Update documentation and refactor comments in iaf_psc_exp current input #2229
Conversation
models/iaf_psc_exp.cpp
Outdated
| @@ -248,8 +248,6 @@ nest::iaf_psc_exp::calibrate() | |||
|
|
|||
| const double h = Time::get_resolution().get_ms(); | |||
|
|
|||
| // numbering of state vaiables: i_0 = 0, i_syn_ = 1, V_m_ = 2 | |||
|
|
|||
| // commented out propagators: forward Euler | |||
There was a problem hiding this comment.
The Euler propagators should go. If at all, I would add a remark to the model documentation pointing out that this neuron model differs from the implemented in [1]_ by using exact integration instead of forward Euler.
There was a problem hiding this comment.
The comment says:
// commented out propagators: forward Euler
// needed to exactly reproduce Tsodyks network
Perhaps it makes sense for people coming after us to leave this in? We could also wrap it in some preprocessor conditions.
There was a problem hiding this comment.
I am strictly against leaving commented-out code in. Anyone with a minimal understanding of Euler's method should be able to re-Eulerize the code if they want to (we should put a remark into the model doc). Compiler switches are too cumbersome, since we'd need to propagate them from the CMake level. The only alternative I would consider would be a runtime switch in the model, similar to the consistent_integration switch in the Izhikevich model.
heplesser
added
I: No breaking change
Co-authored-by: jessica-mitchell <mitchell20j@gmail.com>
Co-authored-by: jessica-mitchell <mitchell20j@gmail.com>
Co-authored-by: jessica-mitchell <mitchell20j@gmail.com>
Co-authored-by: jessica-mitchell <mitchell20j@gmail.com>
…af_psc_exp_current_in
Co-authored-by: jessica-mitchell <mitchell20j@gmail.com>
Co-authored-by: jessica-mitchell <mitchell20j@gmail.com>
…af_psc_exp_current_in
|
@heplesser, @jessica-mitchell: thanks for the comments! I updated the code, could you have another look? |
Fixes #1226.
The meaning of the
i_syn_exp_was left unchanged, but a warning about its meaning was added in two places in the documentation.