A new efficient algorithm for computing Gröbner bases without reduction to zero

Colloque sur invitation. nationale.National audienceGoal of F5 Computing Gröbner bases of $(f_1,\ldots,f_m)$: Buchberger algorithm or F4 Open issue: remove useless computations. 90% of the time is spent in computing zero --> a more powerful criterion to remove useless critical pairs. Goal of F5: theoretical and practical answer