From coin-trac at coin-or.org Sun Jun 24 21:43:31 2007 From: coin-trac at coin-or.org (Volume Algorithm) Date: Sun Jun 24 21:43:34 2007 Subject: [Vol-tickets] Re: [Volume Algorithm] #3: Incorrect solutions for p0033, tp3 In-Reply-To: <050.1ae260a1b82b37682654d8acfbdad919@coin-or.org> References: <050.1ae260a1b82b37682654d8acfbdad919@coin-or.org> Message-ID: <059.2588b6650f40712aab74cfec209ceeeb@coin-or.org> #3: Incorrect solutions for p0033, tp3 -------------------------+-------------------------------------------------- Reporter: fmargot | Owner: somebody Type: defect | Status: closed Priority: major | Milestone: Component: component1 | Version: Resolution: invalid | Keywords: -------------------------+-------------------------------------------------- Changes (by ladanyi): * status: new => closed * resolution: => invalid Comment: This is entirely possible. The Volume Algorithm does not deliver an optimal solution, just a near-optimal one. If the tolerance is within the limits (likely for tp3) or the max number of iterations is reached (likely for p0033) then Vol exits. The reason for this behavior is that usually Vol converges to within ~2% of optimality very fast, but gets really slow afterwards. There is no point in continuing to run the algorithm. The iteration limit is there, because sometimes it can't even get that close, so we abort. If you reopen the ticket and attach a full output of the run I can take a look at it and decide whether it is normal or not. Thanks, --Laci -- Ticket URL: Volume Algorithm A subgradient algorithm.