Hi Gilles, Le 11/03/2012 16:57, Gilles Sadowski a écrit : > On Sat, Mar 10, 2012 at 07:15:14PM -0600, Lance Finney wrote: >> We used LaguerreSolver.solveAll() in version 2.1, but we never upgraded to >> 2.2. As such, we didn't notice that it was deprecated in that release. >> >> Now that 3.0 is out, we want to upgrade, but that method has been >> completely removed. >> >> Interestingly, the logic is still there in the library in an inner class: >> org.apache.commons.math3.analysis.solvers.LaguerreSolver.ComplexSolver. >> Unfortunately for us, though, ComplexSolver is now private. So, for us as >> users, we've lost access to a useful algorithm that is still in the library. >> >> Please re-enable access to this useful algorithm that provides complex >> roots of polynomial formulae. The simplest approach for our purposes would >> be to make LaguerreSolver.ComplexSolver public, but perhaps another >> approach would be considered better design. > > At first sight, I'd propose to create an "o.a.c.m.complex.solvers" package > where a "ComplexLaguerreSolver" would reside. > > However, I don't know what are the necessary and sufficient API methods that > would define a "BaseUnivariateComplexSolver" interface that would be the > counterpart of > BaseUnivariateSolver<FUNC extends UnivariateFunction> > (in package "o.a.c.m.analysis.solvers").
I guess almost the same methods could be used, except either only one solve method would be needed with a single start value, or min/max interval definition should be replaced by a two-dimensional rectangle definition. Luc > > > Best regards, > Gilles > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
