Toronto, Canada (see http://www.cnsorg.org/). This course will have a
separate materials fee, and it will run in parallel with the CNS*2007
Workshops (i.e. on July 11 or 12, 2007; see CNS WWW site for schedule).
The speakers at this course will include Michael Hines, Ted Carnevale,
Bob Calin-Jageman, and Felix Schürmann. Its primary focus will be on
the use of NEURON in a parallel simulation environment. We will review
key concepts of distributed computing as they relate to computational
modeling in neuroscience, describe their implementation in NEURON,
and present strategies for creating, debugging, and productively using
distributed models of networks and cells. Potential registrants who are
interested in an introduction to NEURON and/or other topics are invited to
contact Ted Carnevale (ted dot carnevale at yale dot edu).
The topics to be covered will include:
- Screensaver parallelization
Bulletin-board (Linda style) parallelization
Discrete spike events in network models
Improving efficiency with bin queue, spike compression, global ID (GID) compression, exchange buffer size
Using a synthetic milieu of afferent events for efficient study of subnets
Gap junction communication
Splitting cells for load balancing of nets and parallel simulation of individual cells
Reproducible, statistically independent randomization
Achieving computational identity regardless of cell distribution or splitting
Practical examples
--Dendrodendritic synapses
--Separating network architecture from cell details (benchmark in (Brette et al. submitted))
--A thalamocortical network model (Traub et al. 2005)
--Finding and fixing bugs in user code
For dates and registration information, see http://www.cnsorg.org/ .