Permalink
Comparing changes
Choose two branches to see what’s changed or to start a new pull request.
If you need to, you can also or
learn more about diff comparisons.
Open a pull request
Create a new pull request by comparing changes across two branches. If you need to, you can also .
Learn more about diff comparisons here.
...
- 20 commits
- 93 files changed
- 4 contributors
Commits on Mar 28, 2025
-
refactor(cli): accept -G and -k as positional arguments, default to p…
…arallel execution instead of sequential execution (#93) * feat(cli): accept -G and -k as positional arguments * feat(app): default to max_concurrency() instead of sequential execution
Commits on Mar 31, 2025
-
fix: update mt-kahypar adapter (#94)
* fix: update mtkahypar refiner * build: add mtkahypar to installation config
Commits on Apr 1, 2025
Commits on Apr 7, 2025
-
-
feat: make kassert optional (#99)
* refactor: always include assert.h instead of kassert.h * feat: add -DKAMINPAR_BUILD_WITH_KASSERT=<On/Off> option to make KASSERT optional * refactor: fix all unused warnings when building without KASSERT
Commits on Apr 24, 2025
-
refactor: cleanup ci workflows (#100)
* ci: publish github release as non-draft * ci: sign debian and rpm packages
-
* build: update nix flake * ci: add nix package build tester * ci: use nix install action instead of container * build: add numactl only to linux platform * ci: pin dependency * ci: fix nix package build tester * build: refactor nix flake
-
feat: edge sparsification during coarsening (#101)
* add uniform random sampling * add forest fire sparsification with networkit * make sparsification method selectable * remove unnessary counters in random sampler * fix --c-sparsification parsing * remove debug prints from cmake * sparsification targets * inital implementation of kN * improve performance of and add cli option for kN * extract steps into seperate methods * improve choosing of k * correct comutation of k * correct assert message * Create UnionFind * Spanning Trees in k-Neighbour * add cli option ofr kN with spanning tree * increase targetBurntRatio * imporve forest fire implementation * add threshold sampler * implement Forest Fire as ThresholdSampler * implement weight threshold * have ThresholdSampler sample exactly the target amount of edges * add util for loop over edges * Spanning Trees in k-Neighbour * Spanning Trees in k-Neighbour * Spanning Trees in k-Neighbour * include Laplacians as normal julia module not as a git submodule * struct on julia side * fix julia and extract data back into C++ * enable garbage colllector and free memory * rename ijv to ijvmatrix * er in julia * only load julia code once * change er sampler to er score * fix julia code and the linker * remove unused julia file * fix typo * inital weighted forest fire impl * fix memory issues * actually use scores * make computation of k more precice * implement alias method * implement alias method * random sampler * refactor cli & have RandomSampler terminate faster * linear search inplementation of weighted ff * wff with segment tree * initialize julia in the main function * add distribution without replacment * random with replacment sampler * independent sampler * cli for random sampler - broken * remove erroneous cast * fix forest fire score curried constructors * make distribution names consistant * fix sort_by_target * use distribution in kn * remove asserting casts * implement unbiased theshold sampler * change calculation of k * binary search for threshold in ut * remove reweighing function * kN: fix double counting at incident_weights and correct index with offset when sampling * parallelize sparsify * parallelize ThesholdSampler * change grainsize to 2000, since this value is also used in label_propagation.h * time ThresholdSampler and sparsify * implement WFF without NetworKit * parallelize with __atomic_add * add statistic and correct logic error * trivial parallization of ir * implement medians_of_medians * initial parallel approx of normalization factor * partial fix of ir * remove comp * add unit test and fix IR * remove debug printf * only approximate with edge weight because no atomic double * test and fix qselect * test & fixes for qselect * faster theshold sampler * deal with equal elements in qselect * some fixed * fix search in approxNormalizationFactor * linear time selection of neighbors in WFF * linear time selection of neighbors in WFF * add no-approx cli option * remove asserts * make threshold sampler sample exactly the target amount * wff fire statistics * cli for wff parameters * new quickselect for threshold sampling * Optimize WFF with preallocation and new exponential clock impl * check if no neighbors were sampled and mark start vertex as visited * add virtual destructor tor SparsificationTarget * add missing = default * add missing destructors * remove costly statistics * remove julia and EffectiveResistanceScore * refactor sparsification targets * laziness factor * remove submodule networkit * remove networkit imports * remove networkit imports * add support for shrink factor to sparsifing_cluster_coarsener * refactor: iterative quick select * refactor: reduce quickselect memory + running time * refactor: save more memory * minor * minor * further reduce memory * replace sample[] array by highest bit of the edge weights * feat: standalone threshold sparsifier * . * . * . * . * . * bugfixes * bugfix * fix * bugfix * cleanup * fix: segfault * test and fix threshold sampler * remove depricated todo * refactor: fix all warnings * test * test * test * initial impl of make_scores_symetric * improve make_scores_symetric * remove asserts * add option for ignoring weights in wff code (not yet in cli) * refactor: reduce code duplication in cluster coarseners * feat: implemented uniform-random with thresold-sparsifying-clustering * refactor: change --s-recontract to be the default * cli options for unweighted forest fire * correct name to keep_only_negative_edges * feat(presets): add mtkahypar-kway preset * fix(thresold-sparsifying-coarsener): add missing cluster bucket recomputation * refactor: fix gcc-arm64 compile warning, preset include paths * handle edge case target=0 * fix(lp): update cluster count during two-hop clustering (clustering+matching) * add 2 hop strategy to preset * fix: fix accidental edge weight negation in sparsifying_cluster_coarsener * refactor: speedup wff * fix: avoid EdgeID overflow in forest fire * fix: overflow in printf() statement * sparsification timer * refactor: merge with main and remove unused sparsification techniques * fix: error * fix: errors and refactor tests * fix: add missing file * refactor: rename cli arg * refactor: improve output * fix: error --------- Co-authored-by: Dominik Rosch <dominik.rosch@student.kit.edu>
Commits on Apr 25, 2025
-
fix: refactor mt-kahypar adapter (#102)
Co-authored-by: Daniel Seemaier <17594350+DanielSeemaier@users.noreply.github.com>
-
fix: update tools and benchmarks (#103)
* fix(shm): correctly map compressed file format * fix(shm): update tools and remove unnecessary ones * fix(shm): update benchmarks and remove unnecessary ones
Commits on Apr 26, 2025
Commits on Apr 30, 2025
-
fix(dist): suppress unused variable warnings when building the distri…
…buted part without KASSERT
Commits on May 5, 2025
Commits on May 21, 2025
Commits on May 27, 2025
-
feat(build): require gtest to be preinstalled by default if building …
…tests (#106) * feat(build): require gtest to be preinstalled by default if building tests * ci: update required dependencies
Commits on Jun 1, 2025
-
build: check system for dependencies before downloading; add force-do…
…wnload option (#107) * build: add option to download mt-kahypar * build: check system for dependencies before downloading; add force-download option
Loading
This comparison is taking too long to generate.
Unfortunately it looks like we can’t render this comparison for you right now. It might be too big, or there might be something weird with your repository.
You can try running this command locally to see the comparison on your machine:
git diff v3.4.1...main