« first day (1611 days earlier)      last day (1316 days later) » 

2:03 AM
@Pickett Thanks very much:) it works well
 
 
4 hours later…
5:48 AM
@Sascha I only see an image so if you are asking me if live profiler of a circut is useful I will say yes. But it's not a pshysics site, don't get me wrong, I like when enthusiastic people share their interests but mma.se has some rules we should try to obey :)

If your code uses some nice tricks which can be of use for wider audience, great, prepare context (shortly about circut demo), a minimal example and a problem and while self-answering about that general problem you can add your whole code where the solution was applied.
 
 
1 hour later…
7:06 AM
@Kuba The thing about the manipulate in the picture is that is uses the second argument of Dynamic to switch between pictures of the circuit where individual components are highlighted. While dragging a slider the corresponding component is thus highlighted in blue. I have not seen this yet somewhere else even though this is actually pretty straight forward
 
7:27 AM
@Sascha as I said, if you find it useful, can present the topic in readable manner then don't ask me but do it :)
 
8:13 AM
0
Q: Testing framework limitations in the FronEnd area

KubaI understand that user interaction with gui can't be tested automatically but if I can write the code using low-level notebook programming functions that checks something, should I expect it to work in testing notebooks? For example, the following code checks if a bug described in InputAliases a...

 
@Rolf You wrote on community in the WB question:
> Also, the disadvantage of the IntelliJ plugin for companies is that one has to buy an IntelliJ license first (which I did of course).
Maybe I'm completely off, but why exactly do you need to buy it? I thought the difference between the community edition and the Ultimate edition is the set of features and not that you cannot write code you sell with it.
(@R.M. is it possible that you super-ping RolfMertig on this?)
 
@halirutan I can super-ping him when I meet him :)
 
@Kuba hit him hard :)
 
8:29 AM
@halirutan btw. is it possible to convert syntax highlighting elements in IntelliJ to those used by SublimeText? (I'm not using terms like lexer etc because I don't know them :P)
 
@Kuba No, I don't think so.
The best thing you could get is the kind of highlighting that is done like in the pygments extension of R.M.
 
@halirutan Thanks, I don't really care so much, was just curious.
 
8:55 AM
@halirutan Probably this is not true (anymore?), at least according to this
@halirutan fixed in the community post. BTW: One thing I like better in WB than in your plugin is error reporting. But maybe this is just habit.
 
They have special licenses for the Ultimate edition where you can use it for free under certain conditions (but of course not commercial uses)
 
@Szabolcs The Ultimate edition has also quite some extra goodies, so I did not mind paying.
 
@RolfMertig What is it that you use from it (and is useful for Mathematica stuff)?
I really don't use it to its full potential ...
Or is it mostly things that are not useful specifically with the Mathematica plugin?
 
9:32 AM
@SjoerdC.deVries that was some awesome stuff @SjoerdC.deVries , thank you ! If you have any more ideas folks would be happy to se them. looking forward to your posts.
 
10:23 AM
@VitaliyKaurov @SjoerdC.deVries This problem can be transformed into a clique finding problem in graphs. Construct a graph where a node represents a given placement of a piece. Two nodes are connected if the corresponding placements do not overlap. Maximal (and largest) cliques are then the optimal arrangements of pieces in a given space.
I'll write it up if I have time.
The point of doing this is to offload the hard part of the algorithm to the clique finder function.
 
10:54 AM
@Szabolcs, maybe you have some ideas on making the neighbor-finding in this answer more efficient? The only sticking point of NeighborhoodGraph[] for me when I tried it out is that it is no longer guaranteed that the vertices are still ordered anticlockwise.
I can't make the entire Graph[] equivalent of the mesh a directed graph, since forcing one face to be directed anticlockwise will mean the adjacent face is clockwise, which I don't want.
I had the idea of using DepthFirstScan[], but I don't know how to implement the equivalent of TopologicalSort[] with it.
 
I'll take a look a bit later :-)
 
 
5 hours later…
4:29 PM
@Szabolcs looking forward to it, @Szabolcs !
 
 
2 hours later…
6:30 PM
@SjoerdC.deVries are you around?
 
6:42 PM
If you have M10.4, try this: FileNames@FileNameJoin[{$InstallationDirectory, "SystemFiles", "Fonts", "TrueType", "*"}] then try them like this.
 
@Szabolcs only on my iPhone...
 
7:09 PM
@SjoerdC.deVries Two questions: (1) Have you found all possible soma cube arrangements for the simple cube case with your program? How long did it take?
@Sjoerd (2) With FindCliques it takes quite a bit longer to find all of them than to find a couple of hundred. It took about 1000 seconds and found 11520 arrangements. Many of those are of course just rotations of each other.
Dividing by 24 (no of rotations) gives 480. But google says there should only be 240 ways. Do you see which symmetry I am missing here?
Must be the reflection ...
 
7:25 PM
@Szabolcs No, the first one was somewhat more than a minute that I recall. A watched a whole set of additional solutions coming in after that but I didn't let it finish. The final solution set must be in the 10,000s (of course, there are many rotationally similar solutions)
 
Yes, I was missing the reflection. I didn't look carefully enough at the pieces.
 
@Szabolcs I don't see how reflections could work if you have physical pieces.
 
@SjoerdC.deVries Most pieces are their own reflection, except for 5 and 6, which are each other's reflection. Thus the reflection of any arrangement can be also built.
 
@Szabolcs But wouldn't the reflections already occur in the form of a set of combined rotations?
I mean, they don't create anything that rotations wouldn't do, right?
 
@SjoerdC.deVries No, reflections and rotations are independent. It is not possible to rotate a left hand into a right hand.
I meant reflections and rotations of a completed arrangement (full cube)
Maybe I am misunderstanding what you are saying.
What is not obvious at all to me is that no arrangement has any rotational or reflection symmetries ...
But this must be the case, otherwise the count just doesn't add up ...
 
7:38 PM
I'm talking about reflections of the individual pieces. At least some of their reflections I can see as being composed of rotations
 
@SjoerdC.deVries Yes, that is what I meant when I said that most of them "are their own reflections". If we reflect them, the result can be rotated back into the original. This is true for all except 5 and 6. But those are each other's reflection.
So if we reflect all 7 pieces, we get the set of the same 7 (just with different rotations)
 
Sorry have to restart. Computer acting up.
 
That means that if we take a completed arrangement, its reflection can be built from the same 7 shapes.
What was a bit surprising to me is that based on this result (this count of arragements plus wikiepdia's claim that only 240 are distinct), none of the arrangements have any rotational or reflection symmetries. Each 24 rotations and each 2 reflections of these are all distinct.
This is not true for any individual piece. Piece 2 is the only one with 24 distinct rotations, but its reflections are not distinct from those.
 
Back again.
@Szabolcs Looking at the results, that doesn't surprise me much. But I have no real background in this kind of stuff.
Did you post your code already?
 
Yes. But only for the cube, not for arbitrary fill spaces.
It builds on yours.
 
7:50 PM
@Szabolcs Something strange happening for me. MMA hangs on my third block of code (the line showing the 7 colored pieces). Looks like the file may be corrupt or so.
Hmmm. Seems MMA broke. My 10.4 version chokes, but 10.3 is doing fine. Strange thing: I had my laptop in sleep in a drawer and when I took it out it was running. It might have been a windows update which took it out of its sleep, but it has wrecked MMA.
 
8:12 PM
Installing 10.4.1 now.
 
8:27 PM
Not sure what went wrong. The new 10.4.1 also choked on any line that would generate/replace 3D output. After removing all output everything seems to work fine.
 
8:53 PM
@Szabolcs I finally had time to check your post. Amazing out-of-the-box idea!
 
@SjoerdC.deVries :-) with IGraph/M I can get all solutions in merely 60 seconds. Just added a note to the post. But IGraph/M doesn't support finding only a limited number of cliques, e.g. to stop after finding a single one. I was planning to add this feature for a long time (and it is partially done), but I didn't have time to do it yet. It should be part of the next release ...
60 s is much better than 1000 s
 
9:12 PM
@Sjoerd Actually, IGCliques[g, {7}] from IGraph/M gives me the result, all 11520 of them, in only 0.8 seconds.
 
@Szabolcs That's fast. I started the full search on the cube half an hour ago.
 
9:42 PM
@Szabolcs It took 65 minutes for all 11520 (indeed) variations.
 

« first day (1611 days earlier)      last day (1316 days later) »