Tips on using Magic 3D Hyperbolic Tile {6,3,3} (MHT633 8c solved!)

[Raymond Zhao]

Hi y'all,

I finally solved the 8-coloured puzzle in MHT633! For various reasons, it took me several days.

Here are some issues I found along the way:

1. It's not immediately obvious which stickers correspond to the same piece, and it's not immediately obvious how far a piece is from another. Some pieces that appear to be far are often only a single turn off from where you want them. As you solve the puzzle, you gain more intuition on the distances.
   
2. If you only know to left-click-drag to rotate the view, which is the norm in other hypercubing programs, you really won't get far. Knowing all the navigation, twist, and search controls helps so much. Solving this puzzle feels a bit like playing a space exploration game sometimes; I'd do barrel rolls while zooming in to line up the view to how I wanted it before executing certain algs.
   
3. There's a parity situation on the 8-coloured puzzle that doesn't show up on the 3^4. Knowing how to solve the 3^4 helps, and knowing RKT definitely helps, but there's still a case that doesn't show up on the 3^4. This puzzle therefore isn't just "a larger 3^4". In fact, with the 8-coloured puzzle, I found it was as cramped as the 7-coloured hexagonal tiling in MagicTile.
4. I'm not sure if this was just because I was using a VM, but there were times when the program wouldn't launch because the settings file was corrupted. I never cared enough about the setting file to back it up; I'd just delete the corrupted settings file, and then readjust the settings after opening the program.
5. This might also be because I was using a VM, but sometimes, my log file would get corrupted during the solve. The easiest mitigation for this issue is to back up the log file regularly, which you should be doing anyway with puzzles that you expect to take a while to solve. For those of you who shut down your device, after saving the log file, make sure to close the program first, and only then gracefully shut down the device.
   

Also, I remember chatting on MSN Messenger with Matt (Matthew Sheerin) back when he solved this puzzle in 2012. Matt, I don't know if you're reading this, but back then, I understood absolutely nothing when you were talking about your 8c solve and dealing with parity. Now, I finally get it.

Kind regards,

Raymond

[Raymond Zhao]

Hi again,

I have solved the 12-coloured (12C) puzzle. I used F2L-a (2c/3c pairs) and keyhole F2L-b (3c/4c pairs) and was able to get a 3.3k move solution.

I messed up during several steps of the process, so it should be possible to get an even more efficient solve using my approach.

The log/setting file corruption issue happened much less often this time. Because I'm using a VM, shutting down the instance actually does a force-shutdown, and I think that is what causes the file to be corrupted if I don't close the program beforehand. At this point, I also have separate setting files for each puzzle backed up, because I realized I prefer different colours on different puzzles, and also because I didn't want to have to reconfigure colours if the setting file got corrupted again.

As for parity, I thought of a way last night to calculate whether a puzzle could have parity.

Let N be the number of neighbours of a cell for an MHT633 puzzle with T colours. If 6N/2 is divisible by 2, then there is no 3c parity. If 6N/3 is divisible by 3, then there is no 4c parity. In other words, if N is even, there is no 3c parity, and if 2N is divisible by 3, there is no 4c parity.

Some examples:

The 8C puzzle has N=7, so it can get both 3c and 4c parity.

The 12C puzzle has N=9, so it can get only 3c parity.

The 14C puzzle has N=13, so it can get both 3c and 4c parity.

The 20C pattern A puzzle has N=16, so it can get only 4c parity.

I also realized that if N = T - 1, then all of the cells are adjacent to each other, and it could make sense to solve a belt or border around a cell first, so that that cell becomes the "last layer" to solve.

Kind regards,

Raymond