@techreport{TR-DCC-92-07,
  number = {DCC-92-07},
  author = {Setubal, João C.},
  title = {New Experimental Results For Bipartite Matching},
  month = {November},
  year = {1992}, 
  institution = {Department of Computer Science, University of Campinas},
  note = {In English, 17 pages.
    \par\selectlanguage{english}\textbf{Abstract}
      We present experimental results for 3 bipartite matching
      algorithms on 3 classes of sparse graphs. The push-relabel
      maximum flow algorithm, specialized for unweighted bipartite
      graphs, came out as the most robust algorithm, being the fastest
      by a significant margin in two of the classes and competitive in
      the other one.  The other two algorithms implemented are
      Hopcroft and Karp's and Alt, Blum, Mehlhorn, and Paul's, and it
      is noteworthy that both have better worst-case bounds than the
      push-relabel algorithm. The input classes used are variations of
      random bipartite graphs. We also show that speed-ups of up to
      3.2 with respect to the sequential implementation can be
      obtained by parallelizing the push-relabel algorithm on a
      shared-memory multiprocessor using up to 12 processors.
  }
}