Thank you very much for the update Jarlve! I especially like the faster batch solver.
Largo, I just realized I forgot about your shutdown feature.
Never mind, thank you for considering implementing this feature!
I have added the feature and it works on my machine. You can find it under options, solver or in the settings.ini file as "(Batch ciphers & ngrams) Shutdown computer after task completion". Set it to 1 to activate it.
AZdecrypt 1.091 executable: https://drive.google.com/open?id=1EtJ_W … Xoh9XJ8CYI
Thank you very much Jarlve! The feature works great. Now I can run a long test without having to worry that the computer runs all night long!
Thank you very much Jarlve! The feature works great. Now I can run a long test without having to worry that the computer runs all night long!
No problem, it is your addition.
May I ask, at what settings do you run these tests? Iterations, restarts, ngrams, threads? And what CPU are you using along with the MIPS (million instructions per second) that AZdecrypt gets. I am thinking of a new PC, for a long time now, but not sure if it is worth it yet (2010-2017 stagnation). Also, there are rumours that the consumer Ryzen 2 will have 12 cores @ 5.1ghz: https://www.tweaktown.com/news/60100/am … index.html
May I ask, at what settings do you run these tests? Iterations, restarts, ngrams, threads? And what CPU are you using along with the MIPS (million instructions per second) that AZdecrypt gets. I am thinking of a new PC, for a long time now, but not sure if it is worth it yet (2010-2017 stagnation). Also, there are rumours that the consumer Ryzen 2 will have 12 cores @ 5.1ghz: https://www.tweaktown.com/news/60100/am … index.html
I’ve been thinking about buying a new PC for some time too. Just like you, I’m still waiting.
At the moment I have a relatively weak system with an Intel core i5-3570 4×3.4GHz. The tests I am currently running are using the default settings of AZDecrypt and 4 threads. You have chosen the default settings very well, because even if a run does not immediately lead to the result, you can still see a tendency by means of a slightly higher scoring. So I run the batch solver for large batch files and look at all results that exceeds 20800 points separately. This may not be the safest method, but I will stick to it for now. As soon as the test is complete, I’ll post again in a separate thread. Occasionally I interrupt the test series because I try another idea. So it may take a while yet.
I would like to have much more time for working on z340 and some tools. It would be so cool to have your solver open source and platform independent and/or GPU supported. Do you remember my proof-of-concept solver for monoalphabetic substitutions? This is of course no comparison to your powerful AZDecrypt, but it runs on macOS, Linux (+ Raspberry PI) and Windows. The advantage is that it builds up a network and you can share all the resources of your local IT. I have 4 Raspberry PI standing around, which are slow but together have a total of 13 cores (1x Raspi 1, 2x Raspi 2, 1x Raspi 3). If you keep dreaming about it, you’ll end up with a kind of "Zodiac@Home" (spooky name, I know…), which works exactly like Seti@Home. =)
It would be so nice to have something like that for my home network since I have four laptops and two desktop computers (almost three if I cobble together parts). Most of the time they aren’t working on anything! Lazy machines!
Hi, Jarlve,
I just noticed a small problem: The button "Stop Tasks" also leads to a shutdown when the corresponding option is switched on and the batch solver is running. But there’s no hurry, don’t worry. I just wanted to let you know
Hi, Jarlve,
I just noticed a small problem: The button "Stop Tasks" also leads to a shutdown when the corresponding option is switched on and the batch solver is running. But there’s no hurry, don’t worry. I just wanted to let you know
Nice catch Largo, thank you for your continuing effort to report back these issues to me! It is much appreciated.
Hey Jarlve,
this is a "Fake P19" cipher I’ve generated with my statistics tool (plaintext from project gutenberg "Dracula"). P19 contains significantly more Bigrams than P1. AZDecrypt solves it very fast in the normal solving mode. But when I untranspose it by P19 and feed it to the transposition solver, no solution is found. I’m not sure…does the transposition solver supports "Transpose Period n" or just "Untranspose Period n"? Or is the "fake period" the reason it’s not solved?
mn9QSoWXpjxqTrK0A UsBLkCuutVmMhnYyo bljSpZWqkJrDNsKvG AXzcLl1xt7OBimY2T 4hFJndZPMo3ye0Czb jxpcUkl5VDNwJ!W6u uqSAKvXdTr!O4fs12 tULP8YBVZeCjWymDS nQHAokBM3zpXbJqIc Nw5dvr8C0xeustgGm 7ODinAfoHBp7qCbJg 61DuA2frcIG3T4sBu CKltmjDkAlLP0jUnd JBeJYOVS!1yb2CcJZ 3DHIdGA7o0PupqH1z BWJCXrT!eMkxOUVPS uOT2yb3Jc01sUVtQI Dml2znPGj8OJdNAu! P5JoeKbgO630xpwqL 1HrJcM!P4J5SksTtl
this is a "Fake P19" cipher I’ve generated with my statistics tool (plaintext from project gutenberg "Dracula"). P19 contains significantly more Bigrams than P1. AZDecrypt solves it very fast in the normal solving mode. But when I untranspose it by P19 and feed it to the transposition solver, no solution is found. I’m not sure…does the transposition solver supports "Transpose Period n" or just "Untranspose Period n"? Or is the "fake period" the reason it’s not solved?
The AZdecrypt transposition solver supports both transposition and untransposition for all its operations.
The fake period is getting in the way. That is by design as it uses a bigram beam search to find good operations. This is why it works so well most of the time, but not all of the time. It is possible to replace the bigram beam search with the substitution solver itself but it would slow down the search to a crawl – something for later perhaps. Workaround, go to options, solver, and change the bigram beam states variable from 729 to 1. This will remove the bigram beam search from the equation and then fake bigrams will no longer affect the solver. Take in mind though that the solver will need a much longer time to find the operation. Perhaps change the operation stack size to 1 also.
Thank you for the explanation! I’ll experiment with it, maybe I’ll get new ideas.
But I have another question, if you don’t mind:
Some time ago I worked on a proof-of-concept tool that can solve monoalphabetic substitutions distributed over the network. My solver uses a hillclimber in combination with simulated annealing, based on n-grams. This works quite well, but I have not managed to extend the solver in such a way that homophonic substitutions can be solved too. I think I remember that we talked about it once, but I can’t find any details in the forum. How did you handle this in AZDecrypt? Where do you have the necessary knowledge? Have you thought up the procedures yourself or can you tell me about websites or papers that can help me? Unfortunately, I cannot estimate how much work it takes to write a very basic solver for "simple" homophone substitutions. My monoalphabetic solver should be a solid basis, shouldn’t it?
Some time ago I worked on a proof-of-concept tool that can solve monoalphabetic substitutions distributed over the network. My solver uses a hillclimber in combination with simulated annealing, based on n-grams. This works quite well, but I have not managed to extend the solver in such a way that homophonic substitutions can be solved too.
Just extend it and it should work.
How did you handle this in AZDecrypt?
Randomly initialize a symbol to letter key. Then have a loop that determines the iterations. Inside this loop go through the following procedure. Make one random change to the symbol to letter key, for instance, symbol 63 becomes letter L. Score the change with ngrams + ioc normalization, if improvement then keep the change, if not then discard it + your simulated annealing stuff and repeat.
Where do you have the necessary knowledge? Have you thought up the procedures yourself or can you tell me about websites or papers that can help me?
I did not have any knowledge. I cannot think of a single paper that will actually help you. Just extend your current solver.
Thanks for the informations, I’ll try it!
However, I have another question, if you don’t mind:
The solutions issued by the transposition solver contain information about the transposition used. Like so:
Offset column order (51*7, X: 15) Flip (305*2)
How can I reproduce this via the transposition menu in AZDecrypt? The details do not seem to match the fields A1-A8 and their corresponding instructions. How do I have to do that? I have tried quite a few things, but unfortunately I have not been successful. Thanks in advance.
51*7, these are the dimensions, x=51, y=7. You will need to set it manually through the dimensions menu and the same for 305*2.
Hey all.
Sorry to say that my cryptology work and forum activity/visits will be minimal for 2018, reason is that I want to move into other things. In the meantime I still want to offer support for AZdecrypt and if there is anything else you think I need to see contact me at: jaahrolov.ec@oym (untranspose it with period 2).
Hey all. Sorry to say that my cryptology work and forum activity/visits will be minimal for 2018, reason is that I want to move into other things.
I started feeling the same way several weeks ago. I too need to move on to other endeavors. I won’t be gone completely, but am at a point where I feel that I have few new ideas to share and am just ready for a new chapter. Thanks for everything that you have done. Your contribution was substantial, amazing, and will not be repeated. We had a connection, and I really had a lot of fun working with you, especially when we had the same thoughts at the same time even from geographically so far away. Good luck with whatever it is that you do.
Geoff
