11 #ifndef OOPS_BASE_LINEARMODELBASE_H_ 12 #define OOPS_BASE_LINEARMODELBASE_H_ 17 #include <boost/noncopyable.hpp> 18 #include <boost/scoped_ptr.hpp> 25 #include "oops/util/abor1_cpp.h" 26 #include "oops/util/Duration.h" 27 #include "oops/util/Logger.h" 28 #include "oops/util/ObjectCounter.h" 29 #include "oops/util/Printable.h" 30 #include "oops/util/Timer.h" 45 template <
typename MODEL>
47 private boost::noncopyable {
55 static const std::string
classname() {
return "oops::LinearModelBase";}
79 virtual void setTrajectory(
const typename MODEL::State &,
typename MODEL::State &,
80 const typename MODEL::ModelAuxControl &) = 0;
83 virtual void initializeTL(
typename MODEL::Increment &)
const = 0;
84 virtual void stepTL(
typename MODEL::Increment &,
85 const typename MODEL::ModelAuxIncrement &)
const = 0;
86 virtual void finalizeTL(
typename MODEL::Increment &)
const = 0;
89 virtual void initializeAD(
typename MODEL::Increment &)
const = 0;
90 virtual void stepAD(
typename MODEL::Increment &,
typename MODEL::ModelAuxIncrement &)
const = 0;
91 virtual void finalizeAD(
typename MODEL::Increment &)
const = 0;
94 virtual void print(std::ostream &)
const = 0;
100 template <
typename MODEL>
110 static std::map < std::string, LinearModelFactory<MODEL> * > &
getMakers() {
111 static std::map < std::string, LinearModelFactory<MODEL> * > makers_;
118 template<
class MODEL,
class T>
129 template <
typename MODEL>
131 if (getMakers().find(
name) != getMakers().
end()) {
132 Log::error() <<
name <<
" already registered in the tangent linear model factory." << std::endl;
133 ABORT(
"Element already registered in LinearModelFactory.");
135 getMakers()[
name] =
this;
140 template <
typename MODEL>
142 const eckit::Configuration &
conf) {
143 Log::trace() <<
"LinearModelBase<MODEL>::create starting" << std::endl;
144 const std::string
id =
conf.getString(
"version");
145 typename std::map<std::string, LinearModelFactory<MODEL>*>::iterator
146 jerr = getMakers().find(
id);
147 if (jerr == getMakers().
end()) {
148 Log::error() <<
id <<
" does not exist in the tangent linear model factory." << std::endl;
149 Log::error() <<
"Factory contains " << getMakers().size() <<
" elements:" << std::endl;
151 jj = getMakers().
begin(); jj != getMakers().end(); ++jj) {
152 Log::error() <<
"A " << jj->first <<
" linear model" << std::endl;
154 ABORT(
"Element does not exist in LinearModelFactory.");
157 Log::trace() <<
"LinearModelBase<MODEL>::create done" << std::endl;
163 template<
typename MODEL>
166 Log::trace() <<
"LinearModelBase<MODEL>::setTrajectory starting" << std::endl;
167 util::Timer timer(classname(),
"setTrajectory");
169 Log::trace() <<
"LinearModelBase<MODEL>::setTrajectory done" << std::endl;
174 template<
typename MODEL>
176 Log::trace() <<
"LinearModelBase<MODEL>::initializeTL starting" << std::endl;
177 util::Timer timer(classname(),
"initializeTL");
179 Log::trace() <<
"LinearModelBase<MODEL>::initializeTL done" << std::endl;
184 template<
typename MODEL>
186 Log::trace() <<
"LinearModelBase<MODEL>::stepTL starting" << std::endl;
187 util::Timer timer(classname(),
"stepTL");
189 Log::trace() <<
"LinearModelBase<MODEL>::stepTL done" << std::endl;
194 template<
typename MODEL>
196 Log::trace() <<
"LinearModelBase<MODEL>::finalizeTL starting" << std::endl;
197 util::Timer timer(classname(),
"finalizeTL");
199 Log::trace() <<
"LinearModelBase<MODEL>::finalizeTL done" << std::endl;
204 template<
typename MODEL>
206 Log::trace() <<
"LinearModelBase<MODEL>::initializeAD starting" << std::endl;
207 util::Timer timer(classname(),
"initializeAD");
209 Log::trace() <<
"LinearModelBase<MODEL>::initializeAD done" << std::endl;
214 template<
typename MODEL>
216 Log::trace() <<
"LinearModelBase<MODEL>::stepAD starting" << std::endl;
217 util::Timer timer(classname(),
"stepAD");
219 Log::trace() <<
"LinearModelBase<MODEL>::stepAD done" << std::endl;
224 template<
typename MODEL>
226 Log::trace() <<
"LinearModelBase<MODEL>::finalizeAD starting" << std::endl;
227 util::Timer timer(classname(),
"finalizeAD");
229 Log::trace() <<
"LinearModelBase<MODEL>::finalizeAD done" << std::endl;
236 #endif // OOPS_BASE_LINEARMODELBASE_H_ void stepAD(Increment_ &, ModelAuxIncr_ &) const
Geometry< MODEL > Geometry_
virtual LinearModelBase< MODEL > * make(const Geometry_ &, const eckit::Configuration &)=0
virtual const util::Duration & timeResolution() const =0
virtual ~LinearModelFactory()
************************************************************************GNU Lesser General Public License **This file is part of the GFDL Flexible Modeling System(FMS). ! *! *FMS is free software without even the implied warranty of MERCHANTABILITY or *FITNESS FOR A PARTICULAR PURPOSE See the GNU General Public License *for more details **You should have received a copy of the GNU Lesser General Public *License along with FMS If see< http:! ***********************************************************************!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !! MPP_TRANSMIT !! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! subroutine MPP_TRANSMIT_(put_data, put_len, to_pe, get_data, get_len, from_pe, block, tag, recv_request, send_request)!a message-passing routine intended to be reminiscent equally of both MPI and SHMEM!put_data and get_data are contiguous MPP_TYPE_ arrays!at each call, your put_data array is put to to_pe 's get_data! your get_data array is got from from_pe 's put_data!i.e we assume that typically(e.g updating halo regions) each PE performs a put _and_ a get!special PE designations:! NULL_PE:to disable a put or a get(e.g at boundaries)! ANY_PE:if remote PE for the put or get is to be unspecific! ALL_PES:broadcast and collect operations(collect not yet implemented)!ideally we would not pass length, but this f77-style call performs better(arrays passed by address, not descriptor)!further, this permits< length > contiguous words from an array of any rank to be passed(avoiding f90 rank conformance check)!caller is responsible for completion checks(mpp_sync_self) before and after integer, intent(in) ::put_len, to_pe, get_len, from_pe MPP_TYPE_, intent(in) ::put_data(*) MPP_TYPE_, intent(out) ::get_data(*) logical, intent(in), optional ::block integer, intent(in), optional ::tag integer, intent(out), optional ::recv_request, send_request logical ::block_comm integer ::i MPP_TYPE_, allocatable, save ::local_data(:) !local copy used by non-parallel code(no SHMEM or MPI) integer ::comm_tag integer ::rsize if(.NOT.module_is_initialized) call mpp_error(FATAL, 'MPP_TRANSMIT:You must first call mpp_init.') if(to_pe.EQ.NULL_PE .AND. from_pe.EQ.NULL_PE) return block_comm=.true. if(PRESENT(block)) block_comm=block if(debug) then call SYSTEM_CLOCK(tick) write(stdout_unit,'(a, i18, a, i6, a, 2i6, 2i8)')&'T=', tick, ' PE=', pe, ' MPP_TRANSMIT begin:to_pe, from_pe, put_len, get_len=', to_pe, from_pe, put_len, get_len end if comm_tag=DEFAULT_TAG if(present(tag)) comm_tag=tag!do put first and then get if(to_pe.GE.0 .AND. to_pe.LT.npes) then!use non-blocking sends if(debug .and.(current_clock.NE.0)) call SYSTEM_CLOCK(start_tick)!z1l:truly non-blocking send.! if(request(to_pe).NE.MPI_REQUEST_NULL) then !only one message from pe-> to_pe in queue *PE waiting for to_pe ! call error else get_len so only do gets but you cannot have a pure get with MPI call a get means do a wait to ensure put on remote PE is complete error call increase mpp_nml request_multiply call MPP_TRANSMIT end
Increment< MODEL > Increment_
LinearModelFactory(const std::string &)
static LinearModelBase< MODEL > * create(const Geometry_ &, const eckit::Configuration &)
Encapsulates the model state.
State_ & state()
Interfacing.
The namespace for the main oops code.
Geometry< MODEL > Geometry_
void setTrajectory(const State_ &, State_ &, const ModelAux_ &)
************************************************************************GNU Lesser General Public License **This file is part of the GFDL Flexible Modeling System(FMS). ! *! *FMS is free software without even the implied warranty of MERCHANTABILITY or *FITNESS FOR A PARTICULAR PURPOSE See the GNU General Public License *for more details **You should have received a copy of the GNU Lesser General Public *License along with FMS If see< http:! ***********************************************************************!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !! MPP_TRANSMIT !! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! subroutine MPP_TRANSMIT_(put_data, put_len, to_pe, get_data, get_len, from_pe, block, tag, recv_request, send_request)!a message-passing routine intended to be reminiscent equally of both MPI and SHMEM!put_data and get_data are contiguous MPP_TYPE_ arrays!at each call, your put_data array is put to to_pe 's get_data! your get_data array is got from from_pe 's put_data!i.e we assume that typically(e.g updating halo regions) each PE performs a put _and_ a get!special PE designations:! NULL_PE:to disable a put or a get(e.g at boundaries)! ANY_PE:if remote PE for the put or get is to be unspecific! ALL_PES:broadcast and collect operations(collect not yet implemented)!ideally we would not pass length, but this f77-style call performs better(arrays passed by address, not descriptor)!further, this permits< length > contiguous words from an array of any rank to be passed(avoiding f90 rank conformance check)!caller is responsible for completion checks(mpp_sync_self) before and after integer, intent(in) ::put_len, to_pe, get_len, from_pe MPP_TYPE_, intent(in) ::put_data(*) MPP_TYPE_, intent(out) ::get_data(*) logical, intent(in), optional ::block integer, intent(in), optional ::tag integer, intent(out), optional ::recv_request, send_request logical ::block_comm integer ::i MPP_TYPE_, allocatable, save ::local_data(:) !local copy used by non-parallel code(no SHMEM or MPI) integer ::comm_tag integer ::rsize if(.NOT.module_is_initialized) call mpp_error(FATAL, 'MPP_TRANSMIT:You must first call mpp_init.') if(to_pe.EQ.NULL_PE .AND. from_pe.EQ.NULL_PE) return block_comm=.true. if(PRESENT(block)) block_comm=block if(debug) then call SYSTEM_CLOCK(tick) write(stdout_unit,'(a, i18, a, i6, a, 2i6, 2i8)')&'T=', tick, ' PE=', pe, ' MPP_TRANSMIT begin:to_pe, from_pe, put_len, get_len=', to_pe, from_pe, put_len, get_len end if comm_tag=DEFAULT_TAG if(present(tag)) comm_tag=tag!do put first and then get if(to_pe.GE.0 .AND. to_pe.LT.npes) then!use non-blocking sends if(debug .and.(current_clock.NE.0)) call SYSTEM_CLOCK(start_tick)!z1l:truly non-blocking send.! if(request(to_pe).NE.MPI_REQUEST_NULL) then !only one message from pe-> to_pe in queue *PE waiting for to_pe ! call error else get_len so only do gets but you cannot have a pure get with MPI call a get means do a wait to ensure put on remote PE is complete error call increase mpp_nml request_multiply call MPP_TRANSMIT get_len end if return end subroutine MPP_TRANSMIT_ ! MPP_BROADCAST ! subroutine but that doesn t allow !broadcast to a subset of PEs This version and mpp_transmit will remain !backward compatible intent(inout) & T
static const std::string classname()
const ModelAuxControl_ & modelauxcontrol() const
Interfacing.
Increment_ & increment()
Interfacing.
Base class for encapsulation of the linear forecast model.
LinearModelMaker(const std::string &name)
LinearModelFactory Factory.
const Geometry_ & geometry() const
Interfacing.
const ModelAuxIncrement_ & modelauxincrement() const
Interfacing.
void stepTL(Increment_ &, const ModelAuxIncr_ &) const
Increment Class: Difference between two states.
virtual const oops::Variables & variables() const =0
void finalizeTL(Increment_ &) const
virtual LinearModelBase< MODEL > * make(const Geometry_ &geom, const eckit::Configuration &conf)
ModelAuxControl< MODEL > ModelAux_
static std::map< std::string, LinearModelFactory< MODEL > *> & getMakers()
ModelAuxIncrement< MODEL > ModelAuxIncr_
virtual ~LinearModelBase()
Geometry< MODEL > Geometry_
virtual void print(std::ostream &) const =0
void finalizeAD(Increment_ &) const
void initializeTL(Increment_ &) const
************************************************************************GNU Lesser General Public License **This file is part of the GFDL Flexible Modeling System(FMS). ! *! *FMS is free software without even the implied warranty of MERCHANTABILITY or *FITNESS FOR A PARTICULAR PURPOSE See the GNU General Public License *for more details **You should have received a copy of the GNU Lesser General Public *License along with FMS If see< http:! ***********************************************************************!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !! MPP_TRANSMIT !! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! subroutine MPP_TRANSMIT_(put_data, put_len, to_pe, get_data, get_len, from_pe, block, tag, recv_request, send_request)!a message-passing routine intended to be reminiscent equally of both MPI and SHMEM!put_data and get_data are contiguous MPP_TYPE_ arrays!at each call, your put_data array is put to to_pe 's get_data! your get_data array is got from from_pe 's put_data!i.e we assume that typically(e.g updating halo regions) each PE performs a put _and_ a get!special PE designations:! NULL_PE:to disable a put or a get(e.g at boundaries)! ANY_PE:if remote PE for the put or get is to be unspecific! ALL_PES:broadcast and collect operations(collect not yet implemented)!ideally we would not pass length, but this f77-style call performs better(arrays passed by address, not descriptor)!further, this permits< length > contiguous words from an array of any rank to be passed(avoiding f90 rank conformance check)!caller is responsible for completion checks(mpp_sync_self) before and after integer, intent(in) ::put_len, to_pe, get_len, from_pe MPP_TYPE_, intent(in) ::put_data(*) MPP_TYPE_, intent(out) ::get_data(*) logical, intent(in), optional ::block integer, intent(in), optional ::tag integer, intent(out), optional ::recv_request, send_request logical ::block_comm integer ::i MPP_TYPE_, allocatable, save ::local_data(:) !local copy used by non-parallel code(no SHMEM or MPI) integer ::comm_tag integer ::rsize if(.NOT.module_is_initialized) call mpp_error(FATAL, 'MPP_TRANSMIT:You must first call mpp_init.') if(to_pe.EQ.NULL_PE .AND. from_pe.EQ.NULL_PE) return block_comm=.true. if(PRESENT(block)) block_comm=block if(debug) then call SYSTEM_CLOCK(tick) write(stdout_unit,'(a, i18, a, i6, a, 2i6, 2i8)')&'T=', tick, ' PE=', pe, ' MPP_TRANSMIT begin:to_pe, from_pe, put_len, get_len=', to_pe, from_pe, put_len, get_len end if comm_tag=DEFAULT_TAG if(present(tag)) comm_tag=tag!do put first and then get if(to_pe.GE.0 .AND. to_pe.LT.npes) then!use non-blocking sends if(debug .and.(current_clock.NE.0)) call SYSTEM_CLOCK(start_tick)!z1l:truly non-blocking send.! if(request(to_pe).NE.MPI_REQUEST_NULL) then !only one message from pe-> to_pe in queue *PE waiting for to_pe ! call error else get_len so only do gets but you cannot have a pure get with MPI call a get means do a wait to ensure put on remote PE is complete error call increase mpp_nml request_multiply call MPP_TRANSMIT get_len end if return end subroutine MPP_TRANSMIT_ ! MPP_BROADCAST ! subroutine but that doesn t allow !broadcast to a subset of PEs This version and mpp_transmit will remain !backward compatible intent(inout) MPP_BROADCAST begin
void initializeAD(Increment_ &) const