This guide to programming Lava will provide a growing collection of learning resources to help you become a Lava developer! It will cover all aspects of Lava enabling you to create, compile, execute and understand Lava Processes. Review the Lava Architecture section for an introduction to the fundamental architectural concepts of Lava.
The following two sections currently include application examples that give a broader overview how to use Lava features in practice while the tutorials on the fundamental concepts provide a deeper dive into specific features of Lava and how to use them.
These are standalone tutorials that can be followed in any order.
- MNIST digit classification: The classical 'Hello World!' example illustrating how to build a simple feed-forward image classifier using leaky-integrate-and-fire neurons.
Coming soon:
- Liquid state machine: Another classical canonical example in neuromorphic computing, illustrating a complex, spiking and recurrent neural network for time series classification.
- Learning: Many neuromorphic architectures offer local on-chip learning via bio-inspired STDP learning. This tutorial provides a simple example how such learning rules can be used.
These tutorials walk step-by-step through the fundamental concepts of Lava introduced in Lava Architecture. These tutorials tend to build on each other, therefore it's best to consume them in order.
1. Installing Lava: Quickly get Lava installed, tested and ready to develop on your system.
2. Processes: Learn how to create Processes which are Lava's fundamental computational building blocks.
3. Process models: Learn to implement Process behavior with the ability to run on diverse backends via Lava ProcessModels.
4. Execution: This notebook demonstrates how to configure, start, and stop the execution of a network of Processes.
Coming shortly:
5. Connecting processes: How to connect Processes for message-based communication via channels and to build a network of asynchronously operating and interacting Processes.
6. Hierarchical processes: Processes can be composed into hierarchical processes. Learn how to implement SubProcessModels to build modular Processes of Processes.
7. Direct memory access: Explains how to realize direct memory access between Processes and what to be cautious about.
8. State monitoring: Lava offers probes to monitor the evolution of temporal state during execution. Learn how to create probes, retrieve timeseries and how to create custom visualization of network dynamics.