Fixed jar dependencies in Intelij project ---- ODA-176: Fixed bug in IoObjectStochObserver where exchangeItems with no values at all were assumed to be time-independent. ---- ODA-203: Fixed NullPointerException in TimeSeries.getValuesAsDoubles. ---- The function CTA_Message_Quiet is now made available from Fortran as well. This function allows to "silence"/"unsilence" the native (not Java!) message handler. ---- Added Mac executable of SWAN for the swan_l21triad test Improved matlab script (some outputs seems to have been renamed in the algorithms) ---- Added Mac executable of SWAN for the swan_l21triad test ---- Several additions/improvements to the documentations on parallel computing and performance tuning ---- Added comment only. ---- ODA-176 Convert source files to linux line endings (single LF). ---- ODA-176 Makefile: - Add flags for debug target. ---- ODA-176 model_aser_time.series.f90, model_cser_time_series.f90, openDA_wrapper.f90 : - Remove unused variables. Makefile: - Add extra flags for debug target. ---- ODA-XXX Set the text encoding to UTF-8 for the source files in the javac directives in all build.xml files. Otherwise system defaults are assumed for text encoding, which might result in a ant built failure. ---- ODA-176 EfdcDLLTest.java: - Change the value for the reference period in the assertion as it was changed in the configuration. ---- Fixed bug in function name ---- Support for restarts in SparseDud method. Also added SparseDud restart test. ---- Improved error messages ---- $ODA-200\nMaori fixes ---- Fixes in MAORI observer ---- Improved error messages ---- correction in (Sparse)Dud: remove double initialization testbank: added SparseDud tests in CalibrationTest.java for this purpose: change SimpleStochmodel::getParameters from Vector to TreeVector consequence: cast SimpleStochmodel::getParameters back to Vector in many cases, to get same results as reference ---- ODA-176 readme.txt: - Include information on building the shared library. ---- Changed version number of trunk to 'trunk' so that it is different from any stable releases. ---- ODA-173: Wflow2DMapStateExchangeItem: Fixed incorrect length check. ---- SparseDud: added scaling of previous results (based on cost function) Dud+SparseDud: merged and unified code, by making both inherit most of their code from BaseDud and BaseDudCoreOptimizer ---- ODA-173: Wflow2DMapStateExchangeItem: changed code for handling inactive grid cells. Now the get methods return values for all grid cells and the set methods expect values for all grid cells, also for the inactive grid cells. This is not a problem, because the wflow model ignores the values of the inactive grid cells anyway. ---- ODA-173: NetcdfUtils: Fixed bug where missing values were not replaced by fill value in output netcdf files. ---- ODA-176 Enable unit tests on Linux. ---- ODA-176 Add libEFDC.so for linux32_gnu. ---- Add to DataDumper and DataCopier utilities: driver scripts, series tests, etc. ---- ODA-176 Change reference period to 60 seconds. ---- ODA-176 Disable Unit test for Linux for now. Set NTSCRN in EFDC.INP to 12. ---- ODA-176 Add libEFDC.so for linux64. ---- ODA-176: EfdcModelFactory: fixed bug where EfdcModelFactory did not work when timeHorizon was not set from OpenDaInFews. ---- ODA-170: EfdcModelFactory: Removed Delft_PI.jar and Delft_Util.jar from public part. These are not needed anymore because the timeHorizon can now be set from OpenDaInFews, so no need to read pi run file in EfdcModelFactory anymore. ---- ODA-170: WflowModelFactory: Removed Delft_PI.jar and Delft_Util.jar from public part. These are not needed anymore for two reasons. First the timeHorizon can now be set from OpenDaInFews, so no need to read pi run file in WflowModelFactory anymore. The WflowModelInstance now reads/writes data to/from netcdf files, so no need to read/write pi map stack files anymore. ---- ODA-173: WflowModelInstance: fixed bug with writing output netcdf files. ---- ODA-173: MapsNoiseModelInstance: added option to start with a cold state start. ---- ODA-176 Enable Unit tests on Linux. All tests use linux64_gnu/libEFDC.so. ---- ODA-176 Add linux library for 64bit machines. ---- ODA-176 Add 11.0.72 version of Intel Redistributables. ---- ODA-173: WflowModelInstance: Fixed bug where timeHorizon was not passed from OpenDaInFews. ---- ODA-173: WflowModelInstance: Fixed bug where state data was not written to output netcdf files. ---- ODA-176 Change makefile for efdc_fortran_dll_test to use libEFDC.so from native_bin Remove EXTERNAL RAND for gfortran compatibility in .patch files. ---- ODA-176: EfdcModelInstance: changed code so that netcdf output is not written for startTime of timeHorizon. Updated unit tests. ---- ODA-176: TimeUtils: Fixed bug with incorrect number of timeSteps. ---- Increase number of digits written to model_output.txt for start and end time. ---- Fixed problem with TimeConsumer action of BBStochModel on noise models. ---- Update GfortranFunctionMapper to include all dll functions Use GfortranFunctionMapper when running on linux ---- ODA-176 Return model start and time rounded to minutes. Use doubles to store time for the model instances. ---- ODA-176 Updated makefiles for linux: - use normal make, make install, make clean syntax - only patch files when compiler is gfortran - exclude efdc_fortran_dll_test from targets openDA_wrapper.f90 - check if valid instance id is passed when getting the reference year model_aser_time_series.f90 - use .not. to check call_on_init logical ---- Removed old code. ---- Fixed bug in Dud for combination of obs-selection and custom stop criterion. ---- ODA-176: Fixed efdcDll openDaInFews unit tests. ---- Change IComposbleDataObject to use the new IExchangeItems. Part 2 ---- Extended TimeSeries to new IExchangeItem format. Part 2. ---- Extended TimeSeries to new IExchangeItem format. ---- Change IComposbleDataObject to use the new IExchangeItems ---- Updated eclipse .classpath file. ---- ODA-176: NetcdfDataObject and EfdcNetcdfDataObject: changed code so that data for coordinate variables is written as soon as it is available, instead of just before the netcdf file is closed. ---- Added version messages to the application runner ---- Add compiled DLL ---- model_aser_time_series.f90, openDA_wrapper.f90: - Check on end of compute if we are at a day boundary, if so update daily averages. ---- model_aser_time_series.f90, openDA_wrapper.f90: - Save and restore calculated daily averaged solar intensity when switching instances. model_state.f90: - fix typos in parameter names. WQ3D.for: - on restart round time to midnight when calculating SUNDAY1 ---- ODA-170: fixed bug with timeHorizon in BBModelInstance. ---- openDA_wrapper.f90: - when destroy() is called check if calheat.dia is open and close otherwise. WQ3D.for: - round TIMEDAY to day boundary for calculating daily solar instensity. ---- Fix time initialization in BBModelInstance ---- -Added documentation of performance tuning -Improved documentation on parallel computing ---- Include current test setup in model_efdc_dll/tests ---- ODA-176 Makefile in native/efdc_fortran_dll which calls the makefiles in openDA_wrapper and original_efdc_files directories. ---- Added IntelliJ module file for wflow ---- Updated documentation on parallel computing ---- ODA-170: changed EfdcModelFactory to implement the new ITimeHorizonConsumer interface. ---- ODA-170: Added ITimeHorizonConsumer interface to facilitate passing timeHorizon start and end time from application runner to (stoch)ModelFactories that need this. This is needed e.g. when OpenDA is used from an operational system. ---- ODA-173: Fixed bug in WflowModelFactory with resolving path of input files. Updated unit test WflowModelOpenDaInFewsTest. ---- ODA-176: removed redundant arguments from EfdcNetcdfDataObject. ---- ODA-173: Extended NetcdfDataObject so that it can read data for 2D grid time series from netcdf files. Changed WflowModelFactory and WflowModelInstance code so that it uses netcdf files instead of piMapstack xml files. ---- ODA-173: added config for WflowModelFactory. ---- ODA-176: Corrected typo in phosphorus in efdc exchangeItemIds. ---- Distribute global settings like timePrecision en VectorPrecisionIsReal to the server process as well. ---- Moved files writen for the wrapper to openDA_wrapper directory. Separate makefiles for openDA_wrapper and origanal_efdc_files Updated Visual Studio 2010 project for new file locations. OpenDA_wrapper.f90 : Fix preprocessor directives. ---- Temporary rename, to rename file from uppercase to lowercase ---- Added intelliJ iml file for efdc DLL based wrapper ---- New option in ThreadStochasticModel "nonBlockingAxpy" ---- Add patch per file for gfortran compatibility. Add Makefile (for linux) which automatically applies patch files. ---- ODA-176: EfdcDll: Fixed unit test. ---- ODA-176: EfdcDll: Fixed unit tests. ---- Revised source documentation. Removed comment code parts. ---- detection of 32/64 bit linux in settings script ---- ODA-176: AbstractSequentialAlgorithm: Added option to put "coldstart" instead of timestamp in names of restart files. ---- Add patch for gfortran compatibility. ---- Support (switch) for single/double precision computations in OpenDA (storage and computaions of org.openda.utils.Vector Note other objects could/should be extended for variational precision as well when needed. ---- Made new SwanResultsTimeDependent class Serializable such that it works for parallel runs as well ---- ODA-176: TimeSeriesNoiseModelInstance: added option to start with a cold state start. ---- Using OpenDA from Matlab -Added a java utility class for Matlab. Currently used to solve the problem of the enumeration in the getInstance method of the IstochModelFactory -added a small matlab example where the lorenz model is ran from Matlab ---- Add windows XP DLL ---- openDA_wrapper.f90, IEfdcFortranNativeDLL.java, EfdcDLL.java: - Get reference year at creation of new model instance, instead of at DLL init. - Trim string with directory name on output. original EFDC files: - Added debug output to check time series. - Clean up of unused variables. ---- Add windows XP compiled DLL. ---- Added svml_dispmd.dll ifort redistributable. ---- Add ifort redistributables dlls. openDA_wrapper.f90: multiply PSER with 9.81 on setting and divide when getting. ---- ODA-119: Fixed model_swan castor build script so that it does not generate duplicate classes for types from imported core schemas. ---- Created separate run-method in ApplicationRunnerSingleThreaded, to avoid exceptions in constructor ---- ODA-173: Removed unused jar file netcdf-4.1.jar. All classes in this jar file are already present in the jar file toolsUI-4.1.jar. ---- ODA-176: removed code to re-initialize the efdc dll, because this does not work. ---- ODA-176: added code to garbage collect the efdc dll. ---- openDA_wrapper.f90: - start and end time are now set for each instance instead off on dll init. - when getting the start or end time use the instance ID. - remove conversion for time series as this is done by FEWS. IEfdcFortranNativeDLL.java: - update interface of mp_get_start_time_ and mp_get_start_time_ to pass model instance ID EfdcDLL.java: - pass the model instance ID when getting start and end time. ---- ODA-176: Fixed bug where EfdcDLL.free method was not called to free the dll. ---- Updated project files etc for VS2010 ---- Moved Unit Tests to test directory ---- First version of SimpleOscillator OpenMI ModelFactory now works ---- ODA-173: copied NetcdfDataObject to EfdcNetcdfDataObject (used for EfdcModelInstance). This way possible to refactor/extend NetcdfDataObject without affecting EfdcModelInstance. ---- Replacing : /openda_1/public/trunk/core/java/src/org/openda/exchange/dataobjects/NetcdfDataObject.java (Copy from path: /openda_1/public/trunk/core/java/src/org/openda/exchange/dataobjects/NetcdfDataObject.java, Revision, 3588) [model_cser_time_series.f90, model_aser_time_series.f90, model_pser_time_series.f90, model_qser_time_series.f90] : - Only reallocate if storage for longer time series are required (not for more layers, more locations or more variables). - Explicit use of indexes when copying arrays. model_qser_time_series.f90 : - Correctly set return value to zero if reallocation is successful. - Write debug information to file. model_cser_time_series.f90 : - Restore values after reallocation and deallocate temporary arrays. - Only get and set the values for existing exchange items. - Store offset for water quality indexes here instead of in openDA_wrapper.f90. openDA_wrapper.f90 : - Call model_cser_set_items(). ---- Some if-then conditions are added to check whether analysisTimes is null, with an appropriate action added accordingly. ---- Fixed some bugs for spatially correlated noise. ---- ODA-173: Fixed bug where IStochObserver.free() was never called. Now this is called in IAlgorithm.finish() method for each algorithm. ---- Add DLL compiled for XP with ifort version 11. ---- openDA_wrapper.f90: - Add bottom elevation to water depth to get WaterLevel. (subtract when setting waterdepth) - Output add start of compute to model-output.txt. - On init read values from time series files and store for every state. Is overwritten when using exchange items. original_efdc_files: Fix init of reals (0.0 instead of 0). ---- Vector supports both single as double precision storage and computations. Development version: double precision (old format) is default for all users. ---- Added picture on ikvm usage ---- Removed confusing CET date/time print from logging ---- Minor fix in Obs/Pred logging ---- Added model instance string to logging. Removed values on duplicate time stamps from buffer ---- openDA_wrapper.f90: Print (or less if not available) values to model_output.txt. EfdcDLL.java: get_ and set_values use correct array size if a subrange (instead of the total range) of the grid indices is used. ---- EfdcDLL.java: Pass zero based start and end indices (Java) with set_values and get_values methods to Fortran EFDC wrapper openDA_wrapper.f90: Convert start and end indices to EFDC grid indices (2:LA). Only set or get values for grid points (not the additional values at index 1, LC and LCM). Check that converted index is in range 2:LA. openDA_wrapper.f90: When destroy is called reset instance numbering and total to zero. ---- Corrections on the implementation of the new putValue on other result writers (other than MatlabResultWriter): it calls now the old putValue instead of prompting error message that it has not been implemented. ---- ODA-173: NetcdfDataObject: fixed bug where _FillValue attribute was not used to identify missing values that are read from netcdf files. Also fixed bug where scale_factor and add_offset attributes were not used to read packed data values from netcdf files. ---- EfdcDLL.java: RestoreInstance from memory before compute and save afterwards, instead of switching state before compute. ---- ODA-176: EfdcModelDll: added log messages to EfdcModelInstance. ---- Fixed build. ---- ODA-176: EfdcModelDll: changed EfdcModelInstance so that renames the output state files to the names of the corresponding input state files and puts the renamed versions in a separate folder (to avoid overwriting the input state files, so that run is reproducable). ---- ODA-176: EfdcModelDll: committed new version of efdc dll for Windows XP. ---- ODA-173: NetcdfScalarTimeSeriesExchangeItem: filter out times for which there are missing values. ---- ODA-176 RESTIN1.for, EfdcModelInstance.java: Renamed input restart file from TEMP.RSTI to TEMP.RST. RESTOUT.for, EfdcModelInstance.java: Renamed output restart file from TEMP.RST to TEMP.RSTO. OpenDA_wrapper.f90: Use EFDC.INP to select restart or start from zero. This implies that all instance are initialized with restart files from the 'model' directory. FEWS can override by providing each instance with its own restart files. GFortranFunctionMapper.java: Added name mangling translation for all subroutines. makefile: added gfortran for linux compile option Removed windows dependence from unit tests. Changed settings in EFDC.INP and checks in unit files. ---- Changed parameters names of the save/load/restore persistent state variables after name change (confilicting with function names when fortran uses uppercase calles). This was a missing file in a previous commit (not noticed in Windows builds since no parallel support is compiled for Windows) ---- First implementation of the new results selection, where the result items can be selected according to the following options: id, minSize, maxSize, outputLevel, and context. So far it is only implemented for Dud and the SimulationKwadraticCostFunction. The original implementation of the result selection is still working. ---- Clean up a block of disactivated code from SwanSpectralFile.java. ---- Correct the indexing of a loop within BBStochModelInstance.java, where the axpy operation takes place. ---- Black-box wrapper for DAM (Dike Analysis Module) Flow model: modify EnKF example test1. ---- Add windows XP version for DLL ---- documentation.txt: Documentation for the EFDC DLL model for OpenDA ---- openDA_wrapper.f90: Added log messages, general log messages go to model directory, during the compute method the INP from the model instance directory are used, use unix line endings. model_cser_time_series.f90: Use unix line endings. ---- Added Visual studio 2010 project files ---- Several changes and bugfixed in native part of OpenDA found and solved when recompiling OpenDA for windows/porting to Visual studio 2010 ---- EfdcFortranDLL.dll: Added XP dll ---- model_aser_time_series.f90: only set last time to big number to prevent out of bounds for interpolation openDA_wrapper.f90: add check on missing values, inquire if old log file is still open for a different file unit and close that unit CALCSER.for: output also water quality interpolated values CDIAG.out, keep file recent for last time step ---- Fixed error in calling sorting routine (capital in name) ---- Changed labels and title of the cost function plot. ---- ODA-173: NetcdfDataObject: set isLargeFile true for newly created netcdf files, in case much data needs to be written. ---- Added windows XP build of dll ---- [model_aser_time_series.f90] intialize TASER to 1e38 to prevent unused exchange items to crash the model [model_pser_time_series.f90] intialize TPSER to 1e38 to prevent unused exchange items to crash the model [model_qser_time_series.f90] intialize TPSER to 1e38 to prevent unused exchange items to crash the model [model_cser_time_series.f90] intialize TCSER to 1e38 to prevent unused exchange items to crash the model [openDA_wrapper.f90] Better logging, clean old log file for instances. Use time conversion for scalar exchange items, do not get scalar time series from EFDC as it overwrites ones set by OpenDA. ---- ODA-173: NetcdfDataObject: minor refactoring. ---- ODA-176: EfdcModelDll: committed debug version of efdc dll for testing purposes. ---- Added Mac black box support (including swan-binary) for the SwanSimpleBBEnKF test ---- ODA-176: EfdcModelDll: fixed bug with timeZones. Now the timeZone of the Efdc model dll can be configured in the efdcModelFactoryConfig file. ---- Added pause to error message ---- Cleaned up code for writing ITreeValue's header and values ---- ODA-176: EfdcModelDll: replaced efdc dll with version built with Visual Studio 2008 in 32 bit Windows XP. ---- ODA-158 Improved performance of TreeVector-TreeVector operations Methods like axpy now perform up to 10 times faster! Note: TreeVector-TreeVector operations between different structured TreeVectors are now a bit more strict! Before everything worked as long as the total length of the two treevectoes was the same. Now Treevectors can have different structure BUT at subleafs we need the SAME number of subTreeVectors each having the same length. Different formulated: The structure of the two TreeVectors must be the same until one of the treevectors ends up in a leaf. (Note is is similar to the behaviour of native Treevectors. Some examples x=[[1 2][[3 4 5] [6 7]]], y=[1 2 3 4 5 6 7], z=[[1 2][3 4 5 6 7]] can all be combined. But a=[[1 2 3][4 5 6 7]] cannot be combined with x,y or z (which will probably be a bug combining unrelated values in all our implementations) ---- [openDA_wrapper.f90] Corrected function interface for get_values and set_values functions. Explicit casting to doubles or real. [model_state] Initilazing with zeros of real arrays. ---- ODA-176: EfdcModelDll: fixed bug in EfdcModelInstance.compute where model was run too far. For this had to add a tolerance in the time comparison. Also updated unit tests. ---- ODA-176: EfdcModelDll: updated unit test. ---- ODA-173: NetcdfDataObject: now times are written in netcdf file for NetcdfGridTimeSeriesExchangeItems. ---- ODA-158 Switched synchronization off in EnFK ---- ODA-158 Improved performance java vector-vector computations Improved timing (location/name/subtimers) ---- Implemented default alternative to SCE : LB and UB are 1 stdDev from the parameter values. ---- ODA-158 Made DoubleArraySearch Serializable ---- ODA-158 Made IPrevExchangeItem Serializable ---- ODA-184 Rename Abstract Model Factory ---- Updeted IKVM version Create VS2010 solution ---- ODA-176: EfdcModelDll: now referenceTimePeriod is read from the efdc dll and state input/output files are handled correctly. ---- Added mac exe tot black box wrapper config ---- Add input file for junit tests (sorry). ---- ODA-176: EfdcModelDll: temporarily switched off unit tests for linux. ---- ODA-173: NetcdfGridTimeSeriesExchangeItem: removed dependency with efdc_dll. ---- Update of gfortran libs ---- Forgot some linux64_gnu libs ---- Local settings changed for Martin Verlaan ---- Remove sqlite from cta.h and automake 1.12 too new for many systems. ---- Update of linux64 libs ---- Remove sqlite dependency. Removed remaining -lsqlite in opendabridge makefile ---- Swan filter is now working for wind. Todo: fix test-cases ---- Renamed Unit tests to start with Efdc. Started with separate unit test for scalar and grid exchange items. ---- Removed dependency from efdc (todo: fix in efdc) ---- Initialize arrays for storing state and forcings to zero. ---- Removed sqlite3 header file from include ---- Removed dependency of libsqlite3. The implementation is now compiled in ---- EfdcDLL: Correctly pass the arguments to get_values for grid exchange items IntelFortranFunctionMapper: Added the get_values_ and set_values_ subroutine mangling EfdcRestartTest: Added Unit test for saveInternalState en loadPersistentState methods build.xml: Tests use the JAR files from public/bin model_state.f90: Selective variable import from global.mod Removed several debug output ---- ODA-176: Added a method to get the reference period from EFDC ODA-176: Partial implementation of SaveInternalState and RestoreInternalState methods ---- Added detailed logging to algorithm and cost function ---- ODA-173: NetcdfDataObject: fixed bug in NetcdfGridTimeSeriesExchangeItem where output times were not found because tolerance was too small. ---- ODA-176: EfdcModelDll: now compute method uses referenceTimePeriod. ---- ODA-173: NetcdfDataObject: fixed NullPointerException in writeMetaData method. ---- ODA-173: NetcdfDataObject: added new class NetcdfGridTimeSeriesExchangeItem that can be used to write a grid time series to a netcdf file, in the format of the CF-1.6 conventions. ---- Added latest version of SQLITE3 (single-file version) ---- Added test application (EnKF / Oscillator StochModel) ---- ODA-173: NetcdfDataObject: added new class NetcdfScalarTimeSeriesExchangeItem that can be used to read one or more scalar time series from a netcdf file that is in the format of the Delft-FEWS scalar time series netcdf export. ---- Oscillator StochModel test now works ---- A few index fixes ---- Black box stochmodel / SimpleOscillatorModel test runs (state values not ok yet) ---- Some small fixes ---- Adjusted test data for black box stochmodel based version for dotnet SimpleOscillatorModel ---- Implemented black box stochmodel based version for dotnet SimpleOscillatorModel (test not running yet due to error in noise model ---- Fixed error in model class duplication by proper initialization ---- Reverted test change in Makefile.am ---- Fixed error in latest change (pospone obsdescr creation). When we need it internaly we HAVE to create it ---- Solved problem by adding script autoreconf_fix.sh. This script must be used in stead of autreconf to create a new "configure". See new script for details. Note problem is not in OpenDA but caused by gfortran/automake/autoconf incompatibility Commit2: forgot to update some m4 macros ---- Solved problem by adding script autoreconf_fix.sh. This script must be used in stead of autreconf to create a new "configure". See new script for details. Note problem is not in OpenDA but caused by gfortran/automake/autoconf incompatibility ---- ODA-176: EfdcModelDll: Added efdcModelFactoryConfig.xsd to make EfdcModelFactory configurable. ---- ODA-176: EfdcModelDll: Added model_efdc_dll to public build.xml script. ---- ODA-176: EfdcModelDll: removed unused classes. ---- ODA-176: EfdcModelDll: Changed class EfdcExchangeItemConfig into enum EfdcExchangeItemType. ---- ODA-176: EfdcModelDll: finished EfdcModelInstance. ---- Fixed ANT zip build. ---- Another update of linux64 libs ---- Update of linux64 libs ---- Shifted the efdc_dll down in the ordering for eclipse. ---- ODA-176: EfdcModelDll: fixed compute method. Temporarily fixed unit test. ---- ODA-176: EfdcModelDll: Dll now correctly returns current time. Partial implementation of loadPersistentState and saveInternalState methods Corrected preprocessor directives in Fortran code Switch off some EFDC output in EFDC.INP in the testData Changed reference time to 5 seconds in EFDC.INP Cleaned some code ---- ODA-176: EfdcModelDll: added check for currentTime to ModelDLLTest. ---- ODA-176: EfdcModelDll: Temporarily fixed unit test. ---- Wflow: refactored. ---- ODA-176: EfdcModelDll: renamed EfdcBoundaryExchangeItem to EfdcScalarExchangeItem. ---- ODA-176: EfdcModelDll: Implemented EfdcBoundaryExchangeItem and finished EfdcModelInstance implementation. ---- ODA-176: EfdcModelDll: Renamed EfdcGridStateExchangeItem to EfdcGridExchangeItem and continued EfdcModelInstance implementation. ---- ODA-176: EfdcModelDll: Implemented EfdcGridStateExchangeItem and EfdcModelInstance. ---- Add Visual Studio 2008 Solution file for model_efdc_dll. ---- ODA-176: EfdcModelDll: added comments. ---- Fixed missing or wrong export directives for the Win32 DLL. Changed Visual Studio solution post-build event to copy EFDCFortranDLL.dll to native_bin/win32_ifort/. ---- ODA-176: EfdcModelDll: added EfdcModelFactory. Renamed EfcdInstance to EfdcModelInstance. ---- Changed List usage in C# ---- Added support for changing the length of time series. Removed remainder from SimpleModelFortran. ---- Added unit test for c# version of OscillatorModel (test not yet running) ---- ODA-176: Refactored: moved common modelFactory code to BBUtils for code re-use. ---- Added solution file for vs2010 ---- Minor fix to make everything compile ---- Reorganized visual studio solutions, added SimpleOscillatorModel as test case (not running yet) ---- Attempt to fix dead-lock in threadding ---- Added windows dll (binary). ---- Added support for the OpenDA compute method Changed READWIMS1.for for to corectly do runs shorter than one day Changed EFDC.INP to a shorter reference period Changed WQALGG.INP as it could not be read more then once ---- Moved dotnet solutions one level up, to be able do split code in src and test part ---- Added debug-log-info to time dependent discharge filter ---- Marked EnSSKF and SSKF deprecated. ---- Removed Fortran sources from libcta. Replaced fortran integer sorting with a new C-implementation ---- Added option to cash the state and parallelize the getState-method (only first call) ---- ODA-176: added model_efdc_dll to eclipse .classpath file. ---- SimonaTimeSeriesFormatter: Fixed bug in TimeUtils.mjdToDHMAsString. For irregular simona time series the time stamps were written incorrectly. ---- Update of linux32_gnu libraries ---- Inlcude Makefiles for native EFDC code. Jave method for grid exchange items. ---- Changed name (label in internal native admin) of the maori sobs class ---- Completed reading and writing for SWAN grid files. ---- For native observers; do not directly create the obsDescr at creation of Observer but pospone until it is needed. ---- Support for default user observer ---- Reset parameters numberEvaluation and nextEvaluation in the constructors of SimulationKwadraticCostFunction, RMSECostFunction, SimpleCostFunction, and SimpleLeastSquaresCostFunction classes. This will prevent the iteration number, which is displayed in the OpenDA GUI, from continuing from the last iteration number of the previous run. ---- Made tests run again (adjustes test checks according to the fact that stoch model state files now are stored with extension .zip). ---- Restored DATools castor jars ---- Added Visual Studio project and solution files. Fixed faulty assignment in openDA_wrapper.f90. ---- Add fortran tests for EFDC. ---- Add EFDC source files. Add source files for EFDC exchange items. ---- Changed names of loaded functions for MAORI observer ---- Added automatic loading of USER/MAORI observers ---- Removed maori-Fortran implementation (Observer) ---- Various clean ups in java/dotnet bridge ---- Deleted unused DATools files ---- Created dir for zipping states if it does not yet exist. ---- -Added (hard coded) option to serialize vector in single precision (prototype) -Added (hard coded) option for rmi-model to cash state (prototype) -Made IArray Serializable -Some minor improvements ---- ODA-168: WflowModelInstance: fixed bug where stateInputDir was not set correctly. ---- ODA-168: AbstractSequentialAlgorithm: fixed bug where saved internal state could not be restored (extension was not removed from filename). ---- ODA-168: rebuilt castor classes. ---- Implementation of all time series Exchange Items Added latest data for Unit testing ---- Update of linux32_gnu libraries ---- Added user definable id for combined vectors in regularisation constant. Added noise model state(s) to bb stoch. model instances state ---- Added prototype of generic smoother model ---- Fixed many warnings in the native routines ---- Issure: ODA-165 Initial implementation of narive Array. This version is not fully optimized but all methods are implemented. This allows us to do more testing/usage and prototyping in native code as well. Unit test is added and new F90 interfaces are created. Hidden for normal users is some additional support for user dll's with a default implementation of a native stochastic observer. ---- Partial implementation of ModelInstance and ExchangeItems ---- ODA-154: Wflow: now the Wflow2DMapStateExchangeItem filters out the values for inactive grid cells, because the algorithms cannot cope with inactive grid cells. ---- ODA-154: Wflow: fixed unit test. ---- Issue ODA-166: add an EnKF test example for swan version with netcdf format for the state/restart/HOT file ---- Modify NetcdfDataObject and the corresponding unit test for SWAN netcdf state file. ---- Modify NetcdfDataObject and NetcdfUtils to work with SWAN netcdf state file, add a unit test. ---- Initial commit for EFDC dll wrapper. This is taken from model_example_fortran with IPrevExchangeItem replace with IExchageItem. ---- ODA-154: Wflow: removed unused class. ---- Black-box wrapper for DAM (Dike Analysis Module) Flow model: rearrange class names, exchangeItem IDs, examples. ---- Small improvements for running .Net version ---- Modify xml file examples, that show how to use the following options: "continueWithoutObservation", "skipAtInitialTime", and "skipAtFinalTime". ---- Some extensions for ArrayExchangeItems ---- Fixed a small syntax error to make it compile again ---- ODA-154: Updated unit test. ---- ODA-154: OpenDaInFews: fixed NullPointerException in AbstractSequentialAlgorithm.getTimeHorizon after call to AbstractSequentialAlgorithm.finish(). ---- Fixed typo. ---- ODA-154: Fixed bug where input pi mapstack files always had to be in inmaps folder. ---- ODA-154: Fixed exception when wflow model returns 0 variables to create exchangeItems. ---- Refactoring of IVector and ItreeVector. ---- The xsd file of the sequentialSimulation is modified to accommodate the new option "continueWithoutObservation", as well as "skipAtInitialTime" and "skipAtFinalTime". To avoid confusion, the sequentialEnsembleAlgorithm.xsd is rearranged. ---- Remove SwanWindFile.java, SwanWindFileExchangeItem.java, and SwanWindFileTest.java. These still use IPrevExchangeItem and have been replaced by a newer version SwanField2DXYFile.java, SwanField2DXYFileExchangeItem.java, and SwanWindField2DXYFileTest.java, respectively. ---- Fixed wrong import and a modified settings_local for my laptop ---- ODA-154: OpenDaApplication and OpenDaInFews: when the GUI is not used, then use ApplicationRunnerSingleThreaded instead of ApplicationRunner. This is possible because the pause and resume functionality in ApplicationRunner is only used in the GUI. This way the algorithm and models are initialized and used on the same thread, which is required for model(adapters) that are not thread-safe, e.g. when using JEPP to run python code. ---- Fixed connection between TimeSeriesNoiseModel and blackbox ---- Black-box wrapper for DAM (Dike Analysis Module) Flow model: rearrange class names, exchangeItem IDs, examples. ---- ODA-154: OpenDaInFews: fixed bug in model arguments and improved error messages. ---- Black-box wrapper for DAM (Dike Analysis Module) Flow model: rearrange class names, etc. ---- Black-box wrapper for DAM (Dike Analysis Module) Flow model. ---- Changed native autmake cripts for compilation on OSX-Lion Bridge is now always build jni header files must be present ---- Added warnings in message files when output is not generated due to ProductionRun option ---- Black-box wrapper for DAM (Dike Analysis Module) Flow model. ---- Improved precision of timing ---- fixed bug (with selecting architecture) in settings_local_linux.sh ---- Issue ODA-161: when TRESTART=TEND of simulation, a restart_*.zip file is now created when there is no observation and continueWithoutObservation=true ---- Fixed bug (pointers to local variables) in timestep administration ---- ODA-154: Wflow: fixed memory leak in the way that Jep was used in WflowPythonToJavaAdapter. ---- ODA-154: Updated wflow bin files for unit test. ---- ODA-154: ApplicationRunner: added code to read input data from pi mapstack input files for each timeStep. Fixed bug in data file format written in output pi mapstack xml files. Fixed unit test. ---- ODA-154: Wflow: minor refactoring. ---- ODA-154: Wflow: changed Wflow2DFieldStateExchangeItem so that it returns a 2D TreeVector so that it is possible to use only a subvector of this, e.g. when there are observations for only a single location. ---- Fixed bug where DimensionIndex class did not work for 3 dimensions. ---- ODA-154: Wflow: added code to remove old model instance dirs before each new run. ---- Replace castor jar with version that is compiled with java 1.6 instead of 1.7 ---- Added restart functionality for fews: no timetags option and coldstart option. ---- ODA-154: Wflow: fixed bug where pi mapstack xml files were not written for output maps. ---- changed order of subprojects within public for eclipse ---- ODA-154: Wflow: changed Wflow2DFieldResultExchangeItem. ---- ODA-154: Wflow: fixed bug in Wflow2DFieldStateExchangeItem.getTimes and added Wflow2DFieldResultExchangeItem. ---- ODA-154: Wflow: corrected exchange items for WflowModelInstance so that they get/set the current values and use the current time. ---- Bug fix in time stepping of noise model. ---- Undo one change of r3326: a new swan-exe was accidentally checked in ---- Specification of accuracy of time handling in main oda file Internally stored in the variabele OdaGlobSettings.getTimePrecision() Only used in AbscractSequentualAlgorithm but all time (comparison) handling should be using this user variable. Default value is 1 second (seemed to be used often already) ---- ODA-154: Wflow: corrected state handling in WflowModelInstance. The wflow model can only store one state at a time. This may be good enough for most algorithms. ---- Forgot to check in an input file a long time agowq ---- ODA-154: Wflow: added code to read the timeStepLength from the config file of the model. ---- FileSupport: improved error messages. ---- FileSupport: pass original exception when re-throwing exceptions. ---- Added documentation for new option flag_barrier ---- Fixed missing declaration. ---- Improved comments and writing of errors and warnings. ---- ODA-154: Wflow: minor improvement WflowModelInstance. ---- ODA-154: Wflow: fixed bug. Now each WflowModelInstance runs in its own instance directory. ---- Temporary directory for restart files is now removed after reading. ---- New option "continueWithoutObservation" is added to allow the steady filter to continue running even when no observation is available. This is used e.g. for forecast run, when no observation is assimilated. Syntax: . Default value=false. ---- Fixed warnings ---- ODA-154: Wflow: fixed bug where WflowModelInstance.getCurrentTime and WflowModelInstance.compute did not work correctly. ---- ODA-154: Wflow: fixed bug where wflow model did not work for case directories that start with an f. Made options for wflow model adapter configurable using runtime arguments. ---- ODA-154: Wflow: moved wflow_bin so that it can be used by multiple tests. ---- Introduction of option flag_barrier in modelclass. Syntax This option allows to run a waqua-openda computation for an arbritary number of timesteps: set parameter T_step smaller/equal than the max. period for which the wind forcings are computed. ---- ODA-154: Wflow: updated WflowModelFactory. ---- ODA-154: Wflow: removed unused input data for unit test. ---- ODA-154: Wflow: updated WflowModelInstance. ---- Correct the reference in the main configuration file swanSimulation.oda to "Simulation" algorithm instead of to "Simplex". ---- ODA-154: Added third version of WflowModelInstance. ---- Add 0.5*second to parameter selectionStart for defining (observation) selection. In this way, we make sure that data from previous analysis time is not selected. ---- Aditional option for all Sequential Algorithms. With this switch it is possible to skip the analysis step at t=0 or at t=end. This is of use for e.g. restart runs. Syntax where the two booleans skipAtInitialTime and skipAtFinalTime are new ---- ODA-156: Use absolute path for adding restart files so that it works fine with the newly adjusted FileBasedModelState.java ---- ODA-154: Added WflowPythonToJavaAdapterTest unit test. ---- ODA-154: Added second version of WflowModelInstance (this does not work yet). ---- Cancel the previous 'correction'. This problem occurs so far only for Waqua models. The correction should be somewhere in the level of model-wrapper (not in the level of algorithm). ---- ODA-154: Added first version of WflowModelInstance (this does not work yet). ---- ODA-154: fixed unit test JeppTest. Added jep.dll. Added model_wflow to public build.xml. ---- BBUtils: Add caught Exception to new Exception when re-throwing an Exception. ---- ODA-154: Added public/model_wflow folder. Added unit test to test using JEPP to run Python code from Java. ---- For the case where analysis times are not determined from the observer, the parameter selectionStart is re-defined as previous-analysis-time plus a small-threshold. This is to make sure that the observation selection takes observed data only from the current analysis time (and not from the previous analysis time). ---- Fixed bug in ioCopierTest. ---- Fixed bug in fileBasedModelState. ---- Implementatie van restart steady state Kalman filter vanuit OpenDA. ---- New correction on NoosTimeSeriesFormatter.java: this is valid for with or without space, both after and before the colon character.The example file (waterlevelObs2.txt) is modified to test this. ---- added missing file ---- Correct the way the separatorIndex is determined in NoosTimeSeriesFormatter.java. This makes sure that the header of NOOS files is read properly, with or without space after the colon character (:). An example file (waterlevelObs2.txt) is modified to test this. ---- Introduction of aliasesUsedInActualWorkingDirectory. ---- Add option to set java memory ---- Try to plug a memory leak ---- Finalize update of native build scripts. You can select your own built-type more carefully now. ---- Improve linux build scripts for native part ---- improved netcdf installation ---- added remark about installation path not containing spaces ---- Fixed bug in blackbox wrapper for external noise models. ---- Improved SimpleStochModelInstance such that it can work with more ObservationDescriptions than only TimeSeriesObservationDescriptions ---- Added parallel test configuration in model_swan BBEnKF test ---- Made serveral core classes serializable ---- 1) Swan observer is now serializable 2) New observerDescriptions implementation that can be used for non-serializable observerDescriptions (only key-value data) ---- Improved error handling on double exchange items ----