Dakota  Version 6.20
Explore and Predict with Confidence
Public Member Functions | Protected Member Functions | Protected Attributes | Private Member Functions | Private Attributes | List of all members
ApplicationInterface Class Reference

Derived class within the interface class hierarchy for supporting interfaces to simulation codes. More...

Inheritance diagram for ApplicationInterface:
Interface DirectApplicInterface ProcessApplicInterface MatlabInterface Pybind11Interface PythonInterface ScilabInterface TestDriverInterface ParallelDirectApplicInterface SerialDirectApplicInterface SoleilDirectApplicInterface ProcessHandleApplicInterface SysCallApplicInterface

Public Member Functions

 ApplicationInterface (const ProblemDescDB &problem_db)
 constructor
 
 ~ApplicationInterface ()
 destructor
 
- Public Member Functions inherited from Interface
 Interface ()
 default constructor More...
 
 Interface (ProblemDescDB &problem_db)
 standard constructor for envelope More...
 
 Interface (const Interface &interface_in)
 copy constructor More...
 
virtual ~Interface ()
 destructor
 
Interface operator= (const Interface &interface_in)
 assignment operator
 
virtual int minimum_points (bool constraint_flag) const
 returns the minimum number of points required to build a particular ApproximationInterface (used by DataFitSurrModels).
 
virtual int recommended_points (bool constraint_flag) const
 returns the recommended number of points required to build a particular ApproximationInterface (used by DataFitSurrModels).
 
virtual void active_model_key (const Pecos::ActiveKey &key)
 activate an approximation state based on its key
 
virtual void clear_model_keys ()
 reset initial state by removing all model keys for an approximation
 
virtual void approximation_function_indices (const SizetSet &approx_fn_indices)
 set the (currently active) approximation function index set
 
virtual void update_approximation (const Variables &vars, const IntResponsePair &response_pr)
 updates the anchor point for an approximation
 
virtual void update_approximation (const RealMatrix &samples, const IntResponseMap &resp_map)
 updates the current data points for an approximation
 
virtual void update_approximation (const VariablesArray &vars_array, const IntResponseMap &resp_map)
 updates the current data points for an approximation
 
virtual void append_approximation (const Variables &vars, const IntResponsePair &response_pr)
 appends a single point to an existing approximation
 
virtual void append_approximation (const RealMatrix &samples, const IntResponseMap &resp_map)
 appends multiple points to an existing approximation
 
virtual void append_approximation (const VariablesArray &vars_array, const IntResponseMap &resp_map)
 appends multiple points to an existing approximation
 
virtual void append_approximation (const IntVariablesMap &vars_map, const IntResponseMap &resp_map)
 appends multiple points to an existing approximation
 
virtual void replace_approximation (const IntResponsePair &response_pr)
 replace the response for a single point within an existing approximation
 
virtual void replace_approximation (const IntResponseMap &resp_map)
 replace responses for multiple points within an existing approximation
 
virtual void track_evaluation_ids (bool track)
 assigns trackEvalIds to activate tracking of evaluation ids within surrogate data, enabling id-based lookups for data replacement
 
virtual void build_approximation (const RealVector &c_l_bnds, const RealVector &c_u_bnds, const IntVector &di_l_bnds, const IntVector &di_u_bnds, const RealVector &dr_l_bnds, const RealVector &dr_u_bnds)
 builds the approximation
 
virtual void export_approximation ()
 export the approximation to disk
 
virtual void rebuild_approximation (const BitArray &rebuild_fns)
 rebuilds the approximation after a data update
 
virtual void pop_approximation (bool save_data)
 removes data from last append from the approximation
 
virtual void push_approximation ()
 retrieves approximation data from a previous state (negates pop)
 
virtual bool push_available ()
 queries the approximation for the ability to retrieve a previous increment
 
virtual void finalize_approximation ()
 finalizes the approximation by applying all trial increments
 
virtual void combine_approximation ()
 combine the current approximation with previously stored data sets
 
virtual void combined_to_active (bool clear_combined=true)
 promote the combined approximation to the currently active one
 
virtual void clear_inactive ()
 clear inactive approximation data
 
virtual bool advancement_available ()
 query for available advancements in approximation resolution controls
 
virtual bool formulation_updated () const
 query for change in approximation formulation
 
virtual void formulation_updated (bool update)
 assign an updated status for approximation formulation to force rebuild
 
virtual Real2DArray cv_diagnostics (const StringArray &metric_types, unsigned num_folds)
 approximation cross-validation quality metrics per response function
 
virtual RealArray challenge_diagnostics (const String &metric_type, const RealMatrix &challenge_pts)
 approximation challenge data metrics per response function
 
virtual void clear_current_active_data ()
 clears current data from an approximation interface
 
virtual void clear_active_data ()
 clears all data from an approximation interface
 
virtual SharedApproxDatashared_approximation ()
 retrieve the SharedApproxData within an ApproximationInterface
 
virtual std::vector< Approximation > & approximations ()
 retrieve the Approximations within an ApproximationInterface
 
virtual const Pecos::SurrogateData & approximation_data (size_t fn_index)
 retrieve the approximation data from a particular Approximation within an ApproximationInterface
 
virtual const RealVectorArray & approximation_coefficients (bool normalized=false)
 retrieve the approximation coefficients from each Approximation within an ApproximationInterface
 
virtual void approximation_coefficients (const RealVectorArray &approx_coeffs, bool normalized=false)
 set the approximation coefficients within each Approximation within an ApproximationInterface
 
virtual const RealVector & approximation_variances (const Variables &vars)
 retrieve the approximation variances from each Approximation within an ApproximationInterface
 
virtual const StringArray & analysis_drivers () const
 retrieve the analysis drivers specification for application interfaces
 
virtual const String2DArray & analysis_components () const
 retrieve the analysis components, if available
 
virtual void discrepancy_emulation_mode (short mode)
 set discrepancy emulation mode used for approximating response differences
 
virtual void file_cleanup () const
 clean up any interface parameter/response files when aborting
 
IntResponseMap & response_map ()
 return rawResponseMap
 
void cache_unmatched_response (int raw_id)
 migrate an unmatched response record from rawResponseMap to cachedResponseMap
 
void cache_unmatched_responses ()
 migrate all remaining response records from rawResponseMap to cachedResponseMap
 
void assign_rep (std::shared_ptr< Interface > interface_rep)
 assign letter or replace existing letter with a new one More...
 
void assign_rep (Interface *interface_rep, bool ref_count_incr=false)
 assign letter or replace existing letter with a new one DEPRECATED, but left for library mode clients to migrate: transfers memory ownership to the contained shared_ptr; ref_count_incr is ignored More...
 
unsigned short interface_type () const
 returns the interface type
 
const String & interface_id () const
 returns the interface identifier
 
int evaluation_id () const
 returns the value of the (total) evaluation id counter for the interface
 
void fine_grained_evaluation_counters (size_t num_fns)
 set fineGrainEvalCounters to true and initialize counters if needed
 
void init_evaluation_counters (size_t num_fns)
 initialize fine grained evaluation counters, sizing if needed
 
void set_evaluation_reference ()
 set evaluation count reference points for the interface
 
void print_evaluation_summary (std::ostream &s, bool minimal_header, bool relative_count) const
 print an evaluation summary for the interface
 
bool multi_proc_eval () const
 returns a flag signaling the use of multiprocessor evaluation partitions
 
bool iterator_eval_dedicated_master () const
 returns a flag signaling the use of a dedicated master processor at the iterator-evaluation scheduling level
 
bool is_null () const
 function to check interfaceRep (does this envelope contain a letter?)
 
std::shared_ptr< Interfaceinterface_rep ()
 function to return the letter
 
void eval_tag_prefix (const String &eval_id_str, bool append_iface_id=true)
 set the evaluation tag prefix (does not recurse) More...
 

Protected Member Functions

void init_communicators (const IntArray &message_lengths, int max_eval_concurrency)
 allocate communicator partitions for concurrent evaluations within an iterator and concurrent multiprocessor analyses within an evaluation.
 
void set_communicators (const IntArray &message_lengths, int max_eval_concurrency)
 set the local parallel partition data for an interface (the partitions are already allocated in ParallelLibrary).
 
void init_serial ()
 
int asynch_local_evaluation_concurrency () const
 return asynchLocalEvalConcurrency
 
short interface_synchronization () const
 return interfaceSynchronization
 
bool evaluation_cache () const
 return evalCacheFlag
 
bool restart_file () const
 return evalCacheFlag
 
String final_eval_id_tag (int fn_eval_id)
 form and return the final evaluation ID tag, appending iface ID if needed
 
void map (const Variables &vars, const ActiveSet &set, Response &response, bool asynch_flag=false)
 Provides a "mapping" of variables to responses using a simulation. Protected due to Interface letter-envelope idiom. More...
 
void manage_failure (const Variables &vars, const ActiveSet &set, Response &response, int failed_eval_id)
 manages a simulation failure using abort/retry/recover/continuation
 
const IntResponseMap & synchronize ()
 executes a blocking schedule for asynchronous evaluations in the beforeSynchCorePRPQueue and returns all jobs More...
 
const IntResponseMap & synchronize_nowait ()
 executes a nonblocking schedule for asynchronous evaluations in the beforeSynchCorePRPQueue and returns a partial set of completed jobs More...
 
void serve_evaluations ()
 run on evaluation servers to serve the iterator master More...
 
void stop_evaluation_servers ()
 used by the iterator master to terminate evaluation servers More...
 
bool check_multiprocessor_analysis (bool warn)
 checks on multiprocessor analysis configuration
 
bool check_asynchronous (bool warn, int max_eval_concurrency)
 checks on asynchronous configuration (for direct interfaces)
 
bool check_multiprocessor_asynchronous (bool warn, int max_eval_concurrency)
 checks on asynchronous settings for multiprocessor partitions
 
String final_batch_id_tag ()
 form and return the final batch ID tag
 
virtual void derived_map (const Variables &vars, const ActiveSet &set, Response &response, int fn_eval_id)
 Called by map() and other functions to execute the simulation in synchronous mode. The portion of performing an evaluation that is specific to a derived class.
 
virtual void derived_map_asynch (const ParamResponsePair &pair)
 Called by map() and other functions to execute the simulation in asynchronous mode. The portion of performing an asynchronous evaluation that is specific to a derived class.
 
virtual void wait_local_evaluations (PRPQueue &prp_queue)
 For asynchronous function evaluations, this method is used to detect completion of jobs and process their results. It provides the processing code that is specific to derived classes. This version waits for at least one completion.
 
virtual void test_local_evaluations (PRPQueue &prp_queue)
 For asynchronous function evaluations, this method is used to detect completion of jobs and process their results. It provides the processing code that is specific to derived classes. This version is nonblocking and will return without any completions if none are immediately available.
 
virtual void init_communicators_checks (int max_eval_concurrency)
 perform construct-time error checks on the parallel configuration More...
 
virtual void set_communicators_checks (int max_eval_concurrency)
 perform run-time error checks on the parallel configuration More...
 
void master_dynamic_schedule_analyses ()
 blocking dynamic schedule of all analyses within a function evaluation using message passing More...
 
void serve_analyses_synch ()
 serve the master analysis scheduler and manage one synchronous analysis job at a time More...
 
virtual int synchronous_local_analysis (int analysis_id)
 Execute a particular analysis (identified by analysis_id) synchronously on the local processor. Used for the derived class specifics within ApplicationInterface::serve_analyses_synch().
 
- Protected Member Functions inherited from Interface
 Interface (BaseConstructor, const ProblemDescDB &problem_db)
 constructor initializes the base class part of letter classes (BaseConstructor overloading avoids infinite recursion in the derived class constructors - Coplien, p. 139) More...
 
 Interface (NoDBBaseConstructor, size_t num_fns, short output_level)
 constructor initializes the base class part of letter classes (NoDBBaseConstructor used for on the fly instantiations without a DB)
 
void init_algebraic_mappings (const Variables &vars, const Response &response)
 Define algebraicACVIndices, algebraicACVIds, and algebraicFnIndices.
 
void asv_mapping (const ActiveSet &total_set, ActiveSet &algebraic_set, ActiveSet &core_set)
 define the evaluation requirements for algebraic_mappings() (algebraic_set) and the core Application/Approximation mapping (core_set) from the total Interface evaluation requirements (total_set)
 
void asv_mapping (const ActiveSet &algebraic_set, ActiveSet &total_set)
 map an algebraic ASV back to original total ordering for asynch recovery
 
void algebraic_mappings (const Variables &vars, const ActiveSet &algebraic_set, Response &algebraic_response)
 evaluate the algebraic_response using the AMPL solver library and the data extracted from the algebraic_mappings file
 
void response_mapping (const Response &algebraic_response, const Response &core_response, Response &total_response)
 combine the response from algebraic_mappings() with the response from derived_map() to create the total response More...
 

Protected Attributes

ParallelLibraryparallelLib
 reference to the ParallelLibrary object used to manage MPI partitions for the concurrent evaluations and concurrent analyses parallelism levels
 
bool batchEval
 flag indicating usage of batch evaluation facilities, where a set of jobs is launched and scheduled as a unit rather than individually
 
bool asynchFlag
 flag indicating usage of asynchronous evaluation
 
int batchIdCntr
 maintain a count of the batches
 
bool suppressOutput
 flag for suppressing output on slave processors
 
int evalCommSize
 size of evalComm
 
int evalCommRank
 processor rank within evalComm
 
int evalServerId
 evaluation server identifier
 
bool eaDedMasterFlag
 flag for dedicated master partitioning at ea level
 
int analysisCommSize
 size of analysisComm
 
int analysisCommRank
 processor rank within analysisComm
 
int analysisServerId
 analysis server identifier
 
int numAnalysisServers
 current number of analysis servers
 
bool multiProcAnalysisFlag
 flag for multiprocessor analysis partitions
 
bool asynchLocalAnalysisFlag
 flag for asynchronous local parallelism of analyses
 
int asynchLocalAnalysisConcurrency
 limits the number of concurrent analyses in asynchronous local scheduling and specifies hybrid concurrency when message passing
 
int asynchLocalEvalConcSpec
 user specification for asynchronous local evaluation concurrency
 
int asynchLocalAnalysisConcSpec
 user specification for asynchronous local analysis concurrency
 
int numAnalysisDrivers
 the number of analysis drivers used for each function evaluation (from the analysis_drivers interface specification)
 
IntSet completionSet
 the set of completed fn_eval_id's populated by wait_local_evaluations() and test_local_evaluations()
 
String failureMessage
 base message for managing failed evals; will be followed with more details in screen output
 
- Protected Attributes inherited from Interface
unsigned short interfaceType
 the interface type: enum for system, fork, direct, grid, or approximation
 
String interfaceId
 the interface specification identifier string from the DAKOTA input file
 
bool algebraicMappings
 flag for the presence of algebraic_mappings that define the subset of an Interface's parameter to response mapping that is explicit and algebraic.
 
bool coreMappings
 flag for the presence of non-algebraic mappings that define the core of an Interface's parameter to response mapping (using analysis_drivers for ApplicationInterface or functionSurfaces for ApproximationInterface).
 
short outputLevel
 output verbosity level: {SILENT,QUIET,NORMAL,VERBOSE,DEBUG}_OUTPUT
 
int currEvalId
 identifier for the current evaluation, which may differ from the evaluation counters in the case of evaluation scheduling; used on iterator master as well as server processors. Currently, this is set prior to all invocations of derived_map() for all processors.
 
bool fineGrainEvalCounters
 controls use of fn val/grad/hess counters for detailed evaluation report
 
int evalIdCntr
 total interface evaluation counter
 
int newEvalIdCntr
 new (non-duplicate) interface evaluation counter
 
int evalIdRefPt
 iteration reference point for evalIdCntr
 
int newEvalIdRefPt
 iteration reference point for newEvalIdCntr
 
IntArray fnValCounter
 number of value evaluations by resp fn
 
IntArray fnGradCounter
 number of gradient evaluations by resp fn
 
IntArray fnHessCounter
 number of Hessian evaluations by resp fn
 
IntArray newFnValCounter
 number of new value evaluations by resp fn
 
IntArray newFnGradCounter
 number of new gradient evaluations by resp fn
 
IntArray newFnHessCounter
 number of new Hessian evaluations by resp fn
 
IntArray fnValRefPt
 iteration reference point for fnValCounter
 
IntArray fnGradRefPt
 iteration reference point for fnGradCounter
 
IntArray fnHessRefPt
 iteration reference point for fnHessCounter
 
IntArray newFnValRefPt
 iteration reference point for newFnValCounter
 
IntArray newFnGradRefPt
 iteration reference point for newFnGradCounter
 
IntArray newFnHessRefPt
 iteration reference point for newFnHessCounter
 
IntResponseMap rawResponseMap
 Set of responses returned by either a blocking or nonblocking schedule. More...
 
IntResponseMap cachedResponseMap
 Set of available asynchronous responses completed within a blocking or nonblocking scheduler that cannot be processed in a higher level context and need to be stored for later.
 
StringArray fnLabels
 response function descriptors (used in print_evaluation_summary() and derived direct interface classes); initialized in map() functions due to potential updates after construction
 
bool multiProcEvalFlag
 flag for multiprocessor evaluation partitions (evalComm)
 
bool ieDedMasterFlag
 flag for dedicated master partitioning at the iterator level
 
String evalTagPrefix
 set of period-delimited evaluation ID tags to use in evaluation tagging
 
bool appendIfaceId
 whether to append the interface ID to the prefix during map (default true)
 
String2DArray analysisComponents
 Analysis components for interface types that support them.
 

Private Member Functions

bool duplication_detect (const Variables &vars, Response &response, bool asynch_flag)
 checks data_pairs and beforeSynchCorePRPQueue to see if the current evaluation request has already been performed or queued More...
 
void init_default_asv (size_t num_fns)
 initialize default ASV if needed; this is done at run time due to post-construct time Response size changes. More...
 
void master_dynamic_schedule_evaluations ()
 blocking dynamic schedule of all evaluations in beforeSynchCorePRPQueue using message passing on a dedicated master partition; executes on iteratorComm master More...
 
void peer_static_schedule_evaluations ()
 blocking static schedule of all evaluations in beforeSynchCorePRPQueue using message passing on a peer partition; executes on iteratorComm master More...
 
void peer_dynamic_schedule_evaluations ()
 blocking dynamic schedule of all evaluations in beforeSynchCorePRPQueue using message passing on a peer partition; executes on iteratorComm master More...
 
void asynchronous_local_evaluations (PRPQueue &prp_queue)
 perform all jobs in prp_queue using asynchronous approaches on the local processor More...
 
void synchronous_local_evaluations (PRPQueue &prp_queue)
 perform all jobs in prp_queue using synchronous approaches on the local processor More...
 
void master_dynamic_schedule_evaluations_nowait ()
 execute a nonblocking dynamic schedule in a master-slave partition More...
 
void peer_static_schedule_evaluations_nowait ()
 execute a nonblocking static schedule in a peer partition More...
 
void peer_dynamic_schedule_evaluations_nowait ()
 execute a nonblocking dynamic schedule in a peer partition More...
 
void asynchronous_local_evaluations_nowait (PRPQueue &prp_queue)
 launch new jobs in prp_queue asynchronously (if capacity is available), perform nonblocking query of all running jobs, and process any completed jobs (handles both local master- and local peer-scheduling cases) More...
 
void broadcast_evaluation (const ParamResponsePair &pair)
 convenience function for broadcasting an evaluation over an evalComm
 
void broadcast_evaluation (int fn_eval_id, const Variables &vars, const ActiveSet &set)
 convenience function for broadcasting an evaluation over an evalComm
 
void send_evaluation (PRPQueueIter &prp_it, size_t buff_index, int server_id, bool peer_flag)
 helper function for sending sendBuffers[buff_index] to server
 
void receive_evaluation (PRPQueueIter &prp_it, size_t buff_index, int server_id, bool peer_flag)
 helper function for processing recvBuffers[buff_index] within scheduler
 
void launch_asynch_local (PRPQueueIter &prp_it)
 launch an asynchronous local evaluation from a queue iterator
 
void launch_asynch_local (MPIUnpackBuffer &recv_buffer, int fn_eval_id)
 launch an asynchronous local evaluation from a receive buffer
 
void process_asynch_local (int fn_eval_id)
 process a completed asynchronous local evaluation
 
void process_synch_local (PRPQueueIter &prp_it)
 process a completed synchronous local evaluation
 
void assign_asynch_local_queue (PRPQueue &local_prp_queue, PRPQueueIter &local_prp_iter)
 helper function for creating an initial active local queue by launching asynch local jobs from local_prp_queue, as limited by server capacity
 
void assign_asynch_local_queue_nowait (PRPQueue &local_prp_queue, PRPQueueIter &local_prp_iter)
 helper function for updating an active local queue by backfilling asynch local jobs from local_prp_queue, as limited by server capacity
 
size_t test_local_backfill (PRPQueue &assign_queue, PRPQueueIter &assign_iter)
 helper function for testing active asynch local jobs and then backfilling
 
size_t test_receives_backfill (PRPQueueIter &assign_iter, bool peer_flag)
 helper function for testing receive requests and then backfilling jobs
 
void serve_evaluations_synch ()
 serve the evaluation message passing schedulers and perform one synchronous evaluation at a time More...
 
void serve_evaluations_synch_peer ()
 serve the evaluation message passing schedulers and perform one synchronous evaluation at a time as part of the 1st peer More...
 
void serve_evaluations_asynch ()
 serve the evaluation message passing schedulers and manage multiple asynchronous evaluations More...
 
void serve_evaluations_asynch_peer ()
 serve the evaluation message passing schedulers and perform multiple asynchronous evaluations as part of the 1st peer More...
 
void set_evaluation_communicators (const IntArray &message_lengths)
 convenience function for updating the local evaluation partition data following ParallelLibrary::init_evaluation_communicators().
 
void set_analysis_communicators ()
 convenience function for updating the local analysis partition data following ParallelLibrary::init_analysis_communicators().
 
void init_serial_evaluations ()
 set concurrent evaluation configuration for serial operations
 
void init_serial_analyses ()
 set concurrent analysis configuration for serial operations (e.g., for local executions on a dedicated master)
 
const ParamResponsePairget_source_pair (const Variables &target_vars)
 convenience function for the continuation approach in manage_failure() for finding the nearest successful "source" evaluation to the failed "target"
 
void continuation (const Variables &target_vars, const ActiveSet &set, Response &response, const ParamResponsePair &source_pair, int failed_eval_id)
 performs a 0th order continuation method to step from a successful "source" evaluation to the failed "target". Invoked by manage_failure() for failAction == "continuation".
 
void common_input_filtering (const Variables &vars)
 common input filtering operations, e.g. mesh movement
 
void common_output_filtering (Response &response)
 common output filtering operations, e.g. data filtering
 

Private Attributes

int worldSize
 size of MPI_COMM_WORLD
 
int worldRank
 processor rank within MPI_COMM_WORLD
 
int iteratorCommSize
 size of iteratorComm
 
int iteratorCommRank
 processor rank within iteratorComm
 
bool ieMessagePass
 flag for message passing at ie scheduling level
 
int numEvalServers
 current number of evaluation servers
 
int numEvalServersSpec
 user specification for number of evaluation servers
 
int procsPerEvalSpec
 user specification for processors per analysis servers
 
bool eaMessagePass
 flag for message passing at ea scheduling level
 
int numAnalysisServersSpec
 user spec for number of analysis servers
 
int procsPerAnalysisSpec
 user specification for processors per analysis servers
 
int lenVarsMessage
 length of a MPIPackBuffer containing a Variables object; computed in Model::init_communicators()
 
int lenVarsActSetMessage
 length of a MPIPackBuffer containing a Variables object and an ActiveSet object; computed in Model::init_communicators()
 
int lenResponseMessage
 length of a MPIPackBuffer containing a Response object; computed in Model::init_communicators()
 
int lenPRPairMessage
 length of a MPIPackBuffer containing a ParamResponsePair object; computed in Model::init_communicators()
 
short evalScheduling
 user specification of evaluation scheduling algorithm: {DEFAULT,MASTER,PEER_DYNAMIC,PEER_STATIC}_SCHEDULING. Used for manual overrides of auto-configure logic in ParallelLibrary::resolve_inputs().
 
short analysisScheduling
 user specification of analysis scheduling algorithm: {DEFAULT,MASTER,PEER}_SCHEDULING. Used for manual overrides of the auto-configure logic in ParallelLibrary::resolve_inputs().
 
int asynchLocalEvalConcurrency
 limits the number of concurrent evaluations in asynchronous local scheduling and specifies hybrid concurrency when message passing
 
bool asynchLocalEvalStatic
 whether the asynchronous local evaluations are to be performed with a static schedule (default false)
 
BitArray localServerAssigned
 array with one bit per logical "server" indicating whether a job is currently running on the server (used for asynch local static schedules)
 
short interfaceSynchronization
 interface synchronization specification: synchronous (default) or asynchronous
 
bool headerFlag
 used by synchronize_nowait to manage header output frequency (since this function may be called many times prior to any completions)
 
bool asvControlFlag
 used to manage a user request to deactivate the active set vector control. true = modify the ASV each evaluation as appropriate (default); false = ASV values are static so that the user need not check them on each evaluation.
 
bool evalCacheFlag
 used to manage a user request to deactivate the function evaluation cache (i.e., queries and insertions using the data_pairs cache).
 
bool nearbyDuplicateDetect
 flag indicating optional usage of tolerance-based duplication detection (less efficient, but helpful when experiencing restart cache misses)
 
Real nearbyTolerance
 tolerance value for tolerance-based duplication detection
 
bool restartFileFlag
 used to manage a user request to deactivate the restart file (i.e., insertions into write_restart).
 
SharedResponseData sharedRespData
 SharedResponseData of associated Response.
 
String gradientType
 type of gradients present in associated Response
 
String hessianType
 type of Hessians present in associated Response
 
IntSet gradMixedAnalyticIds
 IDs of analytic gradients when mixed gradients present.
 
IntSet hessMixedAnalyticIds
 IDs of analytic gradients when mixed gradients present.
 
ShortArray defaultASV
 the static ASV values used when the user has selected asvControl = off
 
String failAction
 mitigation action for captured simulation failures: abort, retry, recover, or continuation
 
int failRetryLimit
 limit on the number of retries for the retry failAction
 
RealVector failRecoveryFnVals
 the dummy function values used for the recover failAction
 
IntResponseMap historyDuplicateMap
 used to bookkeep asynchronous evaluations which duplicate data_pairs evaluations. Map key is evalIdCntr, map value is corresponding response.
 
std::map< int, std::pair< PRPQueueHIter, Response > > beforeSynchDuplicateMap
 used to bookkeep evalIdCntr, beforeSynchCorePRPQueue iterator, and response of asynchronous evaluations which duplicate queued beforeSynchCorePRPQueue evaluations
 
PRPQueue beforeSynchCorePRPQueue
 used to bookkeep vars/set/response of nonduplicate asynchronous core evaluations. This is the queue of jobs populated by asynchronous map() that is later scheduled in synchronize() or synchronize_nowait().
 
PRPQueue beforeSynchAlgPRPQueue
 used to bookkeep vars/set/response of asynchronous algebraic evaluations. This is the queue of algebraic jobs populated by asynchronous map() that is later evaluated in synchronize() or synchronize_nowait().
 
PRPQueue asynchLocalActivePRPQueue
 used by nonblocking asynchronous local schedulers to bookkeep active local jobs
 
std::map< int, IntSizetPair > msgPassRunningMap
 used by nonblocking message passing schedulers to bookkeep which jobs are running remotely
 
int nowaitEvalIdRef
 fnEvalId reference point for preserving modulo arithmetic-based job assignment in case of peer static nonblocking schedulers
 
MPIPackBuffersendBuffers
 array of pack buffers for evaluation jobs queued to a server
 
MPIUnpackBufferrecvBuffers
 array of unpack buffers for evaluation jobs returned by a server
 
MPI_Request * recvRequests
 array of requests for nonblocking evaluation receives
 

Detailed Description

Derived class within the interface class hierarchy for supporting interfaces to simulation codes.

ApplicationInterface provides an interface class for performing parameter to response mappings using simulation code(s). It provides common functionality for a number of derived classes and contains the majority of all of the scheduling algorithms in DAKOTA. The derived classes provide the specifics for managing code invocations using system calls, forks, direct procedure calls, or distributed resource facilities.

Member Function Documentation

◆ init_serial()

void init_serial ( )
inlineprotectedvirtual

DataInterface.cpp defaults of 0 servers are needed to distinguish an explicit user request for 1 server (serialization of a parallelism level) from no user request (use parallel auto-config). This default causes problems when init_communicators() is not called for an interface object (e.g., static scheduling fails in DirectApplicInterface::derived_map() for NestedModel::optionalInterface). This is the reason for this function: to reset certain defaults for interface objects that are used serially.

Reimplemented from Interface.

References ApplicationInterface::init_serial_analyses(), and ApplicationInterface::init_serial_evaluations().

◆ map()

void map ( const Variables vars,
const ActiveSet set,
Response response,
bool  asynch_flag = false 
)
protectedvirtual

Provides a "mapping" of variables to responses using a simulation. Protected due to Interface letter-envelope idiom.

The function evaluator for application interfaces. Called from derived_evaluate() and derived_evaluate_nowait() in derived Model classes. If asynch_flag is not set, perform a blocking evaluation (using derived_map()). If asynch_flag is set, add the job to the beforeSynchCorePRPQueue queue for execution by one of the scheduler routines in synchronize() or synchronize_nowait(). Duplicate function evaluations are detected with duplication_detect().

Reimplemented from Interface.

References Response::active_set(), Interface::algebraic_mappings(), Interface::algebraicMappings, Interface::asv_mapping(), ApplicationInterface::asvControlFlag, ApplicationInterface::batchEval, ApplicationInterface::beforeSynchAlgPRPQueue, ApplicationInterface::beforeSynchCorePRPQueue, ApplicationInterface::broadcast_evaluation(), Response::copy(), Interface::coreMappings, Interface::currEvalId, Dakota::data_pairs, ApplicationInterface::defaultASV, ApplicationInterface::derived_map(), ApplicationInterface::duplication_detect(), ApplicationInterface::evalCacheFlag, Interface::evalIdCntr, Interface::fineGrainEvalCounters, Interface::fnGradCounter, Interface::fnHessCounter, Interface::fnLabels, Interface::fnValCounter, Response::function_labels(), Interface::init_algebraic_mappings(), ApplicationInterface::init_default_asv(), Interface::init_evaluation_counters(), Interface::interfaceId, ApplicationInterface::manage_failure(), Interface::multiProcEvalFlag, Interface::newEvalIdCntr, Interface::newFnGradCounter, Interface::newFnHessCounter, Interface::newFnValCounter, Interface::outputLevel, ApplicationInterface::parallelLib, ActiveSet::request_vector(), Interface::response_mapping(), ApplicationInterface::restartFileFlag, ApplicationInterface::sharedRespData, and ParallelLibrary::write_restart().

◆ synchronize()

const IntResponseMap & synchronize ( )
protectedvirtual

◆ synchronize_nowait()

const IntResponseMap & synchronize_nowait ( )
protectedvirtual

◆ serve_evaluations()

void serve_evaluations ( )
protectedvirtual

◆ stop_evaluation_servers()

void stop_evaluation_servers ( )
protectedvirtual

used by the iterator master to terminate evaluation servers

This code is executed on the iteratorComm rank 0 processor when iteration on a particular model is complete. It sends a termination signal (tag = 0 instead of a valid fn_eval_id) to each of the slave analysis servers. NOTE: This function is called from the Strategy layer even when in serial mode. Therefore, use iteratorCommSize to provide appropriate fall through behavior.

Reimplemented from Interface.

References ParallelLibrary::bcast_e(), ParallelLibrary::free(), ParallelConfiguration::ie_parallel_level(), Interface::ieDedMasterFlag, ParallelLibrary::isend_ie(), ApplicationInterface::iteratorCommSize, Interface::multiProcEvalFlag, ApplicationInterface::numEvalServers, Interface::outputLevel, ParallelLibrary::parallel_configuration(), and ApplicationInterface::parallelLib.

◆ init_communicators_checks()

void init_communicators_checks ( int  max_eval_concurrency)
protectedvirtual

perform construct-time error checks on the parallel configuration

Override DirectApplicInterface definition if plug-in to allow batch processing in Plugin{Serial,Parallel}DirectApplicInterface.cpp

Reimplemented in SysCallApplicInterface, Pybind11Interface, ProcessHandleApplicInterface, and DirectApplicInterface.

Referenced by ApplicationInterface::init_communicators().

◆ set_communicators_checks()

void set_communicators_checks ( int  max_eval_concurrency)
protectedvirtual

perform run-time error checks on the parallel configuration

Override DirectApplicInterface definition if plug-in to allow batch processing in Plugin{Serial,Parallel}DirectApplicInterface.cpp

Reimplemented in SysCallApplicInterface, SoleilDirectApplicInterface, Pybind11Interface, ProcessHandleApplicInterface, SerialDirectApplicInterface, ParallelDirectApplicInterface, and DirectApplicInterface.

Referenced by ApplicationInterface::set_communicators().

◆ master_dynamic_schedule_analyses()

void master_dynamic_schedule_analyses ( )
protected

blocking dynamic schedule of all analyses within a function evaluation using message passing

This code is called from derived classes to provide the master portion of a master-slave algorithm for the dynamic scheduling of analyses among slave servers. It is patterned after master_dynamic_schedule_evaluations(). It performs no analyses locally and matches either serve_analyses_synch() or serve_analyses_asynch() on the slave servers, depending on the value of asynchLocalAnalysisConcurrency. Dynamic scheduling assigns jobs in 2 passes. The 1st pass gives each server the same number of jobs (equal to asynchLocalAnalysisConcurrency). The 2nd pass assigns the remaining jobs to slave servers as previous jobs are completed. Single- and multilevel parallel use intra- and inter-communicators, respectively, for send/receive. Specific syntax is encapsulated within ParallelLibrary.

References ApplicationInterface::asynchLocalAnalysisConcurrency, ParallelLibrary::free(), ParallelLibrary::irecv_ea(), ParallelLibrary::isend_ea(), ApplicationInterface::numAnalysisDrivers, ApplicationInterface::numAnalysisServers, ApplicationInterface::parallelLib, ParallelLibrary::waitall(), and ParallelLibrary::waitsome().

Referenced by ProcessHandleApplicInterface::create_evaluation_process(), SysCallApplicInterface::create_evaluation_process(), and DirectApplicInterface::derived_map().

◆ serve_analyses_synch()

void serve_analyses_synch ( )
protected

serve the master analysis scheduler and manage one synchronous analysis job at a time

This code is called from derived classes to run synchronous analyses on slave processors. The slaves receive requests (blocking receive), do local derived_map_ac's, and return codes. This is done continuously until a termination signal is received from the master. It is patterned after serve_evaluations_synch().

References ApplicationInterface::analysisCommRank, ParallelLibrary::bcast_a(), ParallelLibrary::isend_ea(), ApplicationInterface::multiProcAnalysisFlag, ApplicationInterface::parallelLib, ParallelLibrary::recv_ea(), ApplicationInterface::synchronous_local_analysis(), and ParallelLibrary::wait().

Referenced by ProcessHandleApplicInterface::create_evaluation_process(), SysCallApplicInterface::create_evaluation_process(), and DirectApplicInterface::derived_map().

◆ duplication_detect()

bool duplication_detect ( const Variables vars,
Response response,
bool  asynch_flag 
)
private

checks data_pairs and beforeSynchCorePRPQueue to see if the current evaluation request has already been performed or queued

Called from map() to check incoming evaluation request for duplication with content of data_pairs and beforeSynchCorePRPQueue.
If duplication is detected, return true, else return false. Manage bookkeeping with historyDuplicateMap and beforeSynchDuplicateMap. Note that the list searches can get very expensive if a long list is searched on every new function evaluation (either from a large number of previous jobs, a large number of pending jobs, or both). For this reason, a user request for deactivation of the evaluation cache results in a complete bypass of duplication_detect(), even though a beforeSynchCorePRPQueue search would still be meaningful. Since the intent of this request is to streamline operations, both list searches are bypassed.

References Response::active_set(), ApplicationInterface::beforeSynchCorePRPQueue, ApplicationInterface::beforeSynchDuplicateMap, Response::copy(), Dakota::data_pairs, ParamResponsePair::eval_id(), Interface::evalIdCntr, ApplicationInterface::historyDuplicateMap, Interface::interfaceId, Dakota::lookup_by_val(), ApplicationInterface::nearbyDuplicateDetect, ApplicationInterface::nearbyTolerance, and Response::update().

Referenced by ApplicationInterface::map().

◆ init_default_asv()

void init_default_asv ( size_t  num_fns)
private

initialize default ASV if needed; this is done at run time due to post-construct time Response size changes.

If the user has specified active_set_vector as off, then map() uses a default ASV which is constant for all function evaluations (so that the user need not check the content of the ASV on each evaluation). Only initialized if needed and not already sized.

References ApplicationInterface::asvControlFlag, ApplicationInterface::defaultASV, ApplicationInterface::gradientType, ApplicationInterface::gradMixedAnalyticIds, ApplicationInterface::hessianType, and ApplicationInterface::hessMixedAnalyticIds.

Referenced by ApplicationInterface::map().

◆ master_dynamic_schedule_evaluations()

void master_dynamic_schedule_evaluations ( )
private

blocking dynamic schedule of all evaluations in beforeSynchCorePRPQueue using message passing on a dedicated master partition; executes on iteratorComm master

This code is called from synchronize() to provide the master portion of a master-slave algorithm for the dynamic scheduling of evaluations among slave servers. It performs no evaluations locally and matches either serve_evaluations_synch() or serve_evaluations_asynch() on the slave servers, depending on the value of asynchLocalEvalConcurrency. Dynamic scheduling assigns jobs in 2 passes. The 1st pass gives each server the same number of jobs (equal to asynchLocalEvalConcurrency). The 2nd pass assigns the remaining jobs to slave servers as previous jobs are completed and returned. Single- and multilevel parallel use intra- and inter-communicators, respectively, for send/receive. Specific syntax is encapsulated within ParallelLibrary.

peer

References ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::beforeSynchCorePRPQueue, Dakota::lookup_by_eval_id(), ApplicationInterface::numEvalServers, Interface::outputLevel, ApplicationInterface::parallelLib, ApplicationInterface::receive_evaluation(), ApplicationInterface::recvBuffers, ApplicationInterface::recvRequests, ApplicationInterface::send_evaluation(), ApplicationInterface::sendBuffers, ParallelLibrary::waitall(), and ParallelLibrary::waitsome().

Referenced by ApplicationInterface::synchronize().

◆ peer_static_schedule_evaluations()

void peer_static_schedule_evaluations ( )
private

blocking static schedule of all evaluations in beforeSynchCorePRPQueue using message passing on a peer partition; executes on iteratorComm master

This code runs on the iteratorCommRank 0 processor (the iterator) and is called from synchronize() in order to manage a static schedule for cases where peer 1 must block when evaluating its local job allocation (e.g., single or multiprocessor direct interface evaluations). It matches serve_evaluations_peer() for any other processors within the first evaluation partition and serve_evaluations_{synch,asynch}() for all other evaluation partitions (depending on asynchLocalEvalConcurrency). It performs function evaluations locally for its portion of the job allocation using either asynchronous_local_evaluations() or synchronous_local_evaluations(). Single-level and multilevel parallel use intra- and inter-communicators, respectively, for send/receive. Specific syntax is encapsulated within ParallelLibrary. The iteratorCommRank 0 processor assigns the static schedule since it is the only processor with access to beforeSynchCorePRPQueue (it runs the iterator and calls synchronize). The alternate design of each peer selecting its own jobs using the modulus operator would be applicable if execution of this function (and therefore the job list) were distributed.

References ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::asynchronous_local_evaluations(), ApplicationInterface::beforeSynchCorePRPQueue, ApplicationInterface::numEvalServers, Interface::outputLevel, ApplicationInterface::parallelLib, ApplicationInterface::receive_evaluation(), ApplicationInterface::recvBuffers, ApplicationInterface::recvRequests, ApplicationInterface::send_evaluation(), ApplicationInterface::sendBuffers, ApplicationInterface::synchronous_local_evaluations(), and ParallelLibrary::waitall().

Referenced by ApplicationInterface::synchronize().

◆ peer_dynamic_schedule_evaluations()

void peer_dynamic_schedule_evaluations ( )
private

blocking dynamic schedule of all evaluations in beforeSynchCorePRPQueue using message passing on a peer partition; executes on iteratorComm master

This code runs on the iteratorCommRank 0 processor (the iterator) and is called from synchronize() in order to manage a dynamic schedule, as enabled by nonblocking management of local asynchronous jobs. It matches serve_evaluations_{synch,asynch}() for other evaluation partitions, depending on asynchLocalEvalConcurrency; it does not match serve_evaluations_peer() since, for local asynchronous jobs, the first evaluation partition cannot be multiprocessor. It performs function evaluations locally for its portion of the job allocation using asynchronous_local_evaluations_nowait(). Single-level and multilevel parallel use intra- and inter-communicators, respectively, for send/receive. Specific syntax is encapsulated within ParallelLibrary.

References ApplicationInterface::assign_asynch_local_queue(), ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::beforeSynchCorePRPQueue, ApplicationInterface::msgPassRunningMap, ApplicationInterface::numEvalServers, Interface::outputLevel, ApplicationInterface::recvBuffers, ApplicationInterface::recvRequests, ApplicationInterface::send_evaluation(), ApplicationInterface::sendBuffers, ApplicationInterface::test_local_backfill(), and ApplicationInterface::test_receives_backfill().

Referenced by ApplicationInterface::synchronize().

◆ asynchronous_local_evaluations()

void asynchronous_local_evaluations ( PRPQueue &  local_prp_queue)
private

perform all jobs in prp_queue using asynchronous approaches on the local processor

This function provides blocking synchronization for the local asynch case (background system call, nonblocking fork, or threads). It can be called from synchronize() for a complete local scheduling of all asynchronous jobs or from peer_{static,dynamic}_schedule_evaluations() to perform a local portion of the total job set. It uses derived_map_asynch() to initiate asynchronous evaluations and wait_local_evaluations() to capture completed jobs, and mirrors the master_dynamic_schedule_evaluations() message passing scheduler as much as possible (wait_local_evaluations() is modeled after MPI_Waitsome()).

References ApplicationInterface::assign_asynch_local_queue(), ApplicationInterface::asynchLocalActivePRPQueue, ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::asynchLocalEvalStatic, ApplicationInterface::batchEval, ApplicationInterface::completionSet, ApplicationInterface::launch_asynch_local(), ApplicationInterface::localServerAssigned, Dakota::lookup_by_eval_id(), ApplicationInterface::numEvalServers, Interface::outputLevel, ApplicationInterface::process_asynch_local(), Interface::rawResponseMap, and ApplicationInterface::wait_local_evaluations().

Referenced by ApplicationInterface::peer_static_schedule_evaluations(), and ApplicationInterface::synchronize().

◆ synchronous_local_evaluations()

void synchronous_local_evaluations ( PRPQueue &  local_prp_queue)
private

perform all jobs in prp_queue using synchronous approaches on the local processor

This function provides blocking synchronization for the local synchronous case (foreground system call, blocking fork, or procedure call from derived_map()). It is called from peer_static_schedule_evaluations() to perform a local portion of the total job set.

References ApplicationInterface::broadcast_evaluation(), Interface::currEvalId, ApplicationInterface::derived_map(), ApplicationInterface::manage_failure(), Interface::multiProcEvalFlag, and ApplicationInterface::process_synch_local().

Referenced by ApplicationInterface::peer_static_schedule_evaluations(), and ApplicationInterface::peer_static_schedule_evaluations_nowait().

◆ master_dynamic_schedule_evaluations_nowait()

void master_dynamic_schedule_evaluations_nowait ( )
private

execute a nonblocking dynamic schedule in a master-slave partition

This code is called from synchronize_nowait() to provide the master portion of a nonblocking master-slave algorithm for the dynamic scheduling of evaluations among slave servers. It performs no evaluations locally and matches either serve_evaluations_synch() or serve_evaluations_asynch() on the slave servers, depending on the value of asynchLocalEvalConcurrency. Dynamic scheduling assigns jobs in 2 passes. The 1st pass gives each server the same number of jobs (equal to asynchLocalEvalConcurrency). The 2nd pass assigns the remaining jobs to slave servers as previous jobs are completed. Single- and multilevel parallel use intra- and inter-communicators, respectively, for send/receive. Specific syntax is encapsulated within ParallelLibrary.

References Dakota::abort_handler(), ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::beforeSynchCorePRPQueue, ApplicationInterface::headerFlag, ApplicationInterface::msgPassRunningMap, ApplicationInterface::numEvalServers, ApplicationInterface::recvBuffers, ApplicationInterface::recvRequests, ApplicationInterface::send_evaluation(), ApplicationInterface::sendBuffers, and ApplicationInterface::test_receives_backfill().

Referenced by ApplicationInterface::synchronize_nowait().

◆ peer_static_schedule_evaluations_nowait()

void peer_static_schedule_evaluations_nowait ( )
private

execute a nonblocking static schedule in a peer partition

This code runs on the iteratorCommRank 0 processor (the iterator) and is called from synchronize_nowait() in order to manage a nonblocking static schedule. It matches serve_evaluations_synch() for other evaluation partitions (asynchLocalEvalConcurrency == 1). It performs blocking local function evaluations, one at a time, for its portion of the static schedule and checks for remote completions in between each local completion. Therefore, unlike peer_dynamic_schedule_evaluations_nowait(), this scheduler will always return at least one job. Single-level and multilevel parallel use intra- and inter-communicators, respectively, for send/receive, with specific syntax as encapsulated within ParallelLibrary. The iteratorCommRank 0 processor assigns the static schedule since it is the only processor with access to beforeSynchCorePRPQueue (it runs the iterator and calls synchronize). The alternate design of each peer selecting its own jobs using the modulus operator would be applicable if execution of this function (and therefore the job list) were distributed.

References Dakota::abort_handler(), ApplicationInterface::assign_asynch_local_queue(), ApplicationInterface::assign_asynch_local_queue_nowait(), ApplicationInterface::asynchLocalActivePRPQueue, ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::beforeSynchCorePRPQueue, ApplicationInterface::headerFlag, Interface::interfaceType, Dakota::lookup_by_eval_id(), ApplicationInterface::msgPassRunningMap, Interface::multiProcEvalFlag, ApplicationInterface::nowaitEvalIdRef, ApplicationInterface::numEvalServers, ApplicationInterface::recvBuffers, ApplicationInterface::recvRequests, ApplicationInterface::send_evaluation(), ApplicationInterface::sendBuffers, ApplicationInterface::synchronous_local_evaluations(), ApplicationInterface::test_local_backfill(), and ApplicationInterface::test_receives_backfill().

Referenced by ApplicationInterface::synchronize_nowait().

◆ peer_dynamic_schedule_evaluations_nowait()

void peer_dynamic_schedule_evaluations_nowait ( )
private

execute a nonblocking dynamic schedule in a peer partition

This code runs on the iteratorCommRank 0 processor (the iterator) and is called from synchronize_nowait() in order to manage a nonblocking static schedule. It matches serve_evaluations_{synch,asynch}() for other evaluation partitions (depending on asynchLocalEvalConcurrency). It performs nonblocking local function evaluations for its portion of the static schedule using asynchronous_local_evaluations(). Single-level and multilevel parallel use intra- and inter-communicators, respectively, for send/receive, with specific syntax as encapsulated within ParallelLibrary. The iteratorCommRank 0 processor assigns the dynamic schedule since it is the only processor with access to beforeSynchCorePRPQueue (it runs the iterator and calls synchronize). The alternate design of each peer selecting its own jobs using the modulus operator would be applicable if execution of this function (and therefore the job list) were distributed.

References Dakota::abort_handler(), ApplicationInterface::assign_asynch_local_queue(), ApplicationInterface::assign_asynch_local_queue_nowait(), ApplicationInterface::asynchLocalActivePRPQueue, ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::beforeSynchCorePRPQueue, ApplicationInterface::headerFlag, Dakota::lookup_by_eval_id(), ApplicationInterface::msgPassRunningMap, ApplicationInterface::numEvalServers, ApplicationInterface::recvBuffers, ApplicationInterface::recvRequests, ApplicationInterface::send_evaluation(), ApplicationInterface::sendBuffers, ApplicationInterface::test_local_backfill(), and ApplicationInterface::test_receives_backfill().

Referenced by ApplicationInterface::synchronize_nowait().

◆ asynchronous_local_evaluations_nowait()

void asynchronous_local_evaluations_nowait ( PRPQueue &  local_prp_queue)
private

launch new jobs in prp_queue asynchronously (if capacity is available), perform nonblocking query of all running jobs, and process any completed jobs (handles both local master- and local peer-scheduling cases)

This function provides nonblocking synchronization for the local asynch case (background system call, nonblocking fork, or threads). It is called from synchronize_nowait() and passed the complete set of all asynchronous jobs (beforeSynchCorePRPQueue). It uses derived_map_asynch() to initiate asynchronous evaluations and test_local_evaluations() to capture completed jobs in nonblocking mode. It mirrors a nonblocking message passing scheduler as much as possible (test_local_evaluations() modeled after MPI_Testsome()). The result of this function is rawResponseMap, which uses eval_id as a key. It is assumed that the incoming local_prp_queue contains only active and new jobs - i.e., all completed jobs are cleared by synchronize_nowait().

Also supports asynchronous local evaluations with static scheduling. This scheduling policy specifically ensures that a completed asynchronous evaluation eval_id is replaced with an equivalent one, modulo asynchLocalEvalConcurrency. In the nowait case, this could render some servers idle if evaluations don't come in eval_id order or some evaluations are cancelled by the caller in between calls. If this function is called with unlimited local eval concurrency, the static scheduling request is ignored.

References ApplicationInterface::assign_asynch_local_queue_nowait(), ApplicationInterface::asynchLocalActivePRPQueue, ApplicationInterface::asynchLocalEvalConcurrency, ApplicationInterface::asynchLocalEvalStatic, ApplicationInterface::headerFlag, and ApplicationInterface::test_local_backfill().

Referenced by ApplicationInterface::synchronize_nowait().

◆ serve_evaluations_synch()

void serve_evaluations_synch ( )
private

serve the evaluation message passing schedulers and perform one synchronous evaluation at a time

This code is invoked by serve_evaluations() to perform one synchronous job at a time on each slave/peer server. The servers receive requests (blocking receive), do local synchronous maps, and return results. This is done continuously until a termination signal is received from the master (sent via stop_evaluation_servers()).

References ParallelLibrary::bcast_e(), Interface::currEvalId, ApplicationInterface::derived_map(), ApplicationInterface::evalCommRank, ParallelLibrary::isend_ie(), ApplicationInterface::lenResponseMessage, ApplicationInterface::lenVarsActSetMessage, ApplicationInterface::manage_failure(), Interface::multiProcEvalFlag, ApplicationInterface::parallelLib, ParallelLibrary::recv_ie(), MPIPackBuffer::reset(), ApplicationInterface::sharedRespData, and ParallelLibrary::wait().

Referenced by ApplicationInterface::serve_evaluations().

◆ serve_evaluations_synch_peer()

void serve_evaluations_synch_peer ( )
private

serve the evaluation message passing schedulers and perform one synchronous evaluation at a time as part of the 1st peer

This code is invoked by serve_evaluations() to perform a synchronous evaluation in coordination with the iteratorCommRank 0 processor (the iterator) for static schedules. The bcast() matches either the bcast() in synchronous_local_evaluations(), which is invoked by peer_static_schedule_evaluations(), or the bcast() in map().

References ParallelLibrary::bcast_e(), Interface::currEvalId, ApplicationInterface::derived_map(), ApplicationInterface::lenVarsActSetMessage, ApplicationInterface::manage_failure(), ApplicationInterface::parallelLib, and ApplicationInterface::sharedRespData.

Referenced by ApplicationInterface::serve_evaluations().

◆ serve_evaluations_asynch()

void serve_evaluations_asynch ( )
private

serve the evaluation message passing schedulers and manage multiple asynchronous evaluations

This code is invoked by serve_evaluations() to perform multiple asynchronous jobs on each slave/peer server. The servers test for any incoming jobs, launch any new jobs, process any completed jobs, and return any results. Each of these components is nonblocking, although the server loop continues until a termination signal is received from the master (sent via stop_evaluation_servers()). In the master-slave case, the master maintains the correct number of jobs on each slave. In the static scheduling case, each server is responsible for limiting concurrency (since the entire static schedule is sent to the peers at start up).

References Dakota::abort_handler(), ApplicationInterface::asynchLocalActivePRPQueue, ApplicationInterface::asynchLocalEvalConcurrency, ParallelLibrary::bcast_e(), ApplicationInterface::completionSet, ApplicationInterface::evalCommRank, ParallelLibrary::irecv_ie(), ApplicationInterface::launch_asynch_local(), ApplicationInterface::lenResponseMessage, ApplicationInterface::lenVarsActSetMessage, Dakota::lookup_by_eval_id(), Interface::multiProcEvalFlag, ApplicationInterface::parallelLib, ParallelLibrary::recv_ie(), ParallelLibrary::send_ie(), ParallelLibrary::test(), and ApplicationInterface::test_local_evaluations().

Referenced by ApplicationInterface::serve_evaluations().

◆ serve_evaluations_asynch_peer()

void serve_evaluations_asynch_peer ( )
private

The documentation for this class was generated from the following files: