CyberGIS Toolkit is a suite of loosely coupled open-source geospatial software components that provide computationally scalable spatial analysis and modeling capabilities enabled by advanced cyberinfrastructure. CyberGIS Toolkit represents a deep approach to CyberGIS software integration research and development and is one of the three key pillars of the CyberGIS software environment, along with CyberGIS Gateway and GISolve Middleware. The integration approach to building CyberGIS Toolkit is focused on developing and leveraging innovative computational strategies needed to solve computing- and data-intensive geospatial problems by exploiting high-end cyberinfrastructure resources such as supercomputing resources provided by the NSF Extreme Science and Engineering Discovery Environment (XSEDE) and high-throughput computing resources on the Open Science Grid (OSG).
A rigorous process of software engineering and computational intensity analysis is applied to integrate an identified software component into the toolkit, including software building, testing, packaging, scalability and performance analysis, and deployment. This process includes three major steps:
- Local build and test by software researchers and developers using continuous integration software or services such as Travis CI;
- Continuous integration testing, portability testing, small-scale scalability testing on the National Middleware Initiative (NMI) build and test facility; and
- XSEDE-based evaluation and testing of software performance, scalability, and portability. By leveraging the high-performance computing expertise in the integration team of the NSF CyberGIS Project, large-scale problem-solving tests are conducted on various supercomputing environments on XSEDE to identify potential computational bottlenecks and achieve maximum problem-solving capabilities of each software installation.
- PGAP. PGAP is a scalable Parallel Genetical Algorithm (PGA) solver for the Generalized Assignment Problem (GAP). This code provides an efficient PGA implementation for combinatorial optimization problem-solving and scaled up to 262K processor cores on BlueWaters with marginal communication cost;
- Parallel Agent-Based Modeling (PABM). PABM is an illustrative software for scalable spatially explicit agent-based modeling (ABM);
- Parallel Kernel Density Estimation A multi-GPU code for data-intensive kernel density estimation (KDE). Spatial computational domain is first estimated for KDE. Then the study area is decompose into sub-regions through an adaptive partitioning approach. Each sub-region is processed by a GPU node. These GPU nodes are communicated through massage passing interface (MPI);
- Parallel Map Algebra This package contains a parallel map algebra code using CUDA, MPI, and Parallel I/O. It is extracted from a parallel geospatial programming models training package;
- Simple Parallel Tiff Writer This is a parallel IO code for writing GeoTIFF file on a parallel file system.
- ScalaGAHealth (Scalable Geographic Analytics for Health-Related Data)
- SpatialEvo (A high-performance evolutionary computing library for solving complex spatial optimization problems)
Please send your suggestions or ideas to CyberGIS Helpdesk (email@example.com).
Each software component integrated in CyberGIS Toolkit is open source and has its own copyright and license.
If you have any questions about CyberGIS Toolkit, please contact CyberGIS Helpdesk (firstname.lastname@example.org).