Sorry folks, I've been very busy lately and I just noticed the mysql database was badly corrupted on Sept. 6th, 2009. The database is updated daily, but all the backups on Sept. 6th and after are unusable. I think the only post lost was Tom Rokicki's.

I try my best to make sure everything is working but this one slipped through the cracks. Somehow the mysql database ballooned in size to over 2 gigabytes. After that happened the subsequent databases were not backed up correctly.

It would be a good idea for any posts to be buffered in some way before uploading to the forum, especially long ones.

I trust that I may be forgiven for being slightly off topic. After all, a watermelon Rubik's cube is not very mathematical. But still, it's an interesting concept.

http://www.watermelon.org/FeaturedRecipe.asp

I am in no way connected with the National Watermelon Promotion Board.

I have done my own independent breadth-first search of the edge group using the face-turn metric. I used symmetry/antisymmetry equivalence classes to reduce the number of elements in the search space. I confirm the "Unique mod M+inv" values for this group/metric that Rokicki reported in 2004.

I reduced the "coordinate space" for the search to 5022205*2048=10285475840 elements by using symmetry/antisymmetry equivalence classes of the edge permutation group. (This gives a much more compact overall coordinate space than using an edge orientation sym-coordinate, at a cost of more time required to calculate representative elements. This allowed me to keep track of reached equivalence classes with a ~1.3 GB bitvector in RAM and 5022205 KB disk files to keep track of distances.)

Just finished running out to a distance of 13 in the face turn metric.

First, the positions at exactly that distance:

 d   mod M + inv          mod M       positions
-- ------------- -------------- ---------------
 0             1              1               1
 1             2              2              18
 2             8              9             243
 3            48             75            3240
 4           509            934           43239
 5          6198          12077          574908
 6         80178         159131         7618438
 7       1053077        2101575       100803036

I've computed the count of positions out to 14 quarter turns.

First, positions at exactly the given distance:

 d  mod M + inv         mod M      positions
-- ------------ ------------- --------------
 0            1             1              1
 1            1             1             12
 2            5             5            114
 3           17            25           1068
 4          130           219          10011
 5         1031          1978          93840
 6         9393         18395         878880
 7        86183        171529        8221632
 8       802788       1601725       76843595

I just completed exploring all positions of the cube out to depth 12 in the face turn metric.

The first table is the count of positions with exactly the given depth.

 d  mod M + inv        mod M      positions
-- ------------ ------------ --------------
 0            1            1              1
 1            2            2             18
 2            8            9            243
 3           48           75           3240
 4          509          934          43239
 5         6198        12077         574908
 6        80178       159131        7618438
 7      1053077      2101575      100803036

With 25,000 QTM cosets proved to have a distance of 25 or less,
we have shown that there are no positions that require 30 or more
quarter turns to solve. All these sets were run on my personal
machines, mostly on a new single i7 920 box.

These sets cover more than 4e16 of the total 4e19 cube positions,
when inverses and symmetries are taken into account, and no new
distance-26 position was found. This indicates that distance-26
positions are extremely rare; I conjecture the known one is the
only distance-26 position.

In order to take the step to a proof of 28, I would need a couple

Any inappropriate link (i.e. not math and/or puzzle related) will be deleted. I'd like to keep the forum completely free of ads with the sole exception of ads for books about puzzles, or at least limited to materials appropriate for the site.

For newbies or younger readers:

If you find some of the posts are too difficult to understand please go ahead and ask questions! The people here are willing to help explain things.

Mark

ok here is the game:
your opponent has a secret number that is 4 digits long. the digits are 0-9 and no digit can be repeated in the number (in other words all 4 numbers are different)
examples: 1234, 1948, 4950
you have to guess the number
when you guess a number: your opponent gives you back 2 numbers (x,y)
number x is the amount of numbers in the right place
number y is the amount of numbers right but in the wrong place
anyone have an algorithm to solve this problem using the 2 numbers given back?

I've been playing around with the Rubik's cube subgroup generated by the turns: (U U' D D' L2 R2 F2 B2) which I refer to as the D4h cube subgroup after the symmetry invariance of the generator set. This, I believe, is the subgroup employed by Kociemba in his two phase algorithm. Anyway, I have performed a partial enumeration of the subgroup and its coset space. I was wondering if anyone might be able to confirm these numbers as a check on my methodology.

Six Face/D4h Coset Enumeration (q turns)
Depth  Cosets     Total
 0         1         1
 1         4         5