#59 used parameters #60

Yuri05 · 2019-02-05T18:09:15Z

No description provided.

…uget

Yuri05 · 2019-02-05T18:25:35Z

src/OSPSuite.SimModel/include/SimModel/SimulationTask.h

@@ -29,6 +29,9 @@ class SimulationTask
 	static void SetValuesBelowAbsTolLevelToZero(double * values, int valuesSize, double absTol);

 	static void CheckForNegativeValues(Species ** odeVariables, int numberOfVariables, double absTol);
+
+	//mark all parameters which are used in any ODE varaible or observer as unused
+	static void MarkUsedParameters(Simulation * sim);


That's the main routine, which is called between Simulation.Load and Simulation.Finalize.
It sets UsedInSimulation flag for every simulation parameter

Yuri05 · 2019-02-05T18:29:20Z

src/OSPSuite.SimModel/src/FormulaChange.cpp

 {
 	// if NOT usedAsValue it is contained in switch statement
-	if (!_useAsValue)
+	if (!_useAsValue && !alwaysAppend)


@hdiedam Not sure why it is used like this for the model CPP-Export.
Thus I've defined an optional parameter alwaysAppend, which is false by default (:arrow_right: CPP Export remains AS IS).
For use cases described in Open-Systems-Pharmacology/PK-Sim#1050 parameters must be always attached

Yuri05 · 2019-02-05T18:31:50Z

src/OSPSuite.SimModel/src/SimulationOptions.cpp

@@ -15,13 +15,14 @@ using namespace std;

 SimulationOptions::SimulationOptions(void)
 {
-	_validateXMLWithSchema = true;
+	_validateXMLWithSchema = false; //perform schema validation only if explicitly set by user


this has nothing to do with used parameters.
Just changed the default from true (="validate XML with SimModel-Schema") to false. Just to increase the performance.
PK-Sim/MoBi always set it to false (assuming models generated there are always valid :)
Matlab/R-Toolbox still can set it to true if required

Yuri05 · 2019-02-05T18:33:13Z

src/OSPSuite.SimModel/src/SimulationOptions.cpp

 	_showProgress = false;
 	_executionTimeLimit = 0.0;    //no limit of simulation execution time
 	_stopOnWarnings = true;       //treat warnings of ODE solver as errors
 	_autoReduceTolerances = true; //automatically reduce solver tolerances
 	_writeLogFile = true;
 	_checkForNegativeValues = true;
+	_identifyUsedParameters = false;


Not sure if we need the flag or if used parameters should be ALWAYS estimated.
Currently there is a posiibility to skip this step (in this case, ALL parameters will be set as used)

Yuri05 · 2019-02-05T18:34:07Z

src/OSPSuite.SimModel/src/SimulationTask.cpp

+	TObjectList<Parameter> & parameters = sim->Parameters();
+	int idx;
+
+	if (!sim->Options().IdentifyUsedParameters())


s. comment regarding this option above.

Yuri05 · 2019-02-05T18:35:40Z

src/OSPSuite.SimModel/src/SimulationTask.cpp

@@ -222,6 +222,54 @@ void SimulationTask::CheckForNegativeValues(Species ** odeVariables, int numberO

 }

+void SimulationTask::MarkUsedParameters(Simulation * sim)


Main routine. MArks all parameters as used, which are used in:

ODE Variables (RHS or initial formula)

Observers

Event conditions

Event assignment formulas

Yuri05 · 2019-02-05T18:37:20Z

src/OSPSuite.SimModel/src/Species.cpp

@@ -448,9 +448,14 @@ void Species::SetODEIndex(int newIndex)
 	m_ODEIndex = newIndex;
 }

-void Species::AppendUsedParameters(std::set<int> & usedParameterIDs) {
+void Species::AppendUsedParameters(std::set<int> & usedParameterIDs, bool includeInitialFormula) 


@hdiedam Not sure why initial formula is skipped when getting used parameters for CPP-Export.
Thus I added optional parameter includeInitialFormula (false per default, so CPP-Export is not affected)

The cppExport uses that function primarily to distinguish between "local" parameters (switched or somehow depending on y and t) and "global" parameters (depending only on free parameters) that can be precomputed. If they are depending only on some initial value y(t_0) and not on the state trajectory y(t) they still can be considered global.

Yuri05 · 2019-02-05T18:37:59Z

src/OSPSuite.SimModel/src/Switch.cpp

@@ -178,14 +178,14 @@ void Switch::AppendUsedVariables(set<int> & usedVariblesIndices)
 		_formulaChangeVector[i]->AppendVariablesUsedInNewFormula(usedVariblesIndices, emptySet);
 }

-void Switch::AppendUsedParameters(std::set<int> & usedParameterIDs)
+void Switch::AppendUsedParameters(std::set<int> & usedParameterIDs, bool alwaysAppendInFormulaChange)


s. comment in FormulaChange above

Yuri05 · 2019-02-05T18:39:20Z

src/OSPSuite.SimModelComp/OSPSuite_SimModelComp.xml

@@ -21,6 +21,14 @@
 						<FieldDef Name="ExecutionTimeLimit" DataType="Double" Description="Execution time limit for single simulation run (0= no limit)" DefaultValue="0"/>
 						<Values>0</Values>
 					</Variable>
+					<Variable Key="ValidateWithXMLSchema">


Configuration file for Matlab/R-Toolbox extend with possibilities to

enable/disable validation with SimModel-Schema

enable/disable identifying used parameters (if false: ALL parameters will be marked as used)

Yuri05 · 2019-02-05T18:39:44Z

src/OSPSuite.SimModelComp/include/SimModelComp/SimModelComp.h

@@ -61,6 +61,9 @@ class SimModelComp : public DCI::Component
 	bool m_ProcessDataFirstRun;

 	void LoadConfigurationFromParameterTable();
+	bool GetBooleanValueFrom(DCI::ITableHandle & hTab, const std::string & columnName, bool defaultValue);
+	double GetDoubleValueFrom(DCI::ITableHandle & hTab, const std::string & columnName, double defaultValue);
+	std::string GetStringValueFrom(DCI::ITableHandle & hTab, const std::string & columnName, const std::string & defaultValue="");


just some refactoring

Yuri05 · 2019-02-05T18:42:31Z

tests/TestData/TestExportUsedParameters02.xml

@@ -0,0 +1,225 @@
+<?xml version="1.0" encoding="utf-8"?>


Simulation used in tests, exported from TestExportUsedParameters01.mbp3

msevestre · 2019-02-05T18:50:17Z

src/OSPSuite.SimModel/include/SimModel/SimulationOptions.h

@@ -22,6 +22,8 @@ namespace SimModelNative
 		bool _keepXMLNodeAsString; //original xml is required only for saving the simulation to XML
 		bool _useFloatComparisonInUserOutputTimePoints; //if set to true, float comparison will be used
 		                                                //for user output time points.Otherwise: double
+		bool _identifyUsedParameters; //if set to false: ALL parameters will be marked as used in ODE variables/observes


is there a reason why comments are not above the variable? Kind of unusual

No arguable reason :)

msevestre · 2019-02-05T18:53:58Z

src/OSPSuite.SimModel/include/SimModel/FormulaChange.h

@@ -55,7 +55,7 @@ class FormulaChange :

 	//used variables are only added into the list if UseAsValue=false
 	void AppendVariablesUsedInNewFormula(std::set<int> & usedVariblesIndices, const std::set<int> & variblesIndicesUsedInSwitchAssignments);
-	void AppendUsedParameters(std::set<int> & usedParameterIDs);
+	void AppendUsedParameters(std::set<int> & usedParameterIDs, bool alwaysAppend);


don't understand what this parameter means alwaysAppend

Is explained here: #60 (review)

msevestre · 2019-02-05T18:54:58Z

src/OSPSuite.SimModel/include/SimModel/Species.h

@@ -100,7 +100,7 @@ class Species :

 	void SetODEIndex(int newIndex);

-	virtual void AppendUsedParameters(std::set<int> & usedParameterIDs);
+	virtual void AppendUsedParameters(std::set<int> & usedParameterIDs, bool includeInitialFormula=false);


What does this parameter mean includeInitialFormula
Sometime this extra parameter has a default value (here false) sometimes not (see FormulaChange above). Why?

Is explained here: #60 (review)
The reason is: Species.AppendUsedParameters is called "on the top level"
a) from the current use case and
b) from the Model-CPP-Export).
As I didn't want to change b), I defined this parameter as optional.

FormulaChange.AppendUsedParameters is called only by it's "parent" (Switch.AppendUsedParameters), thus here there is no need for optional paramter

Yuri05 · 2019-02-07T16:26:16Z

Merging to continue.
@hdiedam you can still review

Yuri05 added 5 commits May 29, 2018 23:58

Fixes #30 packages.config references nuget packages that are not on n…

ee870d6

…uget

Merge branch 'master' of https://github.com/Yuri05/OSPSuite.SimModel

124fc06

SimModel

66f0384

SimModelComp (without test)

dde6f7c

Fixes #59 Provide list of all effectively used parameters

e49cdf8

Yuri05 requested a review from hdiedam February 5, 2019 18:09

Fixes #59 Provide list of all effectively used parameters

fa6f378

Yuri05 commented Feb 5, 2019

View reviewed changes

msevestre reviewed Feb 5, 2019

View reviewed changes

Yuri05 merged commit dd253e7 into Open-Systems-Pharmacology:master Feb 7, 2019

Yuri05 deleted the #59_used_parameters branch February 15, 2019 08:47

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

#59 used parameters #60

#59 used parameters #60

Yuri05 commented Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

hdiedam Feb 7, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

Yuri05 Feb 5, 2019

msevestre Feb 5, 2019

Yuri05 Feb 6, 2019

msevestre Feb 5, 2019

Yuri05 Feb 6, 2019

msevestre Feb 5, 2019

Yuri05 Feb 6, 2019

Yuri05 commented Feb 7, 2019

		@@ -222,6 +222,54 @@ void SimulationTask::CheckForNegativeValues(Species ** odeVariables, int numberO

		}

		void SimulationTask::MarkUsedParameters(Simulation * sim)

#59 used parameters #60

#59 used parameters #60

Conversation

Yuri05 commented Feb 5, 2019

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Yuri05 commented Feb 7, 2019