HNN Release 1.3.2

All releases after this one will include the hnn-core integration work and will depend on hnn-core as an installation prerequisite.

Significant changes since release v1.3.1:

• Allow for data files to be comma delimited in addition to space/tab-delimited
• Update install procedure to use HNN releases instead of code master branch
• Add installation instructions for Windows Subsystem for Linux (WSL)
• Deprecate Docker-based installation methods
• Add some basic pytest unit tests for Qt dialogs

HNN Release 1.3.1 [docker deprecation warning]

Note: docker is not going to be the recommended installation method in the next release and will eventually be deprecated.

This release includes a bug fix as a stop-gap until the new installation method is complete and documented.

• Since Docker Desktop 2.3.0.4, the way paths on the host system are addressed has changed. hnn_docker.sh has been updated for this new convention. PR #218

HNN Release 1.3.0

New features:

• Use hyperthreading cores in determining number of cores to run on for better performance
• Windows: running with WSL is much faster than native install and via Docker

Analysis:

• Dipole data files are saved with higher precision to avoid rounding errors when dipole is very small

Docker:

• Shared directory is now hnn_out. This directory exists in the docker container at the same path as on the host OS (e.g. /Users/me/hnn_out) instead of /home/hnn_user/hnn_out
• When loading files in GUI, there are shortcuts to hnn_out and source code directories
• Improve error checking: xauthority keys, ssh keys, open ports
• Refactoring of hnn_docker.sh that makes it much more robust
• Stop using docker-compose. Only relies on docker to manage containers
• Clean up hnn_docker.log formatting and remove special characters from output

Ubuntu:

• Install script renamed to hnn-ubuntu.sh
• Install script works with ubuntu 14.04, 16.04 and 18.04
• Use the precompiled NEURON package instead of compiling from source
• Log output from install script to ubuntu_install.log

Windows:

• Renamed install script to hnn-windows.ps1
• Support user names with spaces (issue #163)
• Install script will work for users without admin privileges (except MPI)
• Include a windows container Dockerfile that is used by Travis. It can be used to run simulations, but cannot launch GUI
• Support running HNN from WSL and run tests in Travis.

Mac

• Install now tested on OS versions Catalina, Mojave, Sierra, High Sierra, and El Capitan

HNN Release 1.2.5

Since v1.1.0...

New features:

• New hnn_docker.sh installation script for all platforms
• Optimization has range sliders to choose parameter ranges
• Added new param files: Alpha.param, gamma_L5weak_L2weak_bursty.param
• Added new data files: S1_ongoing.txt and 2 files for gamma tutorial
• Option to change the spectrogram color map from the default ‘jet’ scheme
• Better placement of new dialog boxes with large or dual-monitor setups
• Allow certain parameter ranges to be adjusted even after optimization starts

Analysis:

• Updates to weighted RMSE calculation in optimization.
• Remove artifact from Poisson inputs at low event rates

Stability:

• Fixed bug causing crashes when running multiple trials
• Fixed several bugs with optimization
• Fixed HNN crash when stopping optimization
• Fixed bug with save figures option
• Handle exceptions with a warning dialog instead of closing HNN

Installation:

• New Virtualbox image and instructions
• New AWS image and instructions
• Updated installation documentation for clarity

HNN Release 1.1.0

Bug fixes:

• Fixed automatic detection of available cores to not count hyperthreading cores
• Fixed dipole plot y-scaling to look at the currently selected sim
• Fixed a bug introduced in v1.0 where data on disk a successful simulation wasn't loaded
• Fixed which optimized parameter set was plotted at the conclusion of optimization (gray line)
• Redraw the plot canvas when data is removed so that axes can be scaled appropriately
• RMSE calculation will now properly handle experimental data that is shorter than simulation duration
• Handle case where the user changes simulation name during the middle of optimization (original value is used)
• Avoid crashing at the last optimization step when the previous steps bring two inputs close enough together that they become part of the same optimization step

Features:

• The histograms of evoked inputs displayed in the main window are more granular (smaller bin widths)
• Remove unnecessary x-axis tickmarks for histograms
• Build docker image with labels injected for the current version. Allows reusing cache on DockerHub builds
• Make parameter values read-only in optimization configuration window
• Display the delta between initial and optimized parameter values in optimization configuration window

HNN Release 1.0.0

Bug fixes:

• RMSE will only include parts of the data file up until tstop
• Fixed y-scaling of evoked input plots and arrows
• Corrected the unit labeling for CaT and HCN channel density
• Display trailing zeros in RMSE
• Pressing “Stop Stimulation” will terminate orphan nrniv processes
• Remove references to the old install location of /usr/local/

Features:

• Automated ERP model optimization fully integrated into GUI
• Simulation output directory renamed to “hnn_out” and source code directory renamed to “hnn_source_code”
• It is no longer required to specify hnn.cfg on the command line
• Instructions for running HNN on Amazon Web Services (AWS)
• Updated Docker-based installation procedures and instructions for running HNN
• Model visualization will display in Docker container
• Display distributions of evoked inputs if a user has a loaded pram file even before running the simulation
• Present welcome message to user on first startup

HNN Release 0.1.4

Improvements to HNN install:

1. Installation using Docker for Mac, Windows, and Linux is recommended. Instructions for installing on each OS without Docker is provided, called "native install".
2. Documentation is at
   https://jonescompneurolab.github.io/hnn/installer/

HNN Release 0.1.3

Add option to configure output location (dbase parameter in [paths] section of hnn.cfg file)

Fix suprathreshold ERP parameter file (remove overly high precision)

HNN Release 0.1.2

Added S1 suprathreshold ERP data file and corresponding parameters

Couple of adjustments to the installer

HNN Release 0.1.1

Make sure calcium decay time constant label (in L5 Pyramidal biophysics tab) is correct.