From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 13814 invoked by alias); 15 May 2010 08:11:10 -0000 Received: (qmail 13801 invoked by uid 22791); 15 May 2010 08:11:09 -0000 X-SWARE-Spam-Status: No, hits=0.8 required=5.0 tests=BAYES_50 X-Spam-Check-By: sourceware.org Received: from up.nbi.dk (HELO mail2.nbi.dk) (130.225.212.6) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Sat, 15 May 2010 08:11:05 +0000 Received: from [192.168.1.109] (tru75-4-82-227-169-50.fbx.proxad.net [82.227.169.50]) by mail2.nbi.dk (Postfix) with ESMTP id 4C63B3C922 for ; Sat, 15 May 2010 10:11:02 +0200 (CEST) Message-ID: <4BEE5711.4050902@webdrake.net> Date: Sat, 15 May 2010 08:11:00 -0000 From: Joseph Wakeling User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.9) Gecko/20100423 Thunderbird/3.0.4 MIME-Version: 1.0 To: gsl-discuss@sourceware.org Subject: Re: Random sampling extension for GSL References: <4BE99D2E.6060703@webdrake.net> In-Reply-To: <4BE99D2E.6060703@webdrake.net> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Mailing-List: contact gsl-discuss-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gsl-discuss-owner@sourceware.org X-SW-Source: 2010-q2/txt/msg00018.txt.bz2 On 05/11/2010 08:08 PM, Joseph Wakeling wrote: > As discussed in earlier emails, I've created a project page at GitHub > for the new random sampling extension: > http://github.com/WebDrake/GrSL Further to the above: the system now implements Vitter's "Algorithm D" and has a new "choose" function that uses the sampling classes. Running the latter using the former for the sampling functionality shows a sizeable speedup over the gsl_ran_choose function. This is early days and it still all needs much more testing. What would be really useful at this stage is if there could be confirmation as to whether the basic design -- API etc. -- makes a good fit with GSL or if there are recommended changes. Still to come: Nair's "Algorithm E" plus various other less-superior but historically relevant ones (e.g. Knuth's Algorithm S). Best wishes, -- Joe