Skip to content

July 2020

Jump to bottom
Alexandru Savulescu edited this page Jul 18, 2020 · 14 revisions

 Time: 17.7.2020, 16:00 CEST; 10:00 EDT Place: Zoom

NOTE: Please contact one of the NEURON dev members for the link if you are interested to join.

Agenda

  • Introduction from "new" members (5 min)
  • NEURON News / Announcements (2 min)
  • NetPyNE GUI demo / CNS Presentation (10 min)
  • NEURON CNS Presentation Summary (8 min)
  • Documentation - The Next Generation (5 min)
  • NMODL Python Wheels (4 min)
  • NEURON Options : Lazy Loading? (4 min)
  • NEURON + CoreNEURON Integration : Python API #643 (5 min)
  • NetPyNE + CoreNEURON Integration Status (5 min)
  • Summary of Important Issues (4 min)
  • CNS Presentations related to NEURON : Anything to highlight? (5 min)

Slides

See google slides

Attendees

  • Michael Hines (MH)
  • Robert McDougal (RM)
  • Salvador Dura-Bernal (SDB)
  • Bill Lyton (BL)
  • Pramod Kumbhar (PK)
  • Alexandru Savulescu (AS)
  • Fernando Pereira (FP)
  • Ioannis Magkanaris (IM)
  • Alessandro Cattabiani (AC)
  • Wouter Klijn (WK)
  • Blake Caldwell (BC)
  • Nicolas Cornu (NC)
  • Ted Carnevale (TC)
  • Roy Ben-Shalom (RBS)

Meeting Minutes

New members introduction

  • Roy Ben-Shalom working on the NeuroGPU platform for running NEURON on GPUs
  • Blake Caldwell working on Human Neocortical Neurosolver project - Brown University

Notes:

  • RBS would like to present NeuroGPU
  • BC interested in parallelisation and happy about neuron python wheels

No news / announcements :)

NetPyNe GUI Demo by SDB

Tutorial: Detailed Cell Network

  • Importing from HOC or Python templates
  • Section exploring, mechanisms, properties
  • 3D representation of the network and simulation
  • Different plots

Tutorial: Multi-scale/multilayer network with RxD components

  • RxD components are defined from the code and not the GUI.
  • 3D representation
  • Different plots, raster , LFP spectrogram, time-series, cell calcium concentrations

Notes:

  • SDB: Jupiter Notebook integration to interact with simulation network and parameters
  • SDB: export python script from the GUI
  • RM : Proposed working together on synergy between HTML RxD builder from NEURON and NetPyNe

NEURON CNS Presentation summary by RM

  • NEURON Overview
  • NEURON concepts and basic scripting
  • RxD
  • ModelDB

Notes:

  • RM: basically NEURON overview and workshop
  • RM: neurongui2 overview
  • PK discussed with RM about use-case for NMODL introspection (via h.hh())

SBML by RM

  • using libroadrunner as alternative backend is a possibility, nothing important to highlight

NEURON Documentation by AS

  • Consolidate dev+user+scm documentation
  • Faster docs-only Travis PRs
  • Live walkthrough of new documentation
  • Doxygen not yet pluggable into Sphinx layout (current tools not production ready)
  • Easier to search for things, all documentation indexed together

NMODL by AC

  • wheels for linux and Mac only
  • use of skbuild
  • simple example

Notes:

  • MH: interest for a prez on skbuild
  • BL + RM + PK: Windows support discussion. TC raises flag on current Windows adoption. MH + PK: should provide support in the future
  • RM + PK: NMODL dependency when we pip install neuron

NEURON + CORENEURON Integration: Python API by Pramod

  • Downsides of CORENEURON usability before
  • Work by MH and PK in the Python API: new coreneuron module

Notes:

  • WC: CORENEURON warning messages on CORENEURON compatibility for different modules. MH: a bit more work to be done on error/warning messages + RxD is incompatible with CORENEURON; RM: should not be so complicated to make RxD compatible

NetPyNe + CORENEURON Integration Status by IM

  • Minimal changes to NetPyNe
  • NMODL-ISPC backend support
  • Example runs with preliminary results on Tut2 and M1 circuit(with GPU)
  • More optimisation and debugging is planned

Notes:

  • SDB + PK: discussion on GPU details

nrn_options: Lazy loading? by FP

Notes:

  • FP: python version agnostic solution (Guido's official hack)
  • MH: discussion on new 3.7 getattr feature vs hack
  • RM + PK + MH + WK: how about support for 2.x ? Min 3.x version? MH: the hack looks like a good option at the moment

Summary of important issues

Notes:

  • PK: issue with Azure regarding new MPICH version
  • MH: add NRN to open-neuroscience.com -> RM and MH will take care of it
  • MH: Dynamic selection of legacy physical constants : actually a more complex problem, (most probably) not going to make it to 8.0alpha; build-time related, also valid for wheel creation
  • RM: plotly now supported for PlotShape

Extra notes on releasing:

  • RM + MH + PK: 8.0alpha release? MH: some issues must be moved to 8.0 milestone.
  • MH: regression testing against ModelDB not done, to be started; GO / NO-GO; still using legacy units; different results expected
  • PK: Trigger Azure for a pre-release?
  • AS + MH + WK: how is the regression testing done; MH: there are scripts with a lot of manual interactions + manual comparison because of several discrepancies; variable time step method roundoff errors for different versions (not for same version), need to address sundials

Next meeting August 21, 2020, 16:00 CEST; 10:00 EDT

Clone this wiki locally