www.neuron.yale.edu

If you're using NEURON 8.1+, the pure-NEURON solution is to either use h.SaveState (which saves ~everything) or rxd.save_state() and rxd.restore_state(), where the rxd-specific solutions represent state using byte-strings. See the (admittedly brief) discussion at https://nrn.readthedocs.io/en/latest...

Unfortunately rxd doesn't directly support expressions that depend on time.

Algebraic expressions should just work. (There's an example of this in the documentation linked below.)

For math functions, you'll want to use rxdmath see the rxdmath discussion in the documentation

Roughly speaking the HOC code would look exactly like the Python, but drop the h. part. That said, HOC and Python are just two different interfaces to the same simulator. You can mix and match. e.g. from Python, you can run a HOC script called my_script.hoc via: h.load_file("my_script.hoc"...

This looks like an install from the PKG installer?

In particular, you'll find you have a NEURON folder in Applications. Remove that, and then enter the following into the terminal:

Glu.nodes is a list of all Glu nodes. If there is only one, then it is unambiguously safe to use Glu.nodes[0] . Otherwise, you'll want to filter down to a specific node by specifying e.g. segment, possibly region if there are multiple by using () for the filtering... So something like Glu.nodes(som...

I don't understand the distinction that you're making. Synapses are at some specific location, and using the mark method of a matplotlib/plotly h.PlotShape (or the point_mark method of a h.Shape), you can put points at specific locations in whatever style... e.g. with a matplotlib PlotShape, you can...

Why not something like:
You can pass regular matplotlib arguments to mark to specify color, marker type, etc.

You don't have to use pyplot or plotly (I prefer the interactivity of plotly) for visualizing PlotShapes. If you have an existing process that worked with interviews windows, you can still use that. If you do use the pyplot version though, you can save to a png in the usual pyplot way, and use that ...

If you got an error or unexpected result with source_var, can you please post it here?

That should just work.

You can always check out the YouTube playlist to see what's new and jump directly to videos of interest: https://www.youtube.com/playlist?list=P ... cdRSFwRLOU.

Yes. Intracellular reaction-diffusion (both 1D and 3D) parallelizes trivially as all cell-to-cell communication is through the regular spike exchange or gap junction mechanisms. (You still need ParallelContext for those, but you get the full speedup from splitting the cells onto different processors...

I suggest in Python using the object-oriented form ( h.hh ) rather than the string ( "hh" ) because it gives you an object whose properties you can explore (although your exploration options depend on whether or not you have the nmodl Python module installed...); i.e. soma.insert(h.hh) You...

I haven't wrapped my head around all of this, but here's a few points to help: I'm not sure I understand your placement array, but there's a comment "L = Length of each segment = 1/nseg"... the length of a segment is the length of the section (which presumably isn't 1 micron) divided by th...

We don't have a timeline for an official 8.0 release yet, but:

There was an alpha release in January: https://github.com/neuronsimulator/nrn/ ... s/tag/8.0a

On mac and linux, you can get the nightly development build via: pip install neuron-nightly

It would be cleaner to have the Python function run at every timestep using a CVode.extra_scatter_gather and then just transfer values to NMODL (using pointers or directly; both should work.)

You could probably get the call from NMODL to work, but that would require a VERBATIM block.