I have just started learning the Tamariz stack, and alongside that am doing a lot of research into the mathematics of the Faro Shuffle.

I downloaded StackView, and am currently enjoying using it as a tool for looking at the permutations of cards... i.e. from the more mathematical perspective (as a maths student this is interesting stuff...)!

I was wondering what people's thoughts are on using this for magic purposes, and what sort of principles you've noticed through using the programme.

I have found it a great tool for demonstrating why it is that the pack returns to the starting order after 8 faro's, and other things along those lines... But as mentioned, these are more mathematical things.

I'd love to hear what people have done with it.

Owen.

I use it mainly for practise. Since I'm on the computer a lot it's a good thing to have. I use it two or three times a week at the moment.

Joakim

I think one of the great joys of StackView is being able to design your own stacks from the ground up. If it's a full deck memorised stack, it helps you to learn it too.

Just out of interest, has anyone used the search facility in StackView to successfully find sequences of moves to get form a starting stack to a target stack?

I haven't. The most immediate application I can think of would be if there would be some easy way of getting to your stack from new deck order, but that ain't gonna happen! (That said, I think Darwin Ortiz has some sequence of moves to go from new deck to some "simpler" type of stack like Si Stebbins).

The StackView Search capability has some value in your search process if you can limit or constrain the number of different manipulations that you are including. The fewer the better. For example, if you want to test for what combination of In and Out Faros are necessary to get certain cards to target positions, StackView Search is very fast. However, if you allow for every possible manipulation to be included to find the New Deck Order to Aronson Stack sequence, it will take a very, very long time. Rather than link you to an earlier post, I have copied some of my earlier thoughts on that here. Someone had asked if anyone had thought about NDO to Aronson, and here was my response:

"Unfortunately (for me), I have thought about this for a very long time.

When I added the "search" capability to StackView, my specific intent was to have the computer find the "NDO to Aronson" solution. But much to my dismay, even with the most powerful computers today, it would still take millenia to comb through all the possible combinatorics. I have had some friends who have Phd's in mathematics help me with the search algorithms, but we are always left with a brute force approach for one simple reason: there is no measure of "goodness" of any interim deck sequence. Since a complex handling such as a partial faro does so much damage to a deck sequence, you can never tell "when you are getting close" in order to focus the search on any particular logistical branch. even if I was able to develop a test of goodness, it would add a huge amount of overhead in each testing cycle. One of the math wizards indicated that I would in effect be doing recursive searches -- and in the end it would still take forever. Finally, I also explored doing a similar thing that MIT did for space exploration with the SETI project -- however, even if all the computers in the world shared their resources to search for the "Aronson Solution" (if I may be so bold to formally name the challenge), it would still take forever to find.

So, we still need some good old fashioned human intervention and brilliance to arrive at the Aronson Solution."

Nick

NDO to Aronson does sounds complex. Now you've got ME thinking about algorithms for computing it. This going to be stuck in my head for the next day or two--I can tell! (Sounds like I'll be getting off easy compared to you if it is only for a couple of days!)

Since you are here Nick, I was wondering if you've ever thought about doing a web version of StackView. I've thought about putting something together that is web-based, with a much reduced feature set, but that's one of many projects I'd like to get to!

Ashley,
I had thought about haveing a web-based version of StackView at one time. The truth is that I don't know how to do it, and more significantly, I don't have the time to learn any new programming tools at the moment. (As much as I'd love to.)

I know that a web-based version would be welcomed by Apple users. I also have had many requests for PDA/Phone apps. Again, the issue for me is the time to do it.

Some people have offered to take my source code and develop those applications for me. I have generally declined those offers for the same reason - time. Since I am not a sophisticated programer, my code is quite sloppy, and does not follow programming conventions. Also, I don't really know the available data structures in Visual Basic, so I kind of developed my own pseudo object oriented data structure. As such, the 100,000+ lines of code are quite unweildy to understand. There are many parts of the code that I do not remember how it works. And since there is so much interdependency in the functionality of the various modules of code, making a change in one place can adversely impact another part of the program's functionality. I wrote it and can't understand it, so just imaging how much time it would take to try to explain it to someone else.

Sorry for the apparent whining. Someday, when I have the time (hah!), I plan on doing a major re-write to address all of the interests and suggestions I have received.

Nick

Nick (or anyone else experienced with StackView),
Does StacView have the capability to save a series of shuffles,and replay them, without you having to reprogramme each move individually.
i.e. Could I save a series of shuffles (maybe an in-faro followed by an out-faro then a cut at 23rd card, the etc. etc.), and then have the programme replay this so that you can watch the effects of each individual shuffle, without having to recommand each move every time you wish to look at the shuffle?

Thanks.
Owen.

Owen--Yes, StackView calls it recording a "session."

Owen,

Yes, as Ashley mentioned, there is a tab on the main control window that allows you to record "Sessions." Think of this as a "Macro Recorder" for Excel, or "Actions" in Photoshop. Once your Session has been recorded, you can save it as a Session file for later retrieval or sharing with others. You can play back the steps all at once, or you can step through them one at a time. The easiest way to record a session is to turn on the Record Mode, and then just do the steps you want to create. Or if you are more of a programmer, you can write an ascii text file of the appropriate commands. The full set of functions and their syntax is documented in the appendix of the Users Guide. You can see some screen shots at my website.

Nick

Hi Nick,

I'm not a PhD by any means, but my initial thought on reading your post was to see if a neural network could be trained to spot when the state of the stack is getting getting nearer the solution.

Of course, there's a world of difference between idly suggesting something and implementing it.

Tomo,
I know a fellow at work who is an expert in AI, and we talked about applying neural nets to this for a while. The challenge (as I understand it) is in learning about "levels of goodness" -- where goodness is defined as being near the objective stack. This would be relatively straightforward for one or two levels away from the target. However, a brute force search will find the target faster than any other alternative if you are only one or two levels away. The fundamental issue is that some manipulations (such as a partial offset faro) does a tremendous amount of damage to the order, and it is not apparent when you are approaching an interesting intermediate state. Bottom line is that you have to be able to know the goodness of an intermediate state, and then test for that goodness in order to redirect the search. All of this creates a significant amount of computational overhead that slows the search to a grind.

So the issue is knowing how to determine the goodness. Since neural nets are good at this sort of thing, it would appear to be the right path. However, the neural net needs to learn what is good by observing its results and progress. In order to learn about goodness, then it would have had to actually solve the NDO-to-Aronson problem to know how to gauge the goodness. But if we knew the solution, then we wouldn't pursue the neural net path any more.

Another similar path is to apply dynamic programming techniques to soleve the problem backwards. In this case, we would start with the Aronson stack, and apply many inverse manipulations. We would then record the various intermediate states in a database, and then check the database aginast the intermedate states of a forward-looking search process. Here again, I beleive the intermediate step overhead grinds this process to a halt.

I am not an expert in any of these approaches, and I fear that my explanation may not even be correct. All I do know is that it is not an easy problem to solve computationally.

Nick