Initialize revp stresses to previous time step

cicecore/cicedynB/dynamics/ice_dyn_shared.F90

@@ -503,7 +503,7 @@ subroutine dyn_prep2 (nx_block,   ny_block,   &
          taubx    (i,j) = c0
          tauby    (i,j) = c0
 
-         if (revp==1) then               ! revised evp
+         if (icetmask(i,j)==0) then 
             stressp_1 (i,j) = c0
             stressp_2 (i,j) = c0
             stressp_3 (i,j) = c0
@@ -516,20 +516,7 @@ subroutine dyn_prep2 (nx_block,   ny_block,   &
             stress12_2(i,j) = c0
             stress12_3(i,j) = c0
             stress12_4(i,j) = c0
-         else if (icetmask(i,j)==0) then ! classic evp
-            stressp_1 (i,j) = c0
-            stressp_2 (i,j) = c0
-            stressp_3 (i,j) = c0
-            stressp_4 (i,j) = c0
-            stressm_1 (i,j) = c0
-            stressm_2 (i,j) = c0
-            stressm_3 (i,j) = c0
-            stressm_4 (i,j) = c0
-            stress12_1(i,j) = c0
-            stress12_2(i,j) = c0
-            stress12_3(i,j) = c0
-            stress12_4(i,j) = c0
-         endif                  ! revp
+         endif                  
       enddo                     ! i
       enddo                     ! j
 

configuration/scripts/cice.batch.csh

@@ -165,6 +165,14 @@ cat >> ${jobfile} << EOFB
 #SBATCH --qos=standby
 EOFB
 
+else if (${ICE_MACHINE} =~ brooks*) then
+cat >> ${jobfile} << EOFB
+#PBS -N ${ICE_CASENAME}
+#PBS -j oe
+#PBS -l select=${nnodes}:ncpus=${corespernode}:mpiprocs=${taskpernodelimit}:ompthreads=${nthrds}
+#PBS -l walltime=${batchtime}
+EOFB
+
 else if (${ICE_MACHINE} =~ theia*) then
 cat >> ${jobfile} << EOFB
 #SBATCH -J ${ICE_CASENAME}

configuration/scripts/cice.launch.csh

@@ -153,6 +153,18 @@ mpirun -np ${ntasks} ./cice >&! \$ICE_RUNLOG_FILE
 EOFR
 endif
 
+#=======
+else if (${ICE_MACHINE} =~ brooks*) then
+if (${ICE_COMMDIR} =~ serial*) then
+cat >> ${jobfile} << EOFR
+./cice >&! \$ICE_RUNLOG_FILE
+EOFR
+else
+cat >> ${jobfile} << EOFR
+aprun -n ${ntasks} -N ${taskpernodelimit} -d ${nthrds} ./cice >&! \$ICE_RUNLOG_FILE
+EOFR
+endif
+
 #=======
 else if (${ICE_MACHINE} =~ theia*) then
 cat >> ${jobfile} << EOFR

configuration/scripts/machines/Macros.brooks_intel

@@ -0,0 +1,78 @@
+#==============================================================================
+# Makefile macros for ECCC brooks
+#==============================================================================
+# For use with intel compiler
+#==============================================================================
+
+CPP        := fpp
+CPPDEFS    := -DFORTRANUNDERSCORE ${ICE_CPPDEFS}
+CFLAGS     := -c -O2 -fp-model precise
+#-xHost
+
+FIXEDFLAGS := -132
+FREEFLAGS  := -FR
+FFLAGS     := -fp-model source -convert big_endian -assume byterecl -ftz -traceback  -diag-disable 5140
+#-xHost
+FFLAGS_NOOPT:= -O0
+
+ifeq ($(ICE_BLDDEBUG), true)
+  FFLAGS     += -O0 -g -check -fpe0 -ftrapuv -fp-model except -check noarg_temp_created
+# -heap-arrays 1024 
+else
+  FFLAGS     += -O2
+endif
+
+SCC   := cc
+SFC   := ftn
+MPICC := cc
+MPIFC := ftn
+
+ifeq ($(ICE_COMMDIR), mpi)
+  FC := $(MPIFC)
+  CC := $(MPICC)
+else
+  FC := $(SFC)
+  CC := $(SCC)
+endif
+LD:= $(FC)
+
+#NETCDF_PATH := /fs/ssm/hpco/tmp/eccc/201402/04/intel-2016.1.150/ubuntu-14.04-amd64-64/
+
+PIO_CONFIG_OPTS:= --enable-filesystem-hints=gpfs 
+
+#PNETCDF_PATH := $(PNETCDF)
+#PNETCDF_PATH := /glade/u/apps/ch/opt/pio/2.2/mpt/2.15f/intel/17.0.1/lib
+
+INCLDIR := $(INCLDIR)
+
+LIB_NETCDF := $(NETCDF_PATH)/lib
+LIB_PNETCDF := $(PNETCDF_PATH)/lib
+LIB_MPI := $(IMPILIBDIR)
+
+#SLIBS   := -L$(LIB_NETCDF) -lnetcdff -lnetcdf -L$(LIB_PNETCDF) -lpnetcdf -lgptl
+#SLIBS   := -L$(LIB_NETCDF) -lnetcdff -lnetcdf
+
+ifeq ($(ICE_IOTYPE), netcdf)
+   INCLDIR += $(shell nf-config --fflags)
+   SLIBS   := $(shell nf-config --flibs)
+endif
+
+
+ifeq ($(ICE_THREADED), true) 
+   LDFLAGS += -qopenmp 
+   CFLAGS += -qopenmp 
+   FFLAGS += -qopenmp 
+endif
+
+### if using parallel I/O, load all 3 libraries.  PIO must be first!
+ifeq ($(IO_TYPE), pio)
+   PIO_PATH:=/glade/u/apps/ch/opt/pio/2.2/mpt/2.15f/intel/17.0.1/lib
+   INCLDIR += -I/glade/u/apps/ch/opt/pio/2.2/mpt/2.15f/intel/17.0.1/include
+   SLIBS   := $(SLIBS) -L$(PIO_PATH) -lpiof
+
+   CPPDEFS :=  $(CPPDEFS) -Dncdf
+endif
+
+ifeq ($(IO_TYPE), netcdf)
+   CPPDEFS :=  $(CPPDEFS) -Dncdf
+endif

configuration/scripts/machines/env.brooks_intel

@@ -0,0 +1,25 @@
+#!/bin/csh -f
+
+source /opt/modules/default/init/csh
+module load PrgEnv-intel # Intel compiler
+module load cray-mpich # MPI (Cray MPICH)
+module load cray-netcdf # NetCDF
+module load cray-hdf5 # HDF5
+
+setenv ICE_MACHINE_ENVNAME brooks
+setenv ICE_MACHINE_COMPILER intel
+setenv ICE_MACHINE_MAKE make
+setenv ICE_MACHINE_WKDIR ~/data/brooks/cice/runs 
+setenv ICE_MACHINE_INPUTDATA /home/ords/cmdd/cmde/phb001/
+setenv ICE_MACHINE_BASELINE ~/data/brooks/cice/baselines
+setenv ICE_MACHINE_SUBMIT "qsub"
+setenv ICE_MACHINE_TPNODE 36
+setenv ICE_MACHINE_ACCT P0000000
+setenv ICE_MACHINE_QUEUE "development"
+setenv ICE_MACHINE_BLDTHRDS 4
+setenv ICE_MACHINE_QSTAT "qstat "
+
+if (-e ~/.cice_proj) then
+   set account_name = `head -1 ~/.cice_proj`
+   setenv CICE_ACCT ${account_name}
+endif

doc/source/science_guide/sg_dynamics.rst

@@ -557,7 +557,7 @@ Introducing another numerical parameter :math:`\alpha=2T \Delta t_e ^{-1}` :cite
 
 where as opposed to the classic EVP, the second term in each equation is at iteration :math:`k` :cite:`Bouillon13`. Also, as opposed to the classic EVP, 
 :math:`\Delta t_e` times the number of subcycles (or iterations) does not need to be equal to the advective time step :math:`\Delta t`. 
-A last difference between the classic EVP and the revised approach is that the latter one initializes the stresses to 0 at the beginning of each time step, 
-while the classic EVP approach uses the previous time level value. The revised EVP is activated by setting the namelist parameter `revised\_evp` = true. 
+Finally, as with the classic EVP approach, the stresses are initialized using the previous time level values. 
+The revised EVP is activated by setting the namelist parameter `revised\_evp` = true. 
 In the code :math:`\alpha = arlx` and :math:`\beta = brlx`. The values of :math:`arlx` and :math:`brlx` can be set in the namelist. 
 It is recommended to use large values of these parameters and to set :math:`arlx=brlx` :cite:`Kimmritz15`.