<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: Yi Cao wins the prize</title>
	<atom:link href="http://blogs.mathworks.com/contest/2007/05/16/yi-cao-wins-the-prize/feed/" rel="self" type="application/rss+xml" />
	<link>http://blogs.mathworks.com/contest/2007/05/16/yi-cao-wins-the-prize/</link>
	<description>The MATLAB Programming Contest is a semi-annual competition where contestants submit MATLAB code to try to solve a challenge.  For more information, see http://www.mathworks.com/contest/overview.html</description>
	<lastBuildDate>Sat, 21 Jan 2012 14:45:27 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.2.1</generator>
	<item>
		<title>By: Yi Cao</title>
		<link>http://blogs.mathworks.com/contest/2007/05/16/yi-cao-wins-the-prize/#comment-2111</link>
		<dc:creator>Yi Cao</dc:creator>
		<pubDate>Wed, 16 May 2007 21:59:17 +0000</pubDate>
		<guid isPermaLink="false">http://blogs.mathworks.com/contest/?p=73#comment-2111</guid>
		<description>My Twilight code was developed through the following path:

1. Improve the initial Mathwork team code by retuning a maximum score move.
2. Convert the above code to be a loop until no move can be made.
3. Store scores contributed by each move, then at the end cut final negative moves.
4. Implement a way to check available second consective moves.
5. Introduce a tuning parameter to weight the score contributed by second consecutive moves.

The 4th step makes the code is more efficient than others while the last step enables me to win the twilight prize. As I expected, this code immediately was taken by contesters to tune and to improve. The improvement continued until the final. 

During this contest, I did not have much time to work online, hence decided to work offline to concentrate on developing algorithms, for example, three consecutive moves, recurrent consecutive moves, but allmost all were beaten by well tuned leading codes. On Tuesday, I decided to submit my code (CollaborationSolver) with several improvements for the Leap prize. However, it even did not get to the top but was taken by Jin then SY to win the prize. That pushed me to find other improvements to the code and to hold the code until final minutes. I knew it is not enough to win the final by just submit the improved code. Hence, in the final hour, I kept watching the queue to find new codes and successful tuning parameters,  combined these codes with my code into five different versions, and finally submitted them within final 5 minutes.

Allthough I have won several mid-contest prizes before, but this is my first time to win a mid-contest prize before datlight and to win a Grand Prize. This is the best contest I had. Thanks to MAthwork Team to organize this intersting event.

See you all in six months.

Yi Cao</description>
		<content:encoded><![CDATA[<p>My Twilight code was developed through the following path:</p>
<p>1. Improve the initial Mathwork team code by retuning a maximum score move.<br />
2. Convert the above code to be a loop until no move can be made.<br />
3. Store scores contributed by each move, then at the end cut final negative moves.<br />
4. Implement a way to check available second consective moves.<br />
5. Introduce a tuning parameter to weight the score contributed by second consecutive moves.</p>
<p>The 4th step makes the code is more efficient than others while the last step enables me to win the twilight prize. As I expected, this code immediately was taken by contesters to tune and to improve. The improvement continued until the final. </p>
<p>During this contest, I did not have much time to work online, hence decided to work offline to concentrate on developing algorithms, for example, three consecutive moves, recurrent consecutive moves, but allmost all were beaten by well tuned leading codes. On Tuesday, I decided to submit my code (CollaborationSolver) with several improvements for the Leap prize. However, it even did not get to the top but was taken by Jin then SY to win the prize. That pushed me to find other improvements to the code and to hold the code until final minutes. I knew it is not enough to win the final by just submit the improved code. Hence, in the final hour, I kept watching the queue to find new codes and successful tuning parameters,  combined these codes with my code into five different versions, and finally submitted them within final 5 minutes.</p>
<p>Allthough I have won several mid-contest prizes before, but this is my first time to win a mid-contest prize before datlight and to win a Grand Prize. This is the best contest I had. Thanks to MAthwork Team to organize this intersting event.</p>
<p>See you all in six months.</p>
<p>Yi Cao</p>
]]></content:encoded>
	</item>
</channel>
</rss>

