Message #2831

From: Roice Nelson <>
Subject: Re: Re: [MC4D] RE: New puzzles
Date: Tue, 26 Nov 2013 19:02:19 -0600

Hi Andrey,

Sorry I haven’t finished merging the puzzles into the distribution yet (I
have started at least). I’m writing now because I did notice a few of the
puzzles could use a bit more fill-in by adding more identifications.
Upping the tile count also helps, but that has the disadvantage that build
times scale with the tile count (would be so nice to have build caching, as
suggested by Melinda). Probably doing a little of both can give the best

In the worst case for puzzles with poor fill-in, you can pan them off to
the boundary (MT will fail to keep the puzzle centered). Good coverage is
visually better anyway, and for these reasons, I’d like to get the puzzles
included with main program setup as well possible.

I noticed the following could use some improvement:
{12,3} 24C, 6C
{12,4} 8C

For example, here’s a screenshot of what the 6C can look like when you pan
near the orange tile right now:

The {5,5} 12C is pretty good, but could maybe use just a tad extra love
too. I’ll work on these, but I thought I’d mention in case you wanted to
as well. Good to keep in mind for future puzzles in any case.


P.S. Had a long plane ride today and made a change to display the topology
information, which I’ll push out with your new puzzles :)

On Tue, Nov 19, 2013 at 1:12 AM, Roice Nelson <> wrote:

> ok, we’ll do them as separate puzzles then. I won’t repeat the
> face-turning slicing in both places though, since the face-turning versions
> will behave identically. Would be nice if we can come up with some
> distinguishing naming too.
> I’m not sure I followed your last question. Did you mean to ask if you
> can do an identification with an "in place reflection" and "end rotation",
> but without extra *reflections*? If that is what you meant, this is
> possible to configure, though I’m not sure it can lead to sensical
> topologies. I just tried it on a couple puzzles including the {8,4} 9C and
> only achieved strange results, but maybe there is some case where it would
> work. To have no reflections, use the same trick I suggested above and
> make the EdgeSet 0. Internally, this reflects the tile twice, but the
> second reflection just undoes the first one. You can do this and set the
> other properties however you want.
> {8,4} 9C has 9 faces, 16 edges, and 8 vertices, so the Euler
> Characteristic is 1 and the topology is the projective plane. As
> configured, there are some vertices and edges that look identical (same
> colors) but are physically different. Not all the faces are octagons.
> Looks like 4 are squares and 4 are digons. It is a strange puzzle for
> sure.
> Roice
> On Mon, Nov 18, 2013 at 1:48 PM, <> wrote:
>> Roice,
>> I think, it’s better to add new puzzles with additional identifications
>> on vertex- and edge-centered twists. Because they are really different
>> puzzles, and some of them may be much more difficult than old ones.
>> I’m looking at {8,4} 9 colors, and can’t understand it. In
>> face-centered puzzle it works like non-oriented non-uniform puzzle with
>> some two-side edges. In vertex-centered variant some vertices of the same
>> structure are identified, but sometimes there is only half of them… and
>> the identified ones don’t all have the same orientation.
>> Is it possible to add identification with "in place reflection" and
>> "end rotation", but without extra rotations?
>> Andrey
>> —In, <roice3@…> wrote:
>> Hi Andrey,
>> Thanks for your observation about this. To get the "mathematically pure"
>> behavior you are wanting, we can add an additional identification to each
>> of these puzzles, one that is a rotation only (no reflections). We can
>> effectively get that by marking EdgeSet 0, and using the appropriate
>> EndRotation… 4 for {8,3} and 5 for {10,3}.
>> Because of the solutions listed in the table, there is the question of
>> whether to edit the existing puzzles or add new ones. The existing
>> definitions are valid configurations too, just with a different topology.
>> But they are similar enough that I’m thinking we wouldn’t want separate
>> definitions with only this difference.
>> If it is ok with you and others, I will just change the behavior of the
>> existing puzzles and not worry about the table, but if anyone disagrees
>> please let me know. I’ll push the change out at the same time as the
>> addition of all the new colorings you’ve been making.
>> Thanks again,
>> Roice
>> On Sun, Nov 17, 2013 at 3:22 PM, <andreyastrelin@…> wrote:
>> Roice,
>> something is wrong with {10,3} 6C edge-rotated puzzles. When I select
>> some edge, I expect that edges on opposite sides of its decagons will be
>> selected too (because mathematically they are the same). But that edges
>> remain non-selected. Same is true for vertex-rotated 6C, and also for
>> {10,3} 12color.
>> Is there something missing in puzzle description?
>> I see the same in {8,3} 6C… and I don’t like it because there are
>> solutions of these puzzles in the table (including some of my own ones)…
>> Looks like we solved puzzles that are not as "mathematically pure" as they
>> should be.
>> Andrey