[source]

From this

n!/(m!*(n-m)!)
]]> http://qvaq.com/algorithms/3284 mezod Fri, 21 Aug 2009 14:50:18 +0200 http://qvaq.com/algorithms/3284 http://qvaq.com/algorithms/ sorry i didnt understand this: n=4 -> x=6, n=6 -> x=20, n=8 -> x=70, n=10 -> x=252.<br/>which example of numbers are you using? which example of numbers are you using?]]> http://qvaq.com/algorithms/3179 mwikegard Fri, 21 Aug 2009 05:06:55 +0200 http://qvaq.com/algorithms/3179 http://qvaq.com/algorithms/ I this should be done for a large set of numbers i would select a totally different approach.<br/>I would try to maintain a correct mean value when picking values.<br/>Perhaps this is best done in SQL.<br/>first calculate mean value.<br/>select all values with values=mean<br/>then select +-n pairs until you reach half total count.<br/>This would require a Normal distribution</p><p> I would try to maintain a correct mean value when picking values.
Perhaps this is best done in SQL.
first calculate mean value.
select all values with values=mean
then select +-n pairs until you reach half total count.
This would require a Normal distribution

]]> http://qvaq.com/algorithms/3178 mwikegard Fri, 21 Aug 2009 03:54:10 +0200 http://qvaq.com/algorithms/3178 http://qvaq.com/algorithms/ Instead of writing everything i found a neat Java class here<br/><a href="http://www.merriampark.com/comb.htm" target="_blank" rel="nofollow">http://www.merriampark.com/comb.htm</a></p><p>if we summarize our numbers and remember the combination closest to half grand total we are done.<br/>At least it is np-c.<br/> http://www.merriampark.com/comb.htm

if we summarize our numbers and remember the combination closest to half grand total we are done.
At least it is np-c.
]]> http://qvaq.com/algorithms/3177 mwikegard Fri, 21 Aug 2009 03:48:06 +0200 http://qvaq.com/algorithms/3177 http://qvaq.com/algorithms/ Ok, so I was a little fast when I said it would be best with a binary tree. It would be a lot of branches to cut.<br/>What we have is a combinational problem.<br/>summarize n/2 numbers out of n numbers and select the one closest to Total sum / 2<br/>Ok now we need to pick out the right combinations.<br/>The number of combinations x would be (n over m) where m=n/2<br/>n=4 -> x=6, n=6 -> x=20, n=8 -> x=70, n=10 -> x=252.<br/>Continued> What we have is a combinational problem.
summarize n/2 numbers out of n numbers and select the one closest to Total sum / 2
Ok now we need to pick out the right combinations.
The number of combinations x would be (n over m) where m=n/2
n=4 -> x=6, n=6 -> x=20, n=8 -> x=70, n=10 -> x=252.
Continued>]]> http://qvaq.com/algorithms/3090 mezod Thu, 20 Aug 2009 21:31:43 +0200 http://qvaq.com/algorithms/3090 http://qvaq.com/algorithms/ @<a href="/users/mwikegard">mwikegard</a> could u give an example of branch cutting?</p><p>great there's someone else to discuss with! mwikegard could u give an example of branch cutting?

great there's someone else to discuss with!]]> http://qvaq.com/algorithms/2967 bjorn Thu, 20 Aug 2009 13:48:53 +0200 http://qvaq.com/algorithms/2967 http://qvaq.com/algorithms/ Good question. Not sure whether it matters, though, since the algorithm generates the wrong results for some other cases (1 2 3 4 5 6 7 8 9 100, for instance). http://qvaq.com/algorithms/2966 mwikegard Thu, 20 Aug 2009 13:44:32 +0200 http://qvaq.com/algorithms/2966 http://qvaq.com/algorithms/ I think it must be a binary tree solution. <br/>Something like @<a href="/users/mezod">mezod</a> suggested.<br/>You can cut-off some branches but thats it.<br/>Remember sorting have a cost also.<br/> Something like @mezod suggested.
You can cut-off some branches but thats it.
Remember sorting have a cost also.
]]> http://qvaq.com/algorithms/2965 murtlest Thu, 20 Aug 2009 13:35:47 +0200 http://qvaq.com/algorithms/2965 http://qvaq.com/algorithms/ Ah great idea.</p><p>One of the rules for a problem to be np is that it should be easy to very the solution once you've found it. Is that true here?

One of the rules for a problem to be np is that it should be easy to very the solution once you've found it. Is that true here?]]> http://qvaq.com/algorithms/2964 bjorn Thu, 20 Aug 2009 13:31:01 +0200 http://qvaq.com/algorithms/2964 http://qvaq.com/algorithms/ Another version that performs better in the 1699... case but worse in others:</p><p>- Sort numbers.<br/>- Put the first and last number into bucket 1. Remove them from the set of numbers.<br/>- Put the new first and last number into bucket 2. Remove them too.<br/>- Repeat until there are two values left.<br/>- Decide where the last two numbers should go.<br/>

- Sort numbers.
- Put the first and last number into bucket 1. Remove them from the set of numbers.
- Put the new first and last number into bucket 2. Remove them too.
- Repeat until there are two values left.
- Decide where the last two numbers should go.
]]> http://qvaq.com/algorithms/2958 bjorn Thu, 20 Aug 2009 12:40:11 +0200 http://qvaq.com/algorithms/2958 http://qvaq.com/algorithms/ Indeed, but that is what makes the problem interesting, isn't it? ;-) http://qvaq.com/algorithms/2956 murtlest Thu, 20 Aug 2009 12:30:30 +0200 http://qvaq.com/algorithms/2956 http://qvaq.com/algorithms/ I think that we need to sum all numbers and divide by two to know what we aim for.</p><p>Still think it''s np-complete though ;(

Still think it''s np-complete though ;(]]> http://qvaq.com/algorithms/2954 murtlest Thu, 20 Aug 2009 12:24:19 +0200 http://qvaq.com/algorithms/2954 http://qvaq.com/algorithms/ Ah hehe, yes with the other example the sums are never equal. http://qvaq.com/algorithms/2952 bjorn Thu, 20 Aug 2009 12:20:31 +0200 http://qvaq.com/algorithms/2952 http://qvaq.com/algorithms/ I only tried with 1 2 3 4 5 9, with which the new version worked better... http://qvaq.com/algorithms/2951 bjorn Thu, 20 Aug 2009 12:17:17 +0200 http://qvaq.com/algorithms/2951 http://qvaq.com/algorithms/ Darn, should have tried it with your example too... http://qvaq.com/algorithms/2950 bjorn Thu, 20 Aug 2009 12:10:02 +0200 http://qvaq.com/algorithms/2950 http://qvaq.com/algorithms/ - Sort numbers descending.<br/>- For each number:<br/>- If bucket 1 is full, put in bucket 2.<br/>- Otherwise, if bucket 2 is full, put in bucket 1.<br/>- Otherwise, if sum of bucket 1 == sum of bucket 2 and items in bucket 1 > items in bucket2, put in bucket 1<br/>- Otherwise, if sum of bucket 1 == sum of bucket 2, put in bucket 2<br/>- Otherwise, if sum of bucket 1 > sum of bucket 2, put in bucket 2.<br/>- Otherwise, put in bucket 1. - For each number:
- If bucket 1 is full, put in bucket 2.
- Otherwise, if bucket 2 is full, put in bucket 1.
- Otherwise, if sum of bucket 1 == sum of bucket 2 and items in bucket 1 > items in bucket2, put in bucket 1
- Otherwise, if sum of bucket 1 == sum of bucket 2, put in bucket 2
- Otherwise, if sum of bucket 1 > sum of bucket 2, put in bucket 2.
- Otherwise, put in bucket 1.]]> http://qvaq.com/algorithms/2939 murtlest Thu, 20 Aug 2009 11:31:55 +0200 http://qvaq.com/algorithms/2939 http://qvaq.com/algorithms/ @<a href="/users/bjorn">bjorn</a>: That was the solution we thought was right first too. But @<a href="/users/mezod">mezod</a> found out that it doesn't work for this set of numbers for example:<br/>1699 1504 1446 1350 1334 1034</p><p>Suggested algorithm would create these buckets:<br/>1: 1699 + 1350 + 1034 = 4083<br/>2: 1504 + 1446 + 1334 = 4284</p><p>But this is better (don't know if it's the best):<br/>1: 1699 + 1446 + 1034 = 4179<br/>2: 1504 + 1350 + 1334 = 4188<br/> bjorn: That was the solution we thought was right first too. But @mezod found out that it doesn't work for this set of numbers for example:
1699 1504 1446 1350 1334 1034

Suggested algorithm would create these buckets:
1: 1699 + 1350 + 1034 = 4083
2: 1504 + 1446 + 1334 = 4284

But this is better (don't know if it's the best):
1: 1699 + 1446 + 1034 = 4179
2: 1504 + 1350 + 1334 = 4188
]]> http://qvaq.com/algorithms/2936 bjorn Thu, 20 Aug 2009 11:07:17 +0200 http://qvaq.com/algorithms/2936 http://qvaq.com/algorithms/ Possible solution, but it is quite early, so I may be missing something...</p><p>- Sort numbers descending.<br/>- For each number:<br/>- If bucket 1 is full, put in bucket 2.<br/>- Otherwise, if bucket 2 is full, put in bucket 1.<br/>- Otherwise, if sum of bucket 1 > sum of bucket 2, put in bucket 2.<br/>- Otherwise, put in bucket 1.</p><p>Seems to perform decently.

- Sort numbers descending.
- For each number:
- If bucket 1 is full, put in bucket 2.
- Otherwise, if bucket 2 is full, put in bucket 1.
- Otherwise, if sum of bucket 1 > sum of bucket 2, put in bucket 2.
- Otherwise, put in bucket 1.

Seems to perform decently.]]> http://qvaq.com/algorithms/2864 murtlest Wed, 19 Aug 2009 21:16:47 +0200 http://qvaq.com/algorithms/2864 http://qvaq.com/algorithms/ Ah i see. That's a real need for this algortihm. I though he just wanted you to solve his homework problem ;) Now i really want to find out if it is np-complete. http://qvaq.com/algorithms/2860 mezod Wed, 19 Aug 2009 21:12:53 +0200 http://qvaq.com/algorithms/2860 http://qvaq.com/algorithms/ yes, he needed that algorithm for equilibrating teams (i suppose as game rating) something like the !fair command in #cs.gather community i guess...wondering which algorithm do they use http://qvaq.com/algorithms/2857 murtlest Wed, 19 Aug 2009 21:09:43 +0200 http://qvaq.com/algorithms/2857 http://qvaq.com/algorithms/ Your first solution was more beautiful, but then again not working ;D http://qvaq.com/algorithms/2853 mezod Wed, 19 Aug 2009 21:02:45 +0200 http://qvaq.com/algorithms/2853 http://qvaq.com/algorithms/ agree, he wanted it for x=10...but anyway its not a code to be proud of ;D http://qvaq.com/algorithms/2851 murtlest Wed, 19 Aug 2009 20:59:13 +0200 http://qvaq.com/algorithms/2851 http://qvaq.com/algorithms/ ah nice, i think i understand most of it :) it's pretty brute force though, wouldn't want to run that with like a million numbers ;) http://qvaq.com/algorithms/2848 mezod Wed, 19 Aug 2009 20:50:47 +0200 http://qvaq.com/algorithms/2848 http://qvaq.com/algorithms/ i implemented the code in pseudocode for that guy that was asking yesterday, its just he was spanish also and i added the comments in spanish (tho he didnt understand a shit anyway).<br/>This is the code i did in a fast way yesterday night...hope u can understand something :p<br/> <a href="http://pastebin.com/m265b77a2" target="_blank" rel="nofollow">http://pastebin.com/m265b77a2</a><br/>"sumale el resto a grupoA" means add the rest to group A. ignore the comments...note it was a draft :p This is the code i did in a fast way yesterday night...hope u can understand something :p
http://pastebin.com/m265b77a2
"sumale el resto a grupoA" means add the rest to group A. ignore the comments...note it was a draft :p]]> http://qvaq.com/algorithms/2845 murtlest Wed, 19 Aug 2009 20:45:41 +0200 http://qvaq.com/algorithms/2845 http://qvaq.com/algorithms/ Probably to knapsack:<br/><a href="http://en.wikipedia.org/wiki/Knapsack_problem" target="_blank" rel="nofollow">http://en.wikipedia.org/wiki/Knapsack_problem</a> http://en.wikipedia.org/wiki/Knapsack_problem]]> http://qvaq.com/algorithms/2837 mezod Wed, 19 Aug 2009 20:33:41 +0200 http://qvaq.com/algorithms/2837 http://qvaq.com/algorithms/ guess so http://qvaq.com/algorithms/2834 murtlest Wed, 19 Aug 2009 20:32:12 +0200 http://qvaq.com/algorithms/2834 http://qvaq.com/algorithms/ Then we should be able to reduce it to be np complete? http://qvaq.com/algorithms/2832 mezod Wed, 19 Aug 2009 20:29:53 +0200 http://qvaq.com/algorithms/2832 http://qvaq.com/algorithms/ for now yes :| http://qvaq.com/algorithms/2828 murtlest Wed, 19 Aug 2009 20:24:11 +0200 http://qvaq.com/algorithms/2828 http://qvaq.com/algorithms/ Okay :( So its NP complete? http://qvaq.com/algorithms/2827 mezod Wed, 19 Aug 2009 20:18:40 +0200 http://qvaq.com/algorithms/2827 http://qvaq.com/algorithms/ well i did a recursive algorithm using backtracking...it checks all the possibilities and it takes the best one..but its not the fastest...i think there's no better solution :( http://qvaq.com/algorithms/2819 murtlest Wed, 19 Aug 2009 19:48:14 +0200 http://qvaq.com/algorithms/2819 http://qvaq.com/algorithms/ @<a href="/users/mezod">mezod</a>: Hehe, damn, now i don't even know the answer. Have you figured it out? I'm thinking but havn't come up with anything better than your solution. mezod: Hehe, damn, now i don't even know the answer. Have you figured it out? I'm thinking but havn't come up with anything better than your solution.]]> http://qvaq.com/algorithms/2689 mezod Wed, 19 Aug 2009 02:55:32 +0200 http://qvaq.com/algorithms/2689 http://qvaq.com/algorithms/ this will be my first post! this contest sounds familiar to me ;D<br/>i have to admit that the algorithm i suggested in #php wasn't right 100% it was a "greedy algorithm". <br/>its late and i should go to sleep, ill just give an example to murtlest so he can check the algorithm doesnt work for all the cases: (post the algorithm if u have time. i dont ;D)<br/>1699 1504 1446 1350 1334 1034.<br/>did something else, post tomorrow! i have to admit that the algorithm i suggested in #php wasn't right 100% it was a "greedy algorithm".
its late and i should go to sleep, ill just give an example to murtlest so he can check the algorithm doesnt work for all the cases: (post the algorithm if u have time. i dont ;D)
1699 1504 1446 1350 1334 1034.
did something else, post tomorrow!]]> http://qvaq.com/algorithms/2632 murtlest Tue, 18 Aug 2009 23:55:34 +0200 http://qvaq.com/algorithms/2632 http://qvaq.com/algorithms/ Todays algorithm contest:<br/>You have 10 numbers (posetive integers). You need to put these into two buckets, so that the difference between the buckets (the sum of the numbers in the buckets) are as low as possible. There should be 5 numbers in each bucket. You have 10 numbers (posetive integers). You need to put these into two buckets, so that the difference between the buckets (the sum of the numbers in the buckets) are as low as possible. There should be 5 numbers in each bucket. ]]>