New version of Arbor neural network simulation library released

Enhanced capabilities have been added to Arbor v0.8.


Arbor is a multi-compartment neural network simulation library designed to be portable across contemporary high-performance computing architectures. The most recent version is now available on EBRAINS. Some of the new features include:

Sources of white noise

This new feature introduces sources of white noise to Arbor. This means that both point and density mechanisms may now use white noise as part of the state updates - effectively turning ordinary differential equations into stochastic differential equations. The noise sources are provided per connection end point (point mechanism) or control volume (density mechanism) and are uncorrelated (spatially, temporally, and across different mechanism instantiations).

Mutable connection table

Users can now add functionality, documents, and examples while editing the connection table. Further functionality will come as requested.

Editing morphologies

Arbor now allows users to edit morphologies. The following operations are supported:

  • join_at
  • split_at
  • equivalence
  • equality
  • apply isometry

Arbor cable cell exporter and backend support in BluePyOpt

This new functionality supports Arbor simulations with cells optimized with Neuron in BluePyOpt as implemented in BlueBrain/BluePyOpt. BluePyOpt is an extensible framework for data-driven model parameter optimisation that wraps and standardizes several existing open-source tools.
Further information

Learn more about Arbor v.08 on the Arbor website.

Arbor on EBRAINS

Click here to learn more about Arbor on EBRAINS.

Image: Mean (repeated 10x) speed-up factors for Arbor with respect to NEURON of 46,939 mechanism combinations inserted in a single compartment. Credit: Robin De Schepper.