Maze Design (2005)
cgl.uwaterloo.ca
cgl.uwaterloo.ca
> The two problems are clearly dual
> A: since a maze's solution is a non-intersecting path, and
> B: conversely every non-intersecting path is the solution to a corresponding maze
Focus on "B" - the output of a TSP algorithm "is the solution to a corresponding maze".
IOW, the negative space of the TSP algorithms output is the maze itself...
The first is a very short bite-size summary of all of the algorithms that the book has covered, so that you don't have to read through the entire book again to remember the differences between the different algorithms. This is so good I wonder why it is not a standard feature of textbooks in general, such that it has its own standard section name (say "Abridgement"). The key is that it's longer-form than an introduction or abstract or conclusion, but the freedom to occupy its own space allows it to be expressed in human terms, unlike the "real research" parts which have to be a bit more rigorous.
The second is a quantitative comparison of the different algorithms among different measurements, which again is like "why isn't this an appendix everywhere?" and such. It's still not trying to be absolutely rigorous but just to give a rough feel for the different options there.
As a result you can just grab the textbook as a reference, flip through those appendices, say "yeah that's what I'm looking for, longer twistier paths," read up a quick description of how that algorithm works, say "well that's a bit more involved than this other one which looks good enough, maybe I'll start from there."
Well worth it.
Sounds cool. Is there a public implementation?
You can see the write up / output: https://medium.com/@G3Kappa/generating-mazes-from-pictures-o...
Back in the 90s I had a great maze generator/game that generated mazes with over/underpasses between points. I've also played around in some physical 3d mazes made of lumber. Not just walls, but several layers of paths above and below each other like a big rubix cube (great drunken fun). So what would a 4/5/6d maze look like?
(I hope everyone is familiar with both books.)
{., NE, NW, ES, SW, NS, EW, NE/SW, NW/SE, NES, NEW, NSW, ESW, NESW}
In a more visual language you'd have 5x5 tiles which the bigger picture is made out of: \ / \ / \ / \ / \ /
\ / \ / \ / | | ‾‾
X X X X | |
/ \ / \ / \ / | | __
/ \ / \ / \ / / \ / \
\ / \ / \ /
\ / \ / \ /
\ / X X X X .
\ / / \ / \
\ / / \ / \
Once you've figured out each individual vortex I think solving the bigger problem is probably doable?http://www.astrolog.org/labyrnth/maze/larger.gif
Enjoy!
Note: This is from the "Think Labyrinth!" site mentioned on the Maze Design site:
http://www.astrolog.org/labyrnth/maze.htm
It is also not the "largest maze" on that page by far; there are a couple of larger ones (as zip'd files), and likely those aren't the largest today.
I'm pretty certain they are just meant as "art" - and not to actually challenge someone to solve them; but if you are so inclined, go for it!
EDIT: I just wanted to add a bit more about the "dense maze" - the start/end is at the upper-left/lower-right of the maze; the white line is the "path" of the maze (black being the walls). Note how the image "changes" depending on the level of "zoom" - if you fit it to a smaller size, the "white" of the image will take a "cross shape", while other areas are "darker" - zoom in on those various areas, and notice how the path changes; it's a very complex construct!
Update final: After just over 4 hours it appears to have given up without actually filling anything.
Cue "I don't know what I expected" meme.