v3.5.0

What's Changed

• manual_setup and backend-related fixes by @kenblu24 in #38
• Add SNN.to_networkx() function by @kenblu24 in #37
• Better zeroing/resetting/activating of neuron states by @kenblu24 in #36
• Resolve #35 by zeroing refractory periods on reset() by @kenblu24 in #39

Full Changelog: v3.4.0...v3.5.0

v3.4.0

What's Changed

• Added doc fixes for draft release by @michabay05 in #13
• Accessors Overhaul by @kenblu24 in #24
• Numpy-like interface for properties of accessor lists by @kenblu24 in #25
• Add and clear spikes on NeuronListView by @kenblu24 in #26
• Allow for deletion of neurons and synapses semi-sanely by @kenblu24 in #27
• don't monkey-patch custom JSONEncoder.default by @kenblu24 in #28

New Contributors

• @michabay05 made their first contribution in #13

Full Changelog: v3.3.0...v3.4.0

v3.4.0b1

What's Changed

• Added doc fixes for draft release by @michabay05 in #13
• Accessors Overhaul by @kenblu24 in #24
• Numpy-like interface for properties of accessor lists by @kenblu24 in #25
• Add and clear spikes on NeuronListView by @kenblu24 in #26
• Allow for deletion of neurons and synapses semi-sanely by @kenblu24 in #27
• don't monkey-patch custom JSONEncoder.default by @kenblu24 in #28

New Contributors

• @michabay05 made their first contribution in #13

Full Changelog: v3.3.0...v3.4.0b1

v3.3.0

Lists are now first-class ways of interacting with lists of Neurons and Synapses.

What's Changed

• Improved accessor classes and Spike Deletion by @kenblu24 in #11

Full Changelog: v3.2.2...v3.3.0

v3.2.2

Bugfixes:

• Fix input_spikes_info printing all spikes even with max_entries set ( #9 )
• Fix minor issues with schema & schema versioning

Full Changelog: v3.2.1...v3.2.2

v3.2.1

Changes:

• Bugfix: Fixed incorrect type when setting snn.spike_train when importing from json.
  Instead of a 2D np.ndarray, it now gets correctly set as a list[np.ndarray] of 1D arrays.
• Minor: When importing or exporting using the skipkeys parameter, users will be warned if they accidentally pass a str, as opposed to a list[str] or set[str].

Full Changelog: v3.2.0...v3.2.1

v3.2.0

What's Changed

• Export & Import SuperNeuroMAT models to/from an .snn.json file. by @kenblu24 in #5
• Bugfix: Neuron.connect_child() and Neuron.connect_parent() now correctly return the virtual Synapse that gets created.

Full Changelog: v3.1.0...v3.2.0

v3.1.0

New Contributors

• @kenblu24 made their first contribution in #4

What's Changed

• SuperNeuroMAT 3.1.0 by @kenblu24 in #4

This is a major release, with some breaking changes. Here is a migration guide for users
coming from SuperNeuroMAT < v3.0.0..

Here is a brief summary of the new features:

• Faster simulate() evaluation
• Ability to run simulate() more than once
• No need to call setup() anymore
• Optional GPU[cuda] and JIT backends for faster evaluation
  • CPU, sparse-CPU, JIT, GPU, backends are automatically chosen based on availability and network size.
• Neuron and Synapse classes
• Ability to reset the model to a prior state
• Better printouts
• Cleaner API
  • Better and more lenient error checking with helpful error messages where it counts
  • Users shouldn't need to inspect or modify _private attributes anymore
• Better management of duplicate synapses and spikes
• Automatic installation of dependencies when installing via pip/uv
  • Only two dependencies are required: numpy>1.26.0, and scipy. pandas is no longer required.
  • GPU and JIT backends require additional libraries.
• Online Documentation!
  • And a getting started guide
• Automated testing and documentation building via GitHub actions

Full Changelog: v2.0.1...v3.1.0

v2.0.1

Major Updates in v2.0.1

1. Install using: pip install superneuromat
2. Update/upgrade using pip install superneuromat --upgrade
3. Import using: import superneuromat as snm
4. The main class could be accessed using model = snm.NeuromorphicModel() previously; however, this API has changed to the following: snn = snm.SNN()
5. Support for sparse computations for all SNN operations is available; this significantly speeds up the simulation of sparsely connected SNNs; sparse computations can be selected automatically; however, to explicitly enable sparse computations, do: snn.setup(sparse=True)
6. Reset functionality is available; to reset the SNN, do snn.reset(); checkout readthedocs for the specifics of all the parameters that can be reset
7. Spike count functionality is implemented; do snn.count_spikes() to get a count of number of spikes that transpired during a simulation
8. Multiple synapses from a neuron to any another neuron are not permitted; the SNN.create_synapse() method throws an error if this is attempted
9. Support for single (32-bit) and double (64-bit) precision computations is now available; select the precision as follows: snn.setup(dtype=32) for single precision and snn.setup(dtype=64) for double precision
10. Refer to the updated documentation on readthedocs for more details on the API

v1.4

Major Updates

1. Updated installation instructions
2. Updated usage instructions
3. Updated API in NeuromorphicModel.create_synapse() function: It now has stdp_enabled parameter instead of the previously enable_stdp parameter
4. Added bare bones docs

Full Changelog: v1.3...v1.4