Is PyPNS working with new versions of NEURON.

[nrnhines]

When I use NEURON-8.1.0 with python3.9 on my linux desktop and install PyPNS from github and run

python -i test.py

I get

	1 
Creating axon 0
Myelinated fiber diameter: [0.97982345]
Number of segments for myelinated axon: 2208

Starting simulation of axon 0
Elapsed time to calculate voltage and membrane current...
352.09 s
Elapsed time to calculate extracellular potential...
21.15 s
Overall processing of axon 0 took 385.83 s.
...

Note: https://www.neuron.yale.edu/phpBB/viewtopic.php?p=19436#p19436

[nrnhines]

Using NEURON -- VERSION 7.6.7 HEAD (603da174f) 2019-04-19
configured with

(pypns-39) hines@hines-t7500:~/tmp/pypns/build$ ../configure --prefix=`pwd`/install --with-nrnpython=python3.9 --without-iv --disable-rx3d
make -j 4 install

Generates the stdout/stderr

(pypns-39) hines@hines-t7500:~/tmp/pypns$ python3.9 -i test.py
	1 
Creating axon 0
Myelinated fiber diameter: [1.28313846]
Number of segments for myelinated axon: 1944

Starting simulation of axon 0
Elapsed time to calculate voltage and membrane current...
185.42 s
Elapsed time to calculate extracellular potential...
22.12 s
Overall processing of axon 0 took 217.44 s.
...

and the figure

[nrnhines]

Using NEURON -- VERSION 7.8.2 HEAD (09b151ecb) 2020-12-16 give yet a third set of numbers

Creating axon 0
Myelinated fiber diameter: [1.59591649]
Number of segments for myelinated axon: 1724

Starting simulation of axon 0
Elapsed time to calculate voltage and membrane current...
218.84 s

And a figure similar to 8.1.0 above (though different in detail).

[nrnhines]

It seems likely that a major part (or all) of the issue stems from #596 change due to issue #509 . Building the version just prior to this change, NEURON -- VERSION 8.0.dev-117-g10da6c2a2 generates

However NEURON -- VERSION 8.0.dev-118-g5a7e42cdc 2020-06-15 exhibits the problem.

Note that to speed things up, I modified test.py so that

--- a/test.py
+++ b/test.py
@@ -9,7 +9,7 @@ import numpy as np
 
 # ----------------------------- simulation params ---------------------------
 
-tStop=50
+tStop=10
 dt=0.0025

Also note that there are many places in the code that make use of randomness and I could not immediately figure out how to make the simulation deterministic.

[nrnhines]

The versions that contain that commit are:

$ git tag --contains 5a7e42cdc
7.8.1
7.8.2
8.0.0
8.0.1
8.0.2
8.0a
8.1.0
8.1a
8.2.dev

[AnaMachadoLuis]

Hello,

I tried to replicate some of the figures (https://github.com/chlubba/PyPNS-PaperFigures) of the PyPNS paper using neuron 8.0, PyPNS (with chlubba/PyPNS#6) and python 3.9 (windows 10), to understand if everything was working and I could use this set up.

The test.py generates the same results that are above represented when it's used neuron 7.6.7.

I was also able to reproduce figure 7 (a, b, c) from the paper. However, I'm not able to replicate the content of figures 8 and 9.

Is it possible that there is still some incompatibility between neuron recent versions and PyPNS?

Figures generated

fig. 8

fig. 9

Figures in the paper

fig. 8

fig. 9

[nrnhines]

@AnaMachadoLuis Thanks, I was hoping someone would check on those figures. I suspect the issue is the same. Can you tell me how to run those
figures. I presume the randomness does not affect the qualitative behavior too severely?

[AnaMachadoLuis]

Yes, of course. First I also considered that the influence of randomness could be an explanation, but then after looking more carefully at the differences between the figures I obtained and the ones in the paper, I realized they are quite significant.

This repository https://github.com/chlubba/PyPNS-PaperFigures contains the scripts to generate the figures in the PyPNS paper. There is a python script (sometimes two in the case when some previous calculations are necessary) for every figure.

In the same directory should be the following elements:

-python script to generate the desired figure (I ran it as a jupyter notebook)
-folder Fields (also in the https://github.com/chlubba/PyPNS-PaperFigures repository)
-compiled mod files (the mods folder is in https://github.com/chlubba/PyPNS.  In the directory of this folder, I ran the command nrnivmodl from the console, after which some files are compiled)

[nrnhines]

@AnaMachadoLuis
I don't understand how to run fig8 and fig9. I'm using Linux Ubuntu and I have folders

/home/hines/tmp/hack/PyPNS-PaperFigures
/home/hines/tmp/hack/pypns # clone of PyPNS

The former is my working directory.
to find PyPNS, I

export PYTHONPATH=$PYTHONPATH:/home/hines/tmp/hack/pypns

and to make sure I have the mod files

nrnivmodl ../pypns/mods

Then I tried for fig8 and got:

(py39) hines@hines-t7500:~/tmp/hack/PyPNS-PaperFigures$ python -i ___paper_fig8_profileZ.py
  File "/home/hines/tmp/hack/PyPNS-PaperFigures/___paper_fig8_profileZ.py", line 9
    rc('text.latex', preamble='\usepackage{sfmath}')
                                                   ^
SyntaxError: (unicode error) 'unicodeescape' codec can't decode bytes in position 0-1: truncated \uXXXX escape
>>> 
(py39) hines@hines-t7500:~/tmp/hack/PyPNS-PaperFigures$ export PYTHONPATH=$PYTHONPATH:/home/hines/tmp/hack/pypns^C
(py39) hines@hines-t7500:~/tmp/hack/PyPNS-PaperFigures$ python -i ___paper_fig8_profileZ.py
  File "/home/hines/tmp/hack/PyPNS-PaperFigures/___paper_fig8_profileZ.py", line 9
    rc('text.latex', preamble='\usepackage{sfmath}')
                                                   ^
SyntaxError: (unicode error) 'unicodeescape' codec can't decode bytes in position 0-1: truncated \uXXXX escape
>>> 

Any suggestions?

[nrnhines]

By using /home/hines/tmp/hack/pypns as my current working directory I was able to

(py39) hines@hines-t7500:~/tmp/hack/pypns$ python -i ../PyPNS-PaperFigures/___paper_fig9A_SFAP_timecourse.py

and see similar results for
NEURON -- VERSION 7.6.7 HEAD (603da17) 2019-04-19

and
NEURON -- VERSION 8.2.dev-54-gfe112466a master (fe11246) 2022-05-25

[AnaMachadoLuis]

@nrnhines
I also got the same error but couldn't find a solution, so I just commented the line of code which was causing problems.

Ok, I can see the results are the same, even though they don't match the contents of the original figures. But based on this, it  should not be a NEURON version problem.

Thank you.

[nrnhines]

@AnaMachadoLuis
Thanks for the hint about Figure 8. Here are my results for

(py39) hines@hines-t7500:~/tmp/hack/pypns$ python -i ../PyPNS-PaperFigures/___paper_fig8_profileZ.py

NEURON -- VERSION 7.6.7 HEAD (603da17) 2019-04-19

'NEURON -- VERSION 8.2.dev-54-gfe112466a master (fe11246) 2022-05-25'