Artificial Intelligence
Please fill the required field.

Cp folder 2024


, which means the directory /pdata/[USER]/2024/ with all the output files from the workers. There are several Mira topology and configuration files in the repo. The `nodes` file (hard coded in `create_nodes` or `create_nodes_extended`) specifies on which hosts the workers are started. The `run.config` file defines the communication between workers and master. The directory `Pegasus` defines the job submission workflow. Note that this is needed for the *cumulated case* (first run only) as mentioned in the heurisitic paper. NEWS ---- Version 1.53 (2021-02-20) ------------------------- - Improved coarsening quality measures: Average Reduction Factor, Proportion of Dropped Chequerboard Cells (PDC) and even/odd ratio - Code refactoring to prepare for general AdaptMesh interface - Removed unused code or functionality - Introduced `parallel_parsers` module which will provides parsing in parallel and sorted or unsorted entries. - New page in documentation → for the [generated static analysis](https://michaelschedler.github.io/marple/heuristic/sphinx_gameoflife.html) of [Game of Life](./example/gameoflife_png_dissection) Version 1.52 (2021-01-28) ------------------------- - Major heurisic bugfixes - Speed improvements: More cell coarsening is done in less steps and cell coarsening is parallelized - Major updates to this readme Version 1.3 (2020-03-11) ------------------------ - Instead of parsing all files at once, only those files are parsed and merged which are not used by the readers yet. This is accomplished using an queue. The added advantage of this is that the output of workers can be spread over several files and is not only be collected in one file. - Minor improvements in code structure to support parallel bugfixes and implementations - Added [example/eigen2d](example/eigen2d) to the repo based on the paper > Arnold, D. N., Batcho, P. (2002). High order temporal discretization in \( L^\infty \) for systems of conservation laws. Also added the new heuristic `half_half`. Version 1.2 (2019-10-19) ------------------------ - Implementation of the suggested heuristics in the [forthcoming paper](.) in the [GridRefinement](heuristic/refinement.py) class. In particular, SG-BBI (`Gauss_Buck_SG`) and anisotropic refinement rules (w.o. subcell). - Tested for dimension 2 with several mock problems - Refactoring: removing code which was not used any more after updating the roadmap (see documention) - Changed and simplified - Density field (`phi`), - Interface ID representation (`code_to_coord_index`) and - Priority computation (`coord_index_prio`) - inside the `MixedRefinement`. Version 1.1 (2019-09-29) ------------------------ - Fixed and improving file parsing - Added depthwise communication for parsing. - Compilation with current `Starpu-Tasks` Optimization results are of slightly lower quality compared to the hash-based parsing in 1.1. Version 1.02 (2019-10-22) ------------------------ Linux bindings improved. Dependencies ------------ Python dependencies are within `marple/src/gol/__init__.py`. See [pip requirements.tar.gz](doc/pip_requirements.tar.gz) for `pip` dependency intallation. PRACE implementation ==================== Now that the C++ version of the heuristic is tested (see forthcoming paper) I - shift to the archived `branches/prace-2021`. - I