mersenneforum.org Mondrian art puzzles - error in Numberphile video
 Register FAQ Search Today's Posts Mark Forums Read

 2016-11-27, 20:31 #1 cuBerBruce     Aug 2012 Mass., USA 2·3·53 Posts Mondrian art puzzles - error in Numberphile video The recent Numberphile video about Mondrian art puzzles has an error. It claims that the best Mondrian score for an 18x18 square is 10. But I've achieved a Mondrian score of 8 for it. Code:  ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ? ? ? ? ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ? ? ? ? % % % < < < < < # # # # # # ? ? ? ? % % % < < < < < # # # # # # ? ? ? ? % % % < < < < < # # # # # # ? ? ? ? % % % < < < < < # # # # # # ? ? ? ? % % % < < < < < # # # # # # ? ? ? ? % % % < < < < < # # # # # # ? ? ? ? % % % > > > > > > > + + + + ? ? ? ? % % % > > > > > > > + + + + = = = = % % % > > > > > > > + + + + = = = = % % % > > > > > > > + + + + = = = = % % % > > > > > > > + + + + = = = = @ @ @ @ @ @ @ @ @ @ + + + + = = = = @ @ @ @ @ @ @ @ @ @ + + + + = = = = @ @ @ @ @ @ @ @ @ @ + + + + = = = = 
 2016-11-27, 21:10 #2 CRGreathouse     Aug 2006 32·5·7·19 Posts I agree! You should send in a correction to A276523 and maybe send Ed an email.
2016-11-27, 21:11   #3
science_man_88

"Forget I exist"
Jul 2009
Dumbassville

203008 Posts

Quote:
 Originally Posted by cuBerBruce The recent Numberphile video about Mondrian art puzzles has an error. It claims that the best Mondrian score for an 18x18 square is 10. But I've achieved a Mondrian score of 8 for it.
the description of that video also said that a new lowest 25 by 25 was found cool find either way. edit: I of course have mostly useless information as to how to make a solution other than if T(x)<n^2<T(x+1) then at most x distinct areas can exist.

Last fiddled with by science_man_88 on 2016-11-27 at 22:10

2016-11-27, 22:08   #4
cuBerBruce

Aug 2012
Mass., USA

2·3·53 Posts

Quote:
 Originally Posted by CRGreathouse I agree! You should send in a correction to A276523 and maybe send Ed an email.
I've now sent Ed an email.

 2016-11-28, 19:43 #5 cuBerBruce     Aug 2012 Mass., USA 4768 Posts Ed has informed me that A276523 has now been corrected. The correction also include new values for 15x15 and 19x19 cases, found by Ed after further checking.
2016-11-28, 20:12   #6
CRGreathouse

Aug 2006

32×5×7×19 Posts

Quote:
 Originally Posted by cuBerBruce Ed has informed me that A276523 has now been corrected. The correction also include new values for 15x15 and 19x19 cases, found by Ed after further checking.
Thanks to both of you! I approved the changes.

 2016-11-28, 23:33 #7 R. Gerbicz     "Robert Gerbicz" Oct 2005 Hungary 5BA16 Posts Really fascinating puzzle! My exahaustive code gives the following better solutions (these are optimal): a(14)=6 (!!!) Code: aaaaaaaaaabbbb aaaaaaaaaabbbb aaaaaaaaaabbbb cccdddddddbbbb cccdddddddbbbb cccdddddddbbbb cccdddddddbbbb cccdddddddbbbb ccceeeeeffffff ccceeeeeffffff ccceeeeeffffff ccceeeeeffffff ccceeeeeffffff ccceeeeeffffff a(16)=8 Code: aaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaa bbbbbbbbbbccccdd bbbbbbbbbbccccdd bbbbbbbbbbccccdd eeefffffffccccdd eeefffffffccccdd eeefffffffccccdd eeefffffffccccdd eeefffffffccccdd eeeggggghhhhhhdd eeeggggghhhhhhdd eeeggggghhhhhhdd eeeggggghhhhhhdd eeeggggghhhhhhdd eeeggggghhhhhhdd a(23)=8 Code: aaaaaaaaaaaaaaaaaabbbbb aaaaaaaaaaaaaaaaaabbbbb aaaaaaaaaaaaaaaaaabbbbb aaaaaaaaaaaaaaaaaabbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb cccccccceeeeffffffbbbbb ddddddddeeeeffffffbbbbb ddddddddeeeeffffffbbbbb ddddddddeeeeffffffbbbbb ddddddddeeeeggggggggggg ddddddddeeeeggggggggggg ddddddddeeeeggggggggggg ddddddddeeeeggggggggggg ddddddddeeeeggggggggggg ddddddddeeeeggggggggggg ddddddddeeeeggggggggggg a(25)=10 Code: aaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaa bbbbbbbbbbbbbbbbbbbbccccc bbbbbbbbbbbbbbbbbbbbccccc ddeeeeeeeffffkkkkkkkccccc ddeeeeeeeffffkkkkkkkccccc ddeeeeeeeffffkkkkkkkccccc ddeeeeeeeffffkkkkkkkccccc ddeeeeeeeffffkkkkkkkccccc ddeeeeeeeffffkkkkkkkccccc ddeeeeeeefffflllllllllmmm ddggghhhhfffflllllllllmmm ddggghhhhfffflllllllllmmm ddggghhhhfffflllllllllmmm ddggghhhhfffflllllllllmmm ddggghhhhjjjjjjjjnnnnnmmm ddggghhhhjjjjjjjjnnnnnmmm ddggghhhhjjjjjjjjnnnnnmmm ddggghhhhjjjjjjjjnnnnnmmm ddggghhhhjjjjjjjjnnnnnmmm ddggghhhhjjjjjjjjnnnnnmmm ddgggiiiiiiiiiiiinnnnnmmm ddgggiiiiiiiiiiiinnnnnmmm ddgggiiiiiiiiiiiinnnnnmmm ddgggiiiiiiiiiiiinnnnnmmm a(26)=9 Code: aaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaaa bbbbbbbbbbbbbbbbccccccccdd bbbbbbbbbbbbbbbbccccccccdd bbbbbbbbbbbbbbbbccccccccdd eeegggggggggggggccccccccdd eeegggggggggggggccccccccdd eeegggggggggggggccccccccdd eeegggggggggggggccccccccdd eeehhhhiiiiiiiiiiiiiiiiidd eeehhhhiiiiiiiiiiiiiiiiidd eeehhhhiiiiiiiiiiiiiiiiidd eeehhhhjjjjjkkkkkkkkkkkkdd eeehhhhjjjjjkkkkkkkkkkkkdd eeehhhhjjjjjkkkkkkkkkkkkdd eeehhhhjjjjjkkkkkkkkkkkkdd eeehhhhjjjjjlllllllmmmmmdd eeehhhhjjjjjlllllllmmmmmdd eeehhhhjjjjjlllllllmmmmmdd eeehhhhjjjjjlllllllmmmmmdd eeehhhhjjjjjlllllllmmmmmdd eeehhhhjjjjjlllllllmmmmmdd eeehhhhjjjjjlllllllmmmmmdd fffffffffffffffffffmmmmmdd fffffffffffffffffffmmmmmdd fffffffffffffffffffmmmmmdd a(27)=10 Code: aaaaaaaaaaaaaaaaaaaaaaaabbb aaaaaaaaaaaaaaaaaaaaaaaabbb ccdddddddddddddddeeeefffbbb ccdddddddddddddddeeeefffbbb ccdddddddddddddddeeeefffbbb ccgghhhhhhiiiiiiieeeefffbbb ccgghhhhhhiiiiiiieeeefffbbb ccgghhhhhhiiiiiiieeeefffbbb ccgghhhhhhiiiiiiieeeefffbbb ccgghhhhhhiiiiiiieeeefffbbb ccgghhhhhhiiiiiiieeeefffbbb ccgghhhhhhiiiiiiieeeefffbbb ccggjjjjjjlllllllllllfffbbb ccggjjjjjjlllllllllllfffbbb ccggjjjjjjlllllllllllfffbbb ccggjjjjjjlllllllllllfffbbb ccggjjjjjjmmmmmmmmmmmmnnnnn ccggjjjjjjmmmmmmmmmmmmnnnnn ccggjjjjjjmmmmmmmmmmmmnnnnn ccggjjjjjjmmmmmmmmmmmmnnnnn ccggkkkkkkkkoooooooooonnnnn ccggkkkkkkkkoooooooooonnnnn ccggkkkkkkkkoooooooooonnnnn ccggkkkkkkkkoooooooooonnnnn ccggkkkkkkkkoooooooooonnnnn ccggppppppppppppppppppppppp ccggppppppppppppppppppppppp a(28)=9 Code: aaaaaaaaaaaaaaaaaaaaaaaaaaaa aaaaaaaaaaaaaaaaaaaaaaaaaaaa bbbbbbbbbbbbbbbbbbbbbbbbbccc bbbbbbbbbbbbbbbbbbbbbbbbbccc ddddddddddddddddddeeeffffccc ddddddddddddddddddeeeffffccc ddddddddddddddddddeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhjjjjjjkkkkkkkeeeffffccc hhhhhllllmmmmmmmmmeeeffffccc hhhhhllllmmmmmmmmmeeeffffccc iiiiillllmmmmmmmmmeeeffffccc iiiiillllmmmmmmmmmeeeggggggg iiiiillllmmmmmmmmmeeeggggggg iiiiillllmmmmmmmmmeeeggggggg iiiiillllnnnnnnnnnnnnggggggg iiiiillllnnnnnnnnnnnnggggggg iiiiillllnnnnnnnnnnnnggggggg iiiiillllnnnnnnnnnnnnggggggg iiiiillllooooooooooooooooooo iiiiillllooooooooooooooooooo iiiiillllooooooooooooooooooo
 2016-11-29, 01:51 #8 science_man_88     "Forget I exist" Jul 2009 Dumbassville 26·131 Posts you can upper bound if for even n by using either 2n if the solution for n/2 is greater than n\4 or 4 times the solution for n\2 otherwise. edit: so for example using the 17 by 17 bound in the video is 8 so we can say with confidence that for 34 by 34 the minimum solution is upper bounded by 32. and for the n=23 example above we can say that now for n=46 the minimal solution is not greater than 32 as well. sorry adding trivialities to the thread. Last fiddled with by science_man_88 on 2016-11-29 at 01:54
2016-11-29, 04:52   #9
cuBerBruce

Aug 2012
Mass., USA

2×3×53 Posts

Quote:
 Originally Posted by R. Gerbicz Really fascinating puzzle! My exahaustive code gives the following better solutions (these are optimal):
Good job, R. Gerbicz! Wow, I am surprised that as many as 6 cases thought to be proven optimal have been improved. I haven't rigorously checked your solutions, but I've verified with my own solver code that the scores for at least the smaller ones (up to 23x23) are achievable.

A little bit more about my initial find...

This web page listed what had been believed to be the optimal Mondrian scores for 4x4 up to 17x17. So after I found solutions that I thought were optimal for squares up to 17x17, I looked for a solution for the 18x18 square with a score that matched the 17x17 value.

After my program generated many solutions (having a score of 8), I tried to find one manually on my own. I struggled for awhile, so I finally peeked at part of one of the solutions found by my computer. I was able to complete the 18x18 square by trial and error from the set of rectangles the computer was using, and that is the solution I posted.

It probably wasn't until a couple days later or so that I realized that this solution had a score that was better by 2 than what was being claimed to be the best possible score. Until then, I didn't think my solution was anything special. Of course, I then realized I should post my solution online right away.

 2016-11-29, 19:44 #10 EdPeggJr   Nov 2016 24 Posts Many thanks for the improved solutions. I found a few bugs in my coding. I'll be posting an update at http://demonstrations.wolfram.com/MondrianArtProblem/ in the near future.
2016-11-30, 21:10   #11
Batalov

"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

3×43×73 Posts

Quote:
 Originally Posted by EdPeggJr Many thanks for the improved solutions. I found a few bugs in my coding. I'll be posting an update at http://demonstrations.wolfram.com/MondrianArtProblem/ in the near future.
Welcome to the forum!

 Similar Threads Thread Thread Starter Forum Replies Last Post lavalamp Lounge 68 2018-09-09 19:01 science_man_88 science_man_88 0 2017-11-17 21:37 Mini-Geek Lounge 3 2017-09-08 06:18 firejuggler Prime Gap Searches 8 2017-07-19 20:22 Mini-Geek Lounge 5 2014-10-29 07:28

All times are UTC. The time now is 07:37.

Fri May 7 07:37:36 UTC 2021 up 29 days, 2:18, 0 users, load averages: 1.68, 1.95, 2.11