20180122, 03:16  #23 
Romulan Interpreter
"name field"
Jun 2011
Thailand
2^{4}×613 Posts 

20180122, 09:20  #24 
Just call me Henry
"David"
Sep 2007
Cambridge (GMT/BST)
3×1,979 Posts 
Very nice. Maybe someone should let numberphile know.
Also I believe papers have been published on less worthwhile subjects. 
20180123, 01:50  #25 
"Matthew Anderson"
Dec 2010
Oregon, USA
953 Posts 
I posted to the numberphile youtube page that this problem has been solved.
Regards, Matt 
20180123, 02:07  #26 
"Forget I exist"
Jul 2009
Dumbassville
20C0_{16} Posts 

20180123, 21:59  #27 
"Robert Gerbicz"
Oct 2005
Hungary
2·3·11·23 Posts 

20180316, 19:12  #28 
Feb 2018
1_{10} Posts 
How do you know that all sequences start with 1, and end with 8 (for even) or 3 (for odd)? Is there a proof?

20180317, 09:05  #29  
"Robert Gerbicz"
Oct 2005
Hungary
2×3×11×23 Posts 
Quote:
I"ve constructed all basic sequences that holds this, and then by induction we still maintain this. Note that all integers in [1,24] is free, because T(c,0) and T(c,1) doesn't use them. So we can use these small (1,3,8) integers at the endpoints. You could ask why we haven't used a shifted and reversed representation, so the sequences starts with 1,3 and 1,8; because in that case we don't know the last term of seq0 and seq1, so we can't glue the sequences. Or why we haven't used the constant 3 at the end for every sequence, because in that case 3=seq0(n)=seq1(n+1), but (n+1n)=1 is odd so the parity position condition wouldn't be true. There are some traps here. Last fiddled with by R. Gerbicz on 20180317 at 09:09 

20211102, 10:51  #30  
"Robert Gerbicz"
Oct 2005
Hungary
2·3·11·23 Posts 
Quote:
First why parity is so important here. It is trivial: for each c=24..24 we choose T(c,0) or T(c,1), exactly one of them. Say c=6,6 if you choose T(6,1) and T(6,1) then we are good, covering the 6 and 6 residues mod 49, the same happens for choosing T(6,0) and T(6,0). But why isn't this broken if we'd select T(6,0) and T(6,1), these contains for all x<=n the 49*x+6 numbers, if you choose 49*x+6 in T(6,0) and 49*x+6 in T(6,1) then the sequence is broken, the same integer appears twice. The trick is that in seq0 and seq1 all x<=n numbers appear in the same parity position, this forces that if one seq you'll choose 49*x+6 then in the other it'll be 49*x6. Let res=47 so we want the solution for N=49*n+47, when we know the good sequence pair for n and n+1. Just for the example remain at c=6,6 the extra n+1 term in seq1 is in the parity position of n+1 mod 2, because with seq0 these contain the 1..n integers in the same parity position, so the largest two terms in T(6,1) and T(6,1) are: y0=49*(n+1)+(1)^(n+1)*6 and y1=49*(n+1)+(1)^(n+1)*(6). First notice that all terms except y0,y1 in these two seq are at most 49*n+24, so not larger than N=49*n+47 and this is true in general, because we recurse for res>=24. What is really not explained, but in the code (quite hidden), that you need to know also the parity of n to know these two terms: say n is even then y0=49*n+43 and y1=49*n+55, but N=49*n+47, so here y0 should be selected but not y1 and this is forced, so you have to select T(6,1) and T(6,0). What is also important: you have to maintain the parity condition for N (to make a working induction), how do you ensure that this will hold? When you glue the sequences you're using 1 integer or n or n+1 integers in a block, if you know the parity of n you'll know the parity between subsequences, and a smaller trick: for a subsequence it is enough to see that the first term goes to the same parity position. ps. in some places we used the reverse of the T(c,) subsequence to give more breath to the method. 

20211112, 14:34  #31 
"Andrew Booker"
Mar 2013
3·29 Posts 
You really should write this up for publication, if only to provide a reference point and a proof that's more than a quick description on an internet forum. You're welcome to send it to me for consideration at JNT, and there are loads of other good options. (The American Mathematical Monthly comes to mind, given that the problem has broad interest and the proof is also very accessible.)

20211115, 15:12  #32  
"Robert Gerbicz"
Oct 2005
Hungary
2×3×11×23 Posts 
Quote:
It is a somewhat recreational Maths problem, but worth a paper. 

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
The SquareDiff problem  firejuggler  Puzzles  1  20180124 23:27 
Square root of 3  Damian  Math  3  20100101 01:56 
Square of Primes  davar55  Puzzles  9  20080521 12:54 
red square  Fusion_power  Puzzles  14  20080425 11:37 
How often is 2^p1 squarefree  ZetaFlux  Math  16  20051214 06:55 