Zak Tonks: Cylindrical algebraic decomposition: algorithmic real algebraic geometry.
Abstract:
Cylindrical Algebraic Decomposition is a powerful method in
computational real algebraic geometry, being `merely#
doubly exponential in the number of variables, rather than
Tarski's original method, whose complexity was not
expressible by any fixed tower of exponentials. But the
original Collins method was very expensive in practice.
McCallum's improvement was substantial, but had the
drawback that it could occasionally fail (input not
well-oriented). Recent improvements to McCallum have made
the computation more problem-sensitive, rather than being a
sledgehammer that solves all problems about the same
polynomials. We describe these, and also the
recently-verified Lazard method, which is similar to
McCallum but has the advantage of never failing.