Continuous LGN Filter State

This release fixes temporal filter-state handling in the Retina/LGN pathway. Filter state is now preserved across consecutive stimuli and blanks include the expected stimulus-offset response.

This causes a numerical break from previous results: responses remain effectively equivalent, but are not numerically identical, similar to using a different random seed. Exact legacy behavior remains available through enabling the original_2024_lgn_mode parameter.

Machine learning data generation

Mozaik release version before adjustments for machine learning data generation.

Rozsa_Cagnol2024

Mozaik release coinciding with publication of a new version of the pre-print on bioRxiv:
Rózsa, T., Cagnol, R., Antolík, J. Iso-orientation bias of layer 2/3 connections: the unifying mechanism of spontaneous, visually and optogenetically driven V1 dynamics (2024)

Rozsa_Cagnol2024

Mozaik release coinciding with publication of new pre-print on bioRxiv:
Rózsa, T., Cagnol, R., Antolík, J. Iso-orientation bias of layer 2/3 connections: the unifying mechanism of spontaneous, visually and optogenetically driven V1 dynamics (2024)

Mozaik version coinciding with submission to Plos. Comp. Bio.

This Mozaik version coincides with submission of the master model to PLOS Computational Biology in Dec. 2023.

If mozaik changes are required due to review, we will release version 0.3.1.

python3 port

Version 0.2.0 after port to python 3.

Main goals for version 0.3.0:

• unittests coverage for majority of the code
• revamped documentation with tutorials