Yes, my last post was hastily done and in error. I looked closer, and there were EDIT: 6 iterations out of 9948 where there were a null and skip in the same place, not two skips in the same place, but never two iterations in a row.
EDIT 2:
For the third restart that was a solve, there were 21 iterations where there were two nulls in the same place ( bottom chart red lines ), 53 iterations with two skips in the same place ( green lines ), and 6 iterations ( blue lines ) with a null and skip in the same place.
The latest build just stopped working in the middle of a run. Need some time to sort things out properly.
Should be okay now: https://drive.google.com/open?id=1lGcJi … qY7-ldk0hd
smokie, can you please test again for nulls and skips or two nulls in the same position please.
Thanks
There is a problem. Iteration after iteration there is no change, always the same null and skip positions, except it does change once in a while. The scores change, just not the symbols, so I wonder if it is the output file.
EDIT: Here are the first few iterations. Scores change but not the symbols.
T: 004 A: 1 Score: 19542 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 003 A: 1 Score: 19618 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 005 A: 1 Score: 19707 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 002 A: 0 Score: 19625 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 006 A: 0 Score: 19410 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 001 A: 1 Score: 20076 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 004 A: 0 Score: 20044 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 002 A: 0 Score: 19886 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 003 A: 1 Score: 20110 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 005 A: 0 Score: 20038 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 001 A: 0 Score: 19390 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 004 A: 0 Score: 18973 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 003 A: 0 Score: 19225 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 002 A: 0 Score: 19854 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 006 A: 0 Score: 20026 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 001 A: 0 Score: 19835 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 003 A: 0 Score: 19478 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 005 A: 1 Score: 20112 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 002 A: 0 Score: 19739 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 004 A: 0 Score: 20051 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 006 A: 0 Score: 19593 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 001 A: 0 Score: 20005 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 003 A: 0 Score: 19940 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 004 A: 0 Score: 19729 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 002 A: 0 Score: 19229 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 001 A: 0 Score: 19802 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 005 A: 0 Score: 19688 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 004 A: 0 Score: 19541 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 006 A: 0 Score: 19739 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 002 A: 0 Score: 19473 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 001 A: 1 Score: 20374 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 005 A: 0 Score: 19736 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 003 A: 0 Score: 19227 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
T: 004 A: 0 Score: 19955 Nulls(0005,0060,0063,0148,0330) Skips(0128,0284,0321)
Sorry. It was something silly I forgot to uncomment in the code. Keep working with the old version for now.
Sorry. It was something silly I forgot to uncomment in the code. Keep working with the old version for now.
Everything seems to be working okay, even the slowdown issue is totally fixed and all outputs are correct, but will hold back a new build until somewhere in the weekend just be sure.
O.k., that is fine. Still working on the general iterations versus restarts data.
Here is the 3 million null skip iteration data, left column iterations per restart, middle number of restarts, and right column number of solves. Last category, 100,000 its with 10 restarts is going now and not done yet *. It will probably net 3 to 7 solves and I will put the results in the spreadsheet in about 8 hours or so. Even though there is a peak at 20,000, It’s not exactly a perfectly smooth curve, considering that there is a dip at 25,000 and 30,000, so I am thinking about dedicating the Acer to another week to see if it levels out or smooths out the curve. I think I will do that. What do you think?
EDIT: done
5000 600 16
10000 300 17
15000 200 20
20000 150 22
25000 120 12
30000 100 12
50000 60 20
100000 20 12
I am thinking about dedicating the Acer to another week to see if it levels out or smooths out the curve. I think I will do that. What do you think?
Many restarts/samples are needed to get a good average. Get at least 500 restarts per.
New build: https://drive.google.com/open?id=1dSU0d … 5ieYshB5zx
Please test for nulls and skips, and two nulls in the same position again.
The new build on jarlve_p20_5nulls_3skips using the standard 5-grams:
Restart: 1000/1000 HC: 8252/10000 Sub: 500000 Development information: over 22500: 77 (7.7%)
And with new Reddit 5-grams compiled from a 705 GB corpus:
Restart: 1000/1000 HC: 7858/10000 Sub: 500000 Development information: over 22500: 110 (11%)
New Reddit 5-gram for AZdecrypt download: https://drive.google.com/open?id=1ixWBc … q0ytZLwHnw
Looks just fine. Here is before the new build, the little colored lines on the bottom graph show N/S, N/N and S/S same positions. Blue = N/S, Red = N/N and Green = S/S.
Here is after, with only S/S same positions little green lines only, no blue or red. 25 out of 10,000 iterations had a S/S.
Looks fantastic. Thanks a lot.
Great, happy that it is working fine now. Your graphs are amazing.
I am done with my tests for now, anything that you want to do? Or want me to help with?
I am still working on my tests and learning how the hillclimber works, which may take some time. Last weekend I made some messages for the future, and they are in the message tested tab on the optimization spreadsheet. There is a P20 8/0, 4/4 and 0/8 to see if any are easier than the others, and I also made a P10 4/4 ( 34 x 10 inscription rectangle ) and P34 4/4 ( 10 x 34 inscription rectangle ) with nulls and skips in exact same places to see if a smaller period is harder to solve than a bigger period. Be my guest when and if you are interested. Otherwise I am fine for now. Thanks a lot.
There is a P20 8/0, 4/4 and 0/8 to see if any are easier than the others
I would like to do this but they have different lengths. That changes the multiplicity and would make the longer ciphers easier to solve. That could affect the test.
and I also made a P10 4/4 ( 34 x 10 inscription rectangle ) and P34 4/4 ( 10 x 34 inscription rectangle ) with nulls and skips in exact same places to see if a smaller period is harder to solve than a bigger period.
These have different substitutions and that could also affect the test.
I understand about the p20 8/0, 4/4/ and 0/8. The P10 4/4 and P34 4/4 are the same message and have the same substitution key, but there is 25% randomization. I am going to experiment with the program for a while, so I won’t be offended if you want to attack the 340 on your own. Thanks for the compliment on the graphs and thanks for the program.
Okay sure. Let me know if you need some help.
I will run the test to determine how the size of the period affects the solve rate with a few ciphers of my own.
