From: "Bruno Luong" <b.luong@fogale.findmycountry>
Newsgroups: comp.soft-sys.matlab
Subject: Re: Derangement: efficient full permutation
Date: Fri, 18 Feb 2011 15:53:04 +0000 (UTC)
Organization: FOGALE nanotech
Lines: 20
Message-ID: <ijm4l0$89i$>
References: <ii8ut8$3fb$> <ii90lg$qmd$> <> <iin0v4$ic2$> <iin5l5$h5n$> <iin77n$rje$> <iio9b0$m0i$> <ijk9se$gcb$> <ijlldg$181$> <ijln5n$o67$> <ijm2l7$pco$>
Reply-To: "Bruno Luong" <b.luong@fogale.findmycountry>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
X-Trace: 1298044384 8498 (18 Feb 2011 15:53:04 GMT)
NNTP-Posting-Date: Fri, 18 Feb 2011 15:53:04 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 390839
Xref: comp.soft-sys.matlab:710928

"Derek O'Connor" wrote in message <ijm2l7$pco$>...

> I have just noticed that I omitted the average time for GRDsta2mex, which is Roger's second method with randperm replaced by Jan's GRPmex.
> GRDsta2mex:  0.21258 secs,  which is more that twice GRDmex, i.e., GRDrej with GRPmex.
> Can you tell me exactly what is wrong with Jan's new derangement algorithm and give me counter-examples that I can test or examine. 
> Please remember, I am not interested in n < 20. I am interested in generating random permutations of various types with n >> 10^6.

Derek, why arguing back and forth I think I have clearly elaborated my position.

You are free to trust Jan's algorithm at larger n with experimental testing. That's after all your choice.

Here are the facts that I stick with:
1) It's unproven algorithm.
2) It's biased at small n.