Hi guys
Please find attached a tex file and a pdf with some improvements to the encoding (thanks
to Bill). The new encoding requires a lot less variables and can probably still be
improved. For the specific case of the N=32 D=3 fold the number of variables in a 2-local
representation is down to about 30,000 for average case sequences (still a lot but better
than a million). The asymptotic scaling of the current version is O(N^2 log N).
Let me know if you have any questions. We are considering coding up a problem generator
that would feed our solver portfolio to see how we do on this instance class (we can
probably handle up to 5,000-ish variables without going to extremes with anything).
In terms of the write-up of the results it should be relatively straightforward to
substitute the encoding here into the existing write-up. Would you prefer that we do this
or would you like to do this?
After the encoding change I would also like to have a pass through edit as I want to
emphasize in the text that the results for encoding presented are not fundamental and with
further research could likely be significantly improved (although as we mention in the
attached it's not likely we can do better than O(N log N) for a 2-local qubo-ish
representation.
Cheers
Geordie & Bill
-----Original Message-----
From: Alejandro Perdomo [mailto:aleperd@gmail.com]
Sent: Sat 1/12/2008 6:50 PM
To: Alan Aspuru-Guzik; Group; Ivan Tubert; C. Truncik; Geordie Rose
Cc: Jacob D. Biamonte; Karolina Maciag; Oscar Perdomo; Salvador Venegas; Julio Arce
Subject: Re: [Aspuru-Guzik group list] Protein Paper.- LATEST VERSION
Dear group and co-authors,
Here is the latest version of the paper with all the corrections some of you
suggested. Any comments would be highly appreciated. Going to the Arxiv soon
(sometime this week).
Thanks!
-Alejandro
On Jan 9, 2008 1:48 PM, Alan Aspuru-Guzik <alan(a)aspuru.com> wrote:
Dear group,
Here is the protein folding paper draft. We plan to send it soon to
the arxiv and PRE, so for all of you that are interested, please give
it a read and provide feedback to myself and Alejandro.
Alan
--
Alan Aspuru-Guzik
Assistant Professor
Department of Chemistry and Chemical Biology
12 Oxford Street
Harvard University
Cambridge, MA 02138
Tel: (617)384-8188
Group URL:
http://aspuru.chem.harvard.edu
_______________________________________________
Aspuru-list mailing list
Aspuru-list(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/aspuru-list