#########################################################################
# This script is provided for
#
# Chen W and De Schutter E (2017) Parallel STEPS: Large Scale Stochastic Spatial Reaction-Diffusion Simulation with High Performance Computers. Front. Neuroinform. 11:13. doi: 10.3389/fninf.2017.00013
#
##########################################################################
import steps.mpi
import steps.utilities.geom_decompose as gd
import steps.rng as srng
import steps.mpi.solver as mpisolver
import time
from extra.constants import *
from extra import data_presets
from steps.utilities import metis_support
import sys
import os
import cPickle
if len(sys.argv) == 2:
RESULT_DIR = sys.argv[1]
else:
RESULT_DIR = "result_branch_influx"
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
MESH_FILE = "meshes/branch.inp"
MORPH_FILE = "meshes/branch.morph"
CA_CONC_PRESET = "extra/CaConc_PRESET1"
CA_P_CURR_DATA_FILE = "extra/CaPCurr_DATASET1"
SIM_TIME = 30.0e-3
PRESET_DATA_START_TIME = 50.0e-3
INFLUX_UPDATE_INTERVAL= 1.0e-3
DATA_RECORD_INTERVAL = 0.02e-3
########################### GET BIOCHEMICAL MODEL ###############################
import CaBurst_model
mdl = CaBurst_model.getModel()
########################### MESH & BRANCH MAPPING ###########################
import CaBurst_geom
mesh, rois, roi_areas, roi_vols = CaBurst_geom.getGeom(MESH_FILE, MORPH_FILE)
########################### Recording ###########################
if steps.mpi.rank == 0:
try: os.mkdir(RESULT_DIR)
except: pass
########################### PARTITIONING ###########################
partition_file = 'meshes/partition/branch.metis.epart.' + str(steps.mpi.nhosts)
mpi_tet_partitions = metis_support.readPartition(partition_file)
mpi_tri_partitions = gd.partitionTris(mesh, mpi_tet_partitions, mesh.getSurfTris())
########################### CREATE SOLVER ###########################
r = srng.create_mt19937(512)
r.initialize(int(time.time()) * steps.mpi.rank)
sim = mpisolver.TetOpSplit(mdl, mesh, r, tet_hosts = mpi_tet_partitions, tri_hosts = mpi_tri_partitions)
########################### LOAD PRESET CALCIUM INFLUX DATA ###########################
# convert P type calcium current data to calcium influx profile
ca_curr_data = data_presets.readData(CA_P_CURR_DATA_FILE)
ca_influx_profile = data_presets.genCaInfluxProfile(ca_curr_data, roi_areas, roi_vols, PRESET_DATA_START_TIME, PRESET_DATA_START_TIME + SIM_TIME + INFLUX_UPDATE_INTERVAL, INFLUX_UPDATE_INTERVAL)
# load preset background calcium concerntrations
ca_conc_preset_file = open(CA_CONC_PRESET, 'r')
ca_conc_preset = cPickle.load(ca_conc_preset_file)
ca_conc_preset_file.close()
if steps.mpi.rank == 0:
conc_result = open(RESULT_DIR + '/calcium_conc_%iprocs.dat' % (steps.mpi.nhosts), 'w', 0)
conc_result.write('#Entries: Time ')
for roi in rois:
conc_result.write('%s ' % (roi))
conc_result.write('\n')
sim.setCompConc('cyto', 'Mg', Mg_conc)
surfarea = sim.getPatchArea('memb')
pumpnbs = 6.022141e12*surfarea
sim.setPatchCount('memb', 'Pump', round(pumpnbs))
sim.setPatchCount('memb', 'CaPump', 0)
sim.setCompConc('cyto', 'iCBsf', iCBsf_conc)
sim.setCompConc('cyto', 'iCBCaf', iCBCaf_conc)
sim.setCompConc('cyto', 'iCBsCa', iCBsCa_conc)
sim.setCompConc('cyto', 'iCBCaCa', iCBCaCa_conc)
sim.setCompConc('cyto', 'CBsf', CBsf_conc)
sim.setCompConc('cyto', 'CBCaf', CBCaf_conc)
sim.setCompConc('cyto', 'CBsCa', CBsCa_conc)
sim.setCompConc('cyto', 'CBCaCa', CBCaCa_conc)
sim.setCompConc('cyto', 'PV', PV_conc)
sim.setCompConc('cyto', 'PVCa', PVCa_conc)
sim.setCompConc('cyto', 'PVMg', PVMg_conc)
for roi in rois:
sim.setROIConc(roi, 'Ca', ca_conc_preset[roi])
n_tpns = int(SIM_TIME / DATA_RECORD_INTERVAL) + 1
update_tpns = int(INFLUX_UPDATE_INTERVAL / DATA_RECORD_INTERVAL)
influx_change_tpns = []
# for each series of the profile, the first data point is the influx start time, following by the influx rate of each roi for the specific time window
for series in ca_influx_profile['Data']:
influx_change_tpns.append(series[0] - PRESET_DATA_START_TIME)
n_influx_changes = len(influx_change_tpns)
next_influx_change_tpn = 0
############################################################################
if steps.mpi.rank == 0:
print "Simulating model, it will take a while if running with small amount of processes..."
print "Note: You can monitor the current simulating time point and data in your result directory, but please do not write anything to the files yourself."
start_time = time.time()
for l in range(n_tpns):
sim_endtime = DATA_RECORD_INTERVAL * l
sim.run(sim_endtime)
if steps.mpi.rank == 0:
conc_result.write('%.6g' %(sim_endtime) + ' ')
for roi in rois:
conc = sim.getROIConc(roi, 'Ca')
if steps.mpi.rank == 0:
conc_result.write('%.6e' %(conc) + ' ')
if steps.mpi.rank == 0:
conc_result.write('\n')
# need to change influx rate
if l % update_tpns == 0:
for r in range(len(rois)):
sim.setROIReacK(rois[r], 'CaInflux', ca_influx_profile['Data'][next_influx_change_tpn][r + 1])
next_influx_change_tpn += 1
time_cost = (time.time() - start_time)
conc_result.close()
proc_file = open(RESULT_DIR + '/proc_%i.csv' % (steps.mpi.rank), 'w', 0)
proc_file.write("SimTime,CompTime,SyncTime,IdleTime,nIteration\n")
proc_file.write("%f,%f,%f,%f,%i\n" % (time_cost, sim.getCompTime(), sim.getSyncTime(), sim.getIdleTime(), sim.getNIteration()))
proc_file.close()
if steps.mpi.rank == 0:
performance_file = open(RESULT_DIR + '/performance_%iprocs.csv' % (steps.mpi.nhosts), 'w')
performance_file.write("Time Cost,%f" % (time_cost))
performance_file.write("\n")
performance_file.close()