From: Bill Dolinar <bdolinar@...>
Date: 2006-02-11T01:01:48+09:00
Subject: Re: [SUMMARY] Splitting the Loot (#65)

--Apple-Mail-1-218816893
Content-Transfer-Encoding: 7bit
Content-Type: text/plain;
	charset=US-ASCII;
	delsp=yes;
	format=flowed

One group of combinations I found that Luke's solution didn't work  
for was when the loot included a single value to be split to one  
partner such as:

ruby loot.rb 3 5 4 4 1 1

It will work if you expand the line in the pick function reading:

i += 1 while i < values.size && values[i] >= sum

to:

if values[lo] == sum
     i += 1
else
     i += 1 while i < values.size && values[i] >= sum
end

I'm not sure how this changed the speed, but for all solvable  
combinations of 3 partners with a sum of 9 Luke's ran in 80 seconds,  
and Manuel's ran in 105 seconds.

Bill


On Feb 9, 2006, at 4:23 PM, Manuel Kasten wrote:

> I want to mention Luke Blanchard's second solution. It is the fastest
> solution of those I tested that worked without flaws. I never ever saw
> it running longer than 0.2 seconds and I did test it a *lot*  
> (because it
> is around factor 10-20 faster than the next correct solution, and I
> thought it must fail sometimes, but it never did). I couldn't believe
> its speed. I studied his code for ca. 30 min just to understand what
> it's doing (luckily it's got a lot of comments, I wouldn't have been
> able to understand it otherwise). Then I could tell it is indeed
> correct, but its speed is nevertheless amazing.


--Apple-Mail-1-218816893--