# --------------------------------------------------------------------------
modeller = PyEMMAAnalysis(pdb_file=pdb_file,
source_folder=File('../staging_area/ntl9/trajs'))
# --------------------------------------------------------------------------
# CREATE THE SESSION
# the instance that runs the simulations on the resource
# --------------------------------------------------------------------------
# add the task generating capabilities
project.register(engine)
project.register(modeller)
project.open()
scheduler = project.get_scheduler(cores=2)
trajs = project.new_trajectory(pdb_file, 100, 4)
# submit the trajectories
scheduler.submit(trajs)
def task_generator():
return [engine.run(traj) for traj in project.new_ml_trajectory(100, 2)]
scheduler.add_event(Event().on(project.on_ntraj(range(
4, 50, 2))).do(task_generator))
# todo: change that this will stop when the first event is done
scheduler.add_event(Event().on(project.on_ntraj(10)).do(
modeller.task_run_msm).repeat().until(project.on_ntraj(20)))
args='-r --report-interval 1 --store-interval 1').named('openmm')
# --------------------------------------------------------------------------
# CREATE AN ANALYZER
# the instance that knows how to compute a msm from the trajectories
# --------------------------------------------------------------------------
modeller = PyEMMAAnalysis(pdb_file=pdb_file).named('pyemma')
project.generators.add(engine)
project.generators.add(modeller)
# --------------------------------------------------------------------------
# CREATE THE CLUSTER
# the instance that runs the simulations on the resource
# --------------------------------------------------------------------------
scheduler = project.get_scheduler('gpu')
# print scheduler.rp_resource_name
# print scheduler.resource.resource
trajectories = project.new_trajectory(engine['pdb_file'], 100, 4)
task = engine.run(trajectories)
scheduler(task)
scheduler.wait()
scheduler.exit()
project.close()
# CREATE AN ANALYZER
# the instance that knows how to compute a msm from the trajectories
# --------------------------------------------------------------------------
modeller = PyEMMAAnalysis(pdb_file=pdb_file).named('pyemma')
project.generators.add(engine)
project.generators.add(modeller)
# --------------------------------------------------------------------------
# CREATE THE CLUSTER
# the instance that runs the simulations on the resource
# --------------------------------------------------------------------------
gpu_scheduler = [
project.get_scheduler('gpu', runtime=4 * 24 * 60) for _ in range(4)
]
pyemma_scheduler = project.get_scheduler('cpu',
cores=1,
runtime=4 * 24 * 60)
# create 4 trajectories
trajectories = project.new_trajectory(pdb_file, 100, 4)
gpu_scheduler[0](trajectories)
gpu_scheduler[0].wait()
# now start adaptive loop
def strategy_trajectory(scheduler, loops, num):
for loop in range(loops):
trajectories = [
project.new_ml_trajectory(length=20, number=4)
# --------------------------------------------------------------------------
# CREATE AN ANALYZER
# the instance that knows how to compute a msm from the trajectories
# --------------------------------------------------------------------------
modeller = PyEMMAAnalysis(pdb_file=pdb_file).named('pyemma')
project.generators.add(engine)
project.generators.add(modeller)
# --------------------------------------------------------------------------
# CREATE THE CLUSTER
# the instance that runs the simulations on the resource
# --------------------------------------------------------------------------
scheduler = project.get_scheduler('gpu', cores=1, runtime=4 * 24 * 60)
# create 4 blocks a 4 trajectories
trajectories = [
project.new_trajectory(engine['pdb_file'], 100, 4) for _ in range(4)
]
tasks = map(engine.run, trajectories)
print trajectories
# submit
scheduler(tasks)
scheduler.wait()
# now start adaptive loop
# --------------------------------------------------------------------------
# CREATE AN ANALYZER
# the instance that knows how to compute a msm from the trajectories
# --------------------------------------------------------------------------
modeller = PyEMMAAnalysis(pdb_file=pdb_file).named('pyemma')
project.generators.add(engine)
project.generators.add(modeller)
# --------------------------------------------------------------------------
# CREATE THE CLUSTER
# the instance that runs the simulations on the resource
# --------------------------------------------------------------------------
scheduler1 = project.get_scheduler('cpu', cores=4, runtime=4 * 24 * 60)
scheduler2 = project.get_scheduler('cpu', cores=16, runtime=4 * 24 * 60)
scheduler3 = project.get_scheduler('cpu', cores=1, runtime=4 * 24 * 60)
# create 4 trajectories
trajectories = project.new_trajectory(pdb_file, 100, 4)
scheduler1(trajectories)
scheduler1.wait()
# now start adaptive loop
def strategy_trajectory(scheduler, loops):
for loop in range(loops):
trajectories = project.new_ml_trajectory(length=20, number=4)
tasks = scheduler(trajectories)
yield tasks.is_done()
# CREATE THE MODELLER
# the instance to create msm models
# --------------------------------------------------------------------------
modeller = PyEMMAAnalysis(
pdb_file=pdb_file,
source_folder=File('../staging_area/ntl9/trajs'))
# add the task generating capabilities
project.register(engine)
project.register(modeller)
# todo: save the task_generators for later in the project
with project:
# lets get the default scheduler from the resource with all the
scheduler = project.get_scheduler('gpu', cores=4)
trajs = project.new_trajectory(pdb_file, 100, 4)
# submit the trajectories
scheduler.submit(trajs)
def task_generator():
return [
engine.run(traj) for traj in
scheduler.new_ml_trajectory(100, 2)]
scheduler.add_event(
Event()
.on(scheduler.on_ntraj(range(10, 50, 10)))
.do(task_generator))