batch_run · batch_save

Running and Saving Batch Jobs

Note

These functions are about running one simulation and saving state variables at regular intervals. While this can be done manually with a series of calls to e.g., n.continuerun() and then saving data to a file, these functions avoid the overhead of returning back to the Python interpreter.

If you are instead looking to run a collection of simulations together using MPI, look at e.g., ParallelContext.runworker() and ParallelContext.submit().

batch_run()

Python

Syntax:

n.batch_run(tstop, tstep, 'filename')
n.batch_run(tstop, tstep, 'filename', 'comment')

Description:

This command replaces the set of commands:

while n.t < tstop:
    for i in range(int(tstep / dt)):
        n.fadvance()
    # print results to filename

and produces the most efficient run on any given neuron model. This command was created specifically for Cray computers in order eliminate the interpreter overhead as the rate limiting simulation step.

This command will save selected variables, as they are changed in the run, into a file whose name is given as the third argument. The 4th comment argument is placed at the beginning of the file. The batch_save() command specifies which variable are to be saved.

The variables are stored in plain text and separated by spaces (see the example output below).

Example:

The following code creates a single compartment neuron, adds Hodgkin-Huxley channels, applies a current clamp, runs, and stores a voltage trace.

from neuron import n

# define a geometry
soma = n.Section("soma")
soma.L = 10
soma.diam = 10

# biophysics: Hodgkin-Huxley channels
soma.insert(n.hh)

# add a stimulus
iclamp = n.IClamp(soma(0.5))
iclamp.dur = 0.2
iclamp.delay = 0.3
iclamp.amp = 0.5

# define variables to be stored (time and the soma's membrane potential)
n.batch_save()
n.batch_save(n._ref_t, soma(0.5)._ref_v)

# initialize, run, and save
n.finitialize(-65)
n.batch_run(2, 0.1, 'hhsim.dat', 'My HH sim')

The output (the time series of an action potential) is stored in the hhsim.dat:

My HH sim
batch_run from t = 0 to 2 in steps of 0.1 with dt = 0.025
 0 -65
 0.1 -64.9971
 0.2 -64.9943
 0.3 -64.9917
 0.4 -49.6876
 0.5 -34.9008
 0.6 -33.426
 0.7 -25.5015
 0.8 -7.00019
 0.9 20.989
 1 38.2226
 1.1 40.9284
 1.2 39.1047
 1.3 35.8921
 1.4 31.8901
 1.5 27.3462
 1.6 22.4496
 1.7 17.3559
 1.8 12.1873
 1.9 7.0331
 2 1.9538

See also

Vector.record()

HOC

Syntax:

batch_run(tstop, tstep, "filename")

batch_run(tstop, tstep, "filename", "comment")

Description:

This command replaces the set of commands:

while (t < tstop) {
    for i=0, tstep/dt {
            fadvance()
    }
    // print results to filename
}

and produces the most efficient run on any given neuron model. This command was created specifically for Cray computers in order eliminate the interpreter overhead as the rate limiting simulation step.

This command will save selected variables, as they are changed in the run, into a file whose name is given as the third argument. The 4th comment argument is placed at the beginning of the file. The batch_save() command specifies which variable are to be saved.

---

batch_save()

Python

Syntax:

n.batch_save()
n.batch_save(varref1, varref2, ...)

Description:

n.batch_save()

starts a new list of variables to save in a batch_run() .

n.batch_save(varref1, varref2, ...)

adds pointers to the list of variables to be saved in a batch_run.

Example:

n.batch_save()    # This clears whatever list existed and starts a new
                      # list of variables to be saved.
n.batch_save(soma(0.5)._ref_v, axon(1)._ref_v)
for i in range(3):
    n.batch_save(dend[i](0.3)._ref_v)

specifies five quantities to be saved from each batch_run().

HOC

Syntax:

batch_save()

batch_save(&var, &var, ...)

Description:

batch_save()

starts a new list of variables to save in a batch_run() .

batch_save(&var, &var, ...)

adds pointers to the list of variables to be saved in a batch_run. A pointer to a range variable, eg. “v”, must have an explicit arc length, eg. axon.v(.5).

Example:

batch_save()        // This clears whatever list existed and starts a new
            // list of variables to be saved.
batch_save(&soma.v(.5), &axon.v(1))
for i=0,2 {
    batch_save(&dend[i].v(.3))
}

specifies five quantities to be saved from each batch_run.