- BBT: Reducing elimination tree height for parallel LU factorization of sparse unsymmetric matrices (see the paper to appear in the proceedings of HiPC2014). Codes and a Matlab interface are available. (February 2015)
- uetree: Constructing elimination trees for sparse unsymmetric matrices (see the paper). The code and a mex file to call the code from MATLAB are available. Newer versions, as of October 2012: code and the mex file. Compilation: "mex uetreemex.c uetree.c -largeArrayDims''.
- ScalingSuit: A set of matrix scaling routines (see the paper for mathematical properties of the proposed algorithms and this one for parallelization of those algorithms).
- In MATLAB:
busymscalinf.m: Inf-norm scaling algorithm of the paper busymscalone.m: 1-norm scaling algorithm of the paper busymscaltwo.m: 2-norm scaling algorithm of the paper knight_sk.m: Sinkhorn-Knopp iterations for 1-norm scaling knight_sk2.m: Sinkhorn-Knopp iterations for 2-norm scaling scalebunch.m: Scaling symmetric matrices in Max norm by Bunch buScaleBunch.m: Max norm scaling (of symmetric or unsymmetric matrices) buScale.m: Auxiliary routine, to scale a matrix with given scaling matrices buScaleSK.m: Calls knight_sk for 1-norm scaling buScaleSK2.m: Calls knight_sk2 for 2-norm scaling - In Fortran, MPI-based parallel implementation:
butestsca.F: driver program showing the use of the scaling routines fac_scalings_simScaleAbs.F: the body fac_scalings_simScale_util.F: utility functions used by the body boratoymtx_5_5.txt: a sample input matrix for butestsca.F

- In MATLAB:
- MatchMaker: A set of augmenting path-based bipartite cardinality matching algorithms (see paper). As of mid 2011, includes a push-relabel algorithm as well, see paper. New version, 2013, matchmaker2 provides implementations on a GPU.
- MUMPS: Parallel scaling algorithms (see paper).
- PaToH: Matlab programming, k-way refinement algorithms for multi-constraint and fixed vertex partitioning (see papers here, there, and there).
- Task scheduling: Communicating tasks modeled with a task interaction graph are to be mapped to heterogeneous computing systems with homogeneous interconnection network (See paper. See also Task Intreaction Graph Data Set and Heuristics).

- ParMxvLib: Building blocks (sparse matrix-vector multiplies and vector operations) for parallel implementation of iterative methods for linear system solutions (Available upon request. See paper).