A unified framework for spiking and gap-junction interactions in distributed neuronal network simulations.

Area of Science:

• Computational neuroscience
• Neuroscience simulation
• High-performance computing

Background:

• Modern neural network simulators support complex topologies and neuron models.
• Advancements enable brain-scale network simulations on supercomputers.
• Distributed simulations optimize spike communication, but impede instantaneous gap-junction interactions.

Purpose of the Study:

• To develop a numerical algorithm for simulating gap junctions in distributed neural networks.
• To integrate gap-junction modeling with existing delayed communication strategies.
• To ensure compatibility with current simulation infrastructures like the NEST simulator.

Main Methods:

• A waveform-relaxation technique was employed to develop the numerical algorithm.
• The algorithm was implemented and integrated into the NEST simulator.
• Benchmarks were conducted on workstations, clusters, and supercomputers to assess performance and accuracy.

Main Results:

• The novel algorithm successfully enables network simulations incorporating gap junctions.
• The implementation integrates smoothly with existing spiking connection infrastructures.
• Benchmarks demonstrate high performance and accuracy across various computing platforms.

Conclusions:

• The unified framework for gap-junction and spiking interactions is efficient and accurate.
• This technology overcomes previous limitations in simulating instantaneous neuronal interactions in distributed networks.
• The approach facilitates more comprehensive and realistic large-scale neural network simulations.