ChaNGa massively parallel N-body code

N-Body Home

About Us

Search

Contact Info

ChaNGa (Charm N-body GrAvity solver) is a code to perform collisionless N-body simulations. It can perform cosmological simulations with periodic boundary conditions in comoving coordinates or simulations of isolated stellar systems. It also can include hydrodynamics using the Smooth Particle Hydrodynamics (SPH) technique. It uses a Barnes-Hut tree to calculate gravity, with hexadecapole expansion of nodes and Ewald summation for periodic forces. Timestepping is done with a leapfrog integrator with individual timesteps for each particle.

This release includes a number of more recent physics improvements including metal cooling , molecular hydrogen based star formation, superbubble feedback, formation and feedback from supermassive black holes, and self-interacting dark matter. This release also provides compatibility with the API for Charm++ version 6.9.0, which includes requiring the use of a C++11 compatible compiler. For GCC, this means version 4.8.1 or later. The SPH physics in this release is comparable to the SPH described in Wadsley et al, 2017.

Differences between version 3.4 and 3.3 include:

• Move to Charm version 6.9.0
• Improved configure/make system.
• Molecular Hydrogen physics for star formation and feedback
• Metal cooling
• Protoplanetary disk cooling
• Superbubble feedback
• Early UV feedback
• Supermassive Black Hole formation and feedback
• Self-Interacting Dark Matter
• Gas splitting
• Local tree walk can be performed on the GPU.
• Better memory use in GPU code.
• More even distribution of work across processors when multistepping
• Refactored Main timestepping routine.

Differences between version 3.3 and 3.2 include:

• Move to Charm version 6.8.0
• Many SPH improvements from Wadsley et al, 2017
• External gravitational potentials
• AGORA (2016) feedback model
• Damping force for glass creation
• Working GPU implementation
• Gravity with single precision floating point

Differences between version 3.2 and 3.1 include:

• Move to Charm version 6.7.1
• Grackle cooling module
• Jeans length pressure floor
• Better scaling for tree builds
• Better multistep load balancing

Differences between version 3.1 and 3.0 include:

• Support for greater than 2 billion particles.
• Support for NChilada file format.
• AVX instruction support.
• SMP load balancing using the Charm++ Ck_Loop Library.
• Faster domain decomposition.
• New load balancers with better scaling.

Differences between version 3.0 and 2.0 include:

• Gas Cooling.
• Star formation and feedback.
• SMP aware.
• Improved neighbor search algorithms
• Improved load balancers.

Differences between version 2.0 and 1.0 include:

• Smooth Particle Hydrodynamics is now implemented.
• Hexadecapole expansions for the gravity calculation.
• Physical gravitational softening lengths for comoving simulations.
• Gravitational timestepping option.
• Non-cubical periodic boundary conditions.
• Adiabatically grow particle masses for equilibrium initial conditions.
• Output of movie files is available.
• Realtime visualization via liveViz.
• Output at specific redshifts or times.
• Support for CUDA.
• SSE2 implementation of force calculations.
• Multistep-aware load balancer.
• Rewrite of inter node caching to use the Charm++ Cache framework.
• Checkpointing using the Charm run time system.

ChaNGa's novel feature is the use of the Charm++ parallel programming system, including its dynamic load balancing schemes, in order to obtain high performance on massively parallel systems. It has been shown to scale well up to 512,000 processors on NCSA Blue Waters.

ChaNGa is being released under the Gnu Public License (GPL). The distribution and development repository is hosted on our N-BodyShop GitHub site. ChaNGa version 3.4 requires version 6.9.0 of Charm++ available from the UIUC PPL download site.

If you choose to use ChaNGa for scientific work, please reference the code papers:

P. Jetley, F. Gioachin, C. Mendes, L. V. Kale, and T. R. Quinn. Massively parallel cosmological simulations with ChaNGa. In Proceedings of IEEE International Parallel and Distributed Processing Symposium 2008, 2008.

P. Jetley, L. Wesolowski, F. Gioachin, L. V. Kale, and T. R. Quinn. Scaling hierarchical n-body simulations on gpu clusters. In Proceedings of the 2010 ACM/IEEE International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2010, Washington, DC, USA, 2010. IEEE Computer Society.

H. Menon, L. Wesolowski, G. Zheng, P. Jetley, L. Kale, T. Quinn, F. Governato. 2015. Adaptive Techniques for Clustered N-Body Cosmological Simulations. Computational Astrophysics and Cosmology, 2, 1.

For more information on obtaining, building and running ChaNGa, please see the Wiki documentation. Help can also be requested from the mailing list changa-users@u.washington.edu. To subscribe to the list, visit the mailman page. Bugs and feature requests can be submitted to our issue tracker.

The development of ChaNGa was supported by a National Science Foundation ITR grant PHY-0205413 and grant AST-1311956 to the University of Washington, and NSF ITR grant NSF-0205611 to the University of Illinois. Contributors to the program include Harshitha Menon, Isaac Backus, Ben Keller, Greg Stinson, Graeme Lufkin, Tom Quinn, Rok Roskar, Gengbin Zheng, Filippo Gioachin, Sayantan Chakravorty, Amit Sharma, Pritish Jetley, Lukasz Wesolowski, Edgar Solomonik, Celso Mendes, Joachim Stadel, and James Wadsley.

	N-Body Shop University of Washington Box 351580 Seattle, WA  98195-1580 (206) 543-2888 voice, (206) 685-0403 FAX