Hans Ekkehard Plesser: Research

Interests

Scientific Computing
Neuronal Network Simulation
Network Descriptions
Reproducible research

Projects

NEST Developement of a versatile, efficient neuronal network simulator in collaboration with the NEST Initiative. The simulator is available for download from the NEST Initiative Homepage. Currently work is under way to improve simulator efficiency further, especially in parallel simulations.

Reproducibility With a number of colleagues, I have analyzed the lack of standards for model sharing and the resulting difficulties in reproducing results in computational neuroscience. In September 2011, I organized a workshop on the Creating, Documenting and Sharing Network Models together with Sharon Crook and Jim Bednar. I also participated in the INCF Program for Multiscale Modeling.

Results

Results

NEST Completely revised the way in which NEST generates connections, providing significant performance improvement, flexible parameterization of connections and extensibility (in collaboration with Hannah Bos and Jochen Eppler, FZ Jülich). 2014

NEST Record-breaking simulation of a network with 1.73 billion neurons and over 10 trillion synapses on the Japanse K supercomputer using NEST by German and Japanese collaborators. I contributed a crucial data structure for compact neuron representation. 2013

NEST Systematic statistical tests for correct generation of random connectivity in large neuronal networks (MSc thesis Daniel Hjertholm, co-supervised by Birgit Kriener). 2013

NEST Topology 3 Module for NEST released as part of NEST 2.2.0, providing support for 3D Networks and signficantly better scaling for > 10k CPUs (with Håkon Enger). 2012

NEST Significantly reduced memory overhead making very large network simulations possible on > 10k CPUs (with Susanne Kunkel, Tobias Potjans, Jochen Eppler, Abigail Morrison, and Markus Diesmann). 2011

NEST New Python interface and new user manual for NEST Topology Module; significantly enhanced user interest. 2010

Modelling Developed Connectivity Pattern Tables and ConnPlotter to visualize connectivity in complex networks. 2010

Modelling Analyzed model description practice and proposed Good Model Description Practice. 2009

NEST NEST now provides a multicompartment neuron and a smart new multimeter recorder. 2009

NEST NEST benchmarked on Norwegian national supercomputing resources Stallo, Titan and Hexagon. 2009

NEST NEST Topology Module makes it easy to create complex neuronal networks (with Kittel Austvoll). 2008

NEST Refactored NEST model code splitting data members into Parameters, State, Buffers, and Variables to enforce careful thinking, provide systematic initialization and reset and prepare for network serialization. 2008

NEST Completely refactored stimulation and recording devices in NEST. 2008

NEST Network connecting times in NEST reduced from quadratic to linear scaling by smart choice of data structure. 2007

NEST First million-neuron network simulation with NEST. 2007

NEST Achieved supralinear scaling in a prototype of the next-generation NEST kernel: run NEST on four processers and get five times as fast! 2004

NEST First Public NEST Release during the Advanced Course in Computational Neuroscience, Obidos, Portugal 2004

CoThaCo CoThaCo architecture defined, implementation in progress. 2003

Caspar Acquired funding for SMP Compute Server (1.7 mill NOK) and installed server as computing resource for my group, IMT and UMB. 2003

NEST Implemented conductance-based synapses in NEST (with my student Johan Hake). Ported NEST to IBM AIX, Linux for SGI Altix, and Mac OS X. 2003

Linear Thalamo-cortical model Complete linear model for the transfer function of relay cells in the dorsal LGN. With Gaute Einevoll. 2002

NEST Implemented new random number library in NEST. 2002

Escape Models Escape noise models equivalent to the leaky integrate-and-fire neuron, but mathematically more convenient. With Wulfram Gerstner. 2000

ModUhl Routines computing first-passage-time distributions of modulated Ornstein-Uhlenbeck processes. 2000

Stochastic Resonance Markov chain analysis of stochastic resonance in integrate-and-fire neurons using numerical methods instead of simulations. Demonstrated SR both in noise amplitude and signal frequency. See my thesis for details. 1999

Markov Classes A C-library for the efficient simulation of large stochastic systems, extending work by Thomas Fricke and Dietmar Wendt. 1995

NEST	Developement of a versatile, efficient neuronal network simulator in collaboration with the NEST Initiative. The simulator is available for download from the NEST Initiative Homepage. Currently work is under way to improve simulator efficiency further, especially in parallel simulations.
Reproducibility	With a number of colleagues, I have analyzed the lack of standards for model sharing and the resulting difficulties in reproducing results in computational neuroscience. In September 2011, I organized a workshop on the Creating, Documenting and Sharing Network Models together with Sharon Crook and Jim Bednar. I also participated in the INCF Program for Multiscale Modeling.

NEST	Completely revised the way in which NEST generates connections, providing significant performance improvement, flexible parameterization of connections and extensibility (in collaboration with Hannah Bos and Jochen Eppler, FZ Jülich).	2014
NEST	Record-breaking simulation of a network with 1.73 billion neurons and over 10 trillion synapses on the Japanse K supercomputer using NEST by German and Japanese collaborators. I contributed a crucial data structure for compact neuron representation.	2013
NEST	Systematic statistical tests for correct generation of random connectivity in large neuronal networks (MSc thesis Daniel Hjertholm, co-supervised by Birgit Kriener).	2013
NEST	Topology 3 Module for NEST released as part of NEST 2.2.0, providing support for 3D Networks and signficantly better scaling for > 10k CPUs (with Håkon Enger).	2012
NEST	Significantly reduced memory overhead making very large network simulations possible on > 10k CPUs (with Susanne Kunkel, Tobias Potjans, Jochen Eppler, Abigail Morrison, and Markus Diesmann).	2011
NEST	New Python interface and new user manual for NEST Topology Module; significantly enhanced user interest.	2010
Modelling	Developed Connectivity Pattern Tables and ConnPlotter to visualize connectivity in complex networks.	2010
Modelling	Analyzed model description practice and proposed Good Model Description Practice.	2009
NEST	NEST now provides a multicompartment neuron and a smart new multimeter recorder.	2009
NEST	NEST benchmarked on Norwegian national supercomputing resources Stallo, Titan and Hexagon.	2009
NEST	NEST Topology Module makes it easy to create complex neuronal networks (with Kittel Austvoll).	2008
NEST	Refactored NEST model code splitting data members into Parameters, State, Buffers, and Variables to enforce careful thinking, provide systematic initialization and reset and prepare for network serialization.	2008
NEST	Completely refactored stimulation and recording devices in NEST.	2008
NEST	Network connecting times in NEST reduced from quadratic to linear scaling by smart choice of data structure.	2007
NEST	First million-neuron network simulation with NEST.	2007
NEST	Achieved supralinear scaling in a prototype of the next-generation NEST kernel: run NEST on four processers and get five times as fast!	2004
NEST	First Public NEST Release during the Advanced Course in Computational Neuroscience, Obidos, Portugal	2004
CoThaCo	CoThaCo architecture defined, implementation in progress.	2003
Caspar	Acquired funding for SMP Compute Server (1.7 mill NOK) and installed server as computing resource for my group, IMT and UMB.	2003
NEST	Implemented conductance-based synapses in NEST (with my student Johan Hake). Ported NEST to IBM AIX, Linux for SGI Altix, and Mac OS X.	2003
Linear Thalamo-cortical model	Complete linear model for the transfer function of relay cells in the dorsal LGN. With Gaute Einevoll.	2002
NEST	Implemented new random number library in NEST.	2002
Escape Models	Escape noise models equivalent to the leaky integrate-and-fire neuron, but mathematically more convenient. With Wulfram Gerstner.	2000
ModUhl	Routines computing first-passage-time distributions of modulated Ornstein-Uhlenbeck processes.	2000
Stochastic Resonance	Markov chain analysis of stochastic resonance in integrate-and-fire neurons using numerical methods instead of simulations. Demonstrated SR both in noise amplitude and signal frequency. See my thesis for details.	1999
Markov Classes	A C-library for the efficient simulation of large stochastic systems, extending work by Thomas Fricke and Dietmar Wendt.	1995