Topographica comes with the following step-by-step guides to running simulations.
The Tk GUI is a good way to explore a model without needing to write any additional code. However, the GUI only supports a limited set of operations, and so after doing an initial exploration using the GUI, we recommend using the more-powerful IPython notebook interface illustrated in the following section.
How to run and test a simple orientation map simulation using the GCAL cortical model from Stevens et al. (J. Neuroscience 2013). The tutorial allows you to present various objects to a saved orientation map network, and to visualize and analyze the responses. It also shows how to develop a new simulation using different input patterns.
How to run a model that develops selectivity for position, mapping the input space to the cortical space. The model uses the abstract SOM algorithm, focusing on the basic principles of self-organization rather than modeling any particular biological system.
In addition to the GUI-based way of interacting with Topographica illustrated in the tutorials above, you can also use IPython Notebook, which provides a command-line prompt that allows you to weave together Python code, textual output, and graphical output. This approach makes it easier to modify a model, look at parts of it not currently exposed in the GUI, and to develop new types of analyses not yet offered in the GUI.
How to run and explore the SOM algorithm in the notebook environment (same topics as the above GUI tutorial, but now showing how the commands are invoked and can be modified). In addition, this notebook includes several videos showing how the SOM develops over time, showcasing a feature only available in the notebook environment.
An exploration of GCAL or LISSOM model in the notebook environment. This tutorial covers the same material as the first GCAL tutorial, but adds animations showing how GCAL develops over time.
A demonstration of how the Collector class can be defined and used to collect measurements from a given model over development. Here the GCAL model is explored using more involved analysis than presented in the basic GCAL tutorial. Analysis over development includes orientation preference histograms, hypercolumn distance estimation, pinwheel finding and the pinwheel density estimation.
The above GCAL tutorials all focus on orientation maps. This tutorial shows how to use GCAL for developing maps for other visual feature dimensions, in any combination. Making all maps develop well together is an ongoing project (and requires much greater computational resources than individual maps), but the individual maps and most combinations should be usable already.
Other GCAL notebooks replicating Stevens et al. (J. Neuroscience 2013) may be found here. These notebooks only aim to replicate the published results and are not up-to-date demonstrations of the recommended workflow for using Topographica in the IPython Notebook environment.
Each of these tutorials should be a good starting point for seeing how Topographica works and how to use a Topographica model in practice. If you develop any other tutorials for your own models, we would love to add them here!