www.neuron.yale.edu

Hi,

I am currently investigating how to record results from NEURON to file (eventually, I want to write the contents of such files to another file format).

So far, I have found the options to store vectors ...
1) via GUI plots to a text file
2) via printf to a text file
3) via vwrite to a binary file

My questions:
- Are there any other methods to store simulation data?
- Is there a standard way to store big amounts of simulation data? For example, the second option above (printf) allows to store in pretty much every format - is there kind of a "standard" format that's adopted by most users?
- In your experience, how often do NEURON users store their raw simulation data (in terms of voltage time series etc) vs directly storing plots or analyzed results?

Thanks in advance!

Johannes Rieke
OpenWorm

jrieke wrote:Are there any other methods to store simulation data?

Nope.

Is there a standard way to store big amounts of simulation data?

There are as many "standards" for data storage as there are groups of people with shared interests. It is possible to get quite carried away with standards. KISS an important principle.

is there kind of a "standard" format that's adopted by most users?

Not really. If you routinely use a particular program that wants a particular data format, e.g. CSV, save your data in that format. Some people like a format something like this

comment
number_of_data_pairs
x0 y0
x1 y1
. . . etc. . . .

or this

label:somestring
number_of_data_pairs
x0 y0
x1 y1
. . . etc. . . .

because they work nicely with NEURON's "clipboard" (part of its standard run system), or

5 201
x ica ica_nacax ica_capump ica_cachan ica_pmp_cadifpmp
0 -0.000422814 -0.00028025 0 -0.000142564 0
0.025 -0.000422814 -0.00028025 0 -0.000142564 0.00083607
0.05 -0.0013857 -0.00208088 0 -0.000140883 0.000875016
0.075 -0.00278452 -0.00351414 0 -0.000145392 0.000954281
. . . etc. for a total of 201 lines . . .

i.e.
number_of_traces number_of_samples_in_time
x y0name y1name . . . yN-1name
x0 y0_0 y1_0 . . . yN-1_0
x1 y0_1 y1_1 . . . yN-1_1
. . . etc. . . .
where the x values are clearly time and the yi are the values of 0..N-1 different variables that change with time.

In your experience

Depends on the individual user's need more than anything else. Some are interested only in spike times, so all they record are
cell_id spiketime
pairs.