Research

Simulation engineering

Performance modeling for the neural simulator NEST

Our work aims at developing performance models for the simulation stage of NEST by a semi-empirical approach. We collect measurements of the runtime performance of NEST under varying parameter settings, and subsequently fit theoretical models to this data.

NESTML - A modeling language for spiking neuron and synapse models for NEST

NESTML is a modular and extensible domain specific language which provides neuroscience domain concepts as first-class language constructs and supports domain experts in creating neuron models for the neural simulation tool NEST.

SpiNNaker Support for the Neural Simulator NEST

The aim of the project is to add the SpiNNaker hardware to the platforms supported by NEST. Exact reproducibility and correctness of simulation results are important.

I/O Performance for the NEST Neural Simulator

The NEST simulator for spiking neuronal networks is meant to be a scalable simulator: it is designed to function on laptops, clusters, current petascale supercomputers and beyond to emerging exascale architectures. At extreme scales, issues of memory utilization and network size have emerged as challenges to be solved.

Structural Plasticity

Structural plasticity in a neural network refers to the physical creation and deletion of synapses. This effect is present during brain development, learning and healing after lesions.

Arbor

Arbor is a performance portable library for the simulation of large networks of multi-compartment, morphologically detailed neurons on emerging HPC architectures. It is developed under an open development model (https://github.com/arbor-sim/arbor) by the Jülich Supercomputing Centre’s SimLab Neuroscience and the Swiss National Supercomputing Center (CSCS), in close collaboration with the neuroscientific community.

Modular Science

The modular science framework is a software for the deployment of complex interactive workflows on supercomputers, which serves as an orchestrator for scientific applications.

Validation methodology for network simulations

In this project, we are investigating methods for quantitative and qualitative validation of neuron models and neural network simulations.

Accelerated neural network simulations: novel neuromorphic system architecture approaches

In this project, we are exploring hard- and software architectural approaches to identify opportunities to accelerate simulations of complex neural networks.

High Performance Computing for the Virtual Brain

In this project, our goal is to enhance the general performance of simulation and analysis tools provided by "The Virtual Brain" (TVB). In collaboration with the research group of Dr. Viktor Jirsa at Aix-Marseille Université, we are working on the development of a modular framework for the implementation of TVB neural mass models on HPC resources including automatic code generation schemes to efficiently port TVB models between computing architectures.

Unified framework for spiking and gap-junction interactions

In the unified framework project, the Jacobi waveform relaxation method is used to solve gap-junction interactions between neurons in the Nest simulator (Hahne et al., 2015). The implemented algorithm based on this method allows for a tradeoff between numerical precision and communication size between neurons in the simulation.

Dynamic structure for the Virtual Connectome

The Virtual Brain (TVB) is a "framework for the simulation of the dynamics of large-scale brain networks with biologically realistic connectivity". Tractography data can be combined with a variety of neural mass models in order to predict experimental and clinical observables such as local field potential, EEG and fMRI measures.

VIGAN: Visualization-guided Interactive Graph Analysis

VIGAN is a tool for high performance analysis and interactive multi-view visualization of graph/network data used in neuroscience research. It incorporates a wide variety of analysis and visualization features from existing tools and expands the applicability of graph analysis to a broader set of use cases, as well as larger datasets, by providing novel functions.

Interactive steering and visualization

We have developed a steering and visualization tool which allows scientists to interact with NEST simulations during run time. The interaction consists of the possibility to change parameters of the network and visualize the effect of these changes in the simulation.The tool is based on the ‘nett’ software framework developed at RWTH Aachen University.

Visual Connectivity Generation

There is a gap between experimental descriptions of connectivity in neuronal networks and information which can be used by simulators to generate network connectivity. With this project, we aim to generate a visual language for connectivity and a framework which can serve as a bridge between experimental datasets and simulation models. Empirically derived data and theoretical models expressed in a shared visual representation called NeuroScheme are used to generate connections for simulation or further processing.

Data analytics for high-throughput image-based cohort phenotyping

In this project, we are generalizing image processing pipelines, data mining techniques, uncertainty management, machine learning and inference methods to HPC systems. These methods are applied by the Connectivity group of INM-1 for high-throughput image-based cohort phenotyping.

Deep Learning for Brain Extraction from MRI scans

This project comprises research and development in application of Deep Learning methods to MRI brain extraction.

Analysis of sub-millimetre resolution fMRI for laminar dissection of the brain function

The project utilizes high-resolution structural and functional MRI to investigate the function of the cerebral cortex along its depth, measuring network connectivity with depth specificity. Achieving precise co-registration between the structural and functional scans of every sulcus and gyrus is essential for this analysis.

Jülich Parameter Exploration (JuPeX)

Carried out in collaboration with Prof. Wolfgang Maass's research group at the Technical University of Graz, this project enables large scale parameter space exploration and optimization of simulations on HPC systems. By extending the UNICORE and JUBE frameworks developed at the JSC, we will provide a flexible platform for parameter exploration using well known optimization algorithms.

Deep Learning for segmentation of 3D-PLI images

In this project we employ the state-of-the-art in deep learning for computer vision to solve the problem of segmentation of ultra-high resolution images of thinly sliced sections of post-mortem brain. The deep neural networks are trained to differentiate brain tissue from background.

Calibration of 3D-PLI measurements

In this project, we contribute algorithmic improvements and high performance implementations thereof, to the 3D-PLI calibration process.

Maximum Likelihood Estimation based Connection Reconstruction (MLECR)

Understanding the synaptic connections and organization of neural networks is a key step in understanding the dynamics of the brain. There are two challenges: simple models are often unable to resolve ambiguities and complex models are computationally very expensive. The MLECR method achieves good performance in both these problem domains.

Quality assurance for the 3D-PLI workflow

In this project we develop methods and tools that help identify errors at various stages of the complex Three-dimensional Polarized Light Imaging (3D-PLI) workflow.

Using structure tensor analysis to measure PLI image registration quality

In this project, we employ the Structure Tensor Analysis (STA) method to verify if the 3D reconstruction of the brain from images of the thinly sliced sections is accurate enough for use in further research.

Last Modified: 04.10.2025

Institute

Jülich Supercomputing Centre (JSC)