anastasios-tampakis.net

Wheel Generator did it again! It has managed this week the following impressive records.
19,6,4,5=111 (was 115 for more than a decade, WG made it at 113 back in 2009 which is also contained in Iliya Bluskov's Pick-6 book) and this week made it to 111 blocks!

20,6,4,5 = 139 (was 142 for many years).
Also WG holds the records 16,6,4,5=52 and 17,6,4,5=66, both also included in Iliya Bluskov's Pick-6 book.
Also WG made some other impressive records, like the 33,9,4,6=84 (previous record was 107) and 34,9,4,6=116 (previous was 121).

Congrats, LA

HI LA

I just recently had WG and still learning, ran for a while, had combinations.. just one silly question, how long should we accept WG to run to produce final combinations, you can have current and best combination but I mean is there limit to run it and how long ?

Hi Rustamshah, this is completely user dependable. If the results obtained are satisfactory, then you can stop the optimization. If you have set the bias sliders to their limits, as per the WG guide posted here, and you see no further improvements, then you can consider the optimization completed. However, WG will never ever stop scanning for two main reasons:
1) Due to the NP-hard problem addressed, we can't be sure if we have ever reached the best of the best possible, unless if it is obvious somehow it is the best of the best[*].
2) Checking for the best of the best is very time consuming practice thus not included as part of the scan to speedup the actual optimization.

[*] For a simple coverage optimization (without filters), the best of the best can be determined if we can achieve a pack design really (check what the main coverage line indicates for that condition - explained in the help file). However, achieving a pack design is really possible only in open-cover constructions up to 30-90% and very few close-cover constructions. Since most users build 90%+ coverage constructions and possibly include some filters, we can't produce a pack design (or have a way to identify a pack design in more complex constructions), therefore it is impossible to check for termination info since there can be an improvement somewhere and the only way to find it is to optimize for more time. For all that reasons, there isn't any scan termination in WG.

Having said that, small coverings (consider k, 1-block cover values) can be constructed within minutes close or exactly to their known records, quite larger coverings may take days to reach a very good state. For example, I have attempted to construct with WG the recent record of 29,10,6,8=576 (previous known record was 610). WG within a few hours beat the older record of 610 and after 4 days of optimization almost equaled the new record. This was attempted from scratch by WG, other build approaches could possibly reach and possibly surpass the new record much quicker. As you can see, there isn't a single answer here. WG does not have any termination checks for a very specific reason.
To contrast the time needed to build coverings, for example the above mentioned records of 19,6,4,5=111 and 20,6,4,5 = 139 where made in less than a day of optimization.

Hi LA

Thanks for info, i started with very simple approach, selected 36 numbers and trying to get a 4 right if 6 are among 36. Just don't know how to see how many combinations are best for playing, how can I see, what WG would consider the best number combination. I am not using any filters and also using only 1 L ...to be matched.

Please also tell if you may, what is the meaning of (36,30,28,30) as Inverse values ?

Thanks

For the inverse, given you know what the covering expression c(v,k,t,m)=b means, the inverse is a covering that can be extracted by the currently processed covering (the help file contains more info on the inverse extraction). If the inverse covering suits your needs, you can export it from the menu. This is mostly for KENO style coverings where the inverse usually generates k > 10 coverings, if you can find a setting for a construction that can be optimized by WG that produces an inverse that suits your needs. if you don't play KENO, don't bother with the inverse.
Typical values for b can be found in repositories such as
http://weefs-lottosysteme.de/systeme,en.htm or
http://www.ccrwest.org/cover.html

This has been discussed a lot in the past about the parameter.

HI

As a matter of fact I do play Keno quite regularly so it will be interesting to use of WG in this regard.

The inverse functionality was added because it is easy to extract the inverse covering. Naturally WG does not support KENO because the required setting of k > 10 goes beyond the capabilities of the engine. The help file contains an equation for the inverse, if you can determine a KENO covering produced as inverse and it is within the capabilities of WG covering limits, then of course you can use the inverse. I'm sure very few KENO coverings can be generated using the inverse for actual play. Unfortunately, it is very hard, if not impossible, to make a usable program that deals with KENO coverings because the computational demands to check the covering qualities grow exponential (and the main reason k,t,m are limited in WG); at least by the way optimization works in WG. There may be mathematical techniques that can construct easily KENO style coverings, and why not to make a program that utilize these techniques.

Hi lottoarchitect,
Fresh 2 new records by WG 1.5

#1/ C(17,8,5,6)=53
previous record was C(17,8,5,6)=54

01 02 03 04 05 12 16 17
01 02 03 06 09 10 15 16
01 02 03 06 10 11 15 16
01 02 03 07 08 12 13 16
01 02 03 07 09 11 13 17
01 02 04 05 06 14 16 17
01 02 04 06 07 08 14 15
01 02 04 09 10 12 13 14
01 02 04 10 11 12 13 14
01 02 05 06 07 09 11 13
01 02 05 08 10 12 15 17
01 03 04 05 06 07 10 12
01 03 04 05 08 09 13 15
01 03 04 05 08 11 13 15
01 03 04 05 10 14 16 17
01 03 06 12 13 14 15 17
01 03 07 08 09 10 11 14
01 03 08 09 11 14 16 17
01 04 06 07 08 09 11 17
01 04 06 08 09 11 12 16
01 04 07 10 13 15 16 17
01 05 06 08 10 13 14 16
01 05 06 09 10 11 13 17
01 05 07 09 12 14 15 16
01 05 07 11 12 14 15 16
02 03 04 06 08 10 13 17
02 03 04 07 09 11 12 15
02 03 05 06 08 09 12 14
02 03 05 06 08 11 12 14
02 03 05 07 10 13 14 15
02 03 06 08 09 11 14 17
02 04 05 06 12 13 15 16
02 04 05 07 08 09 10 16
02 04 05 07 08 10 11 16
02 04 07 09 11 14 15 17
02 05 09 11 13 15 16 17
02 06 07 10 12 14 16 17
02 06 08 13 14 15 16 17
02 06 09 10 11 12 16 17
02 07 08 09 11 12 13 15
03 04 06 07 09 13 14 16
03 04 06 07 11 13 14 16
03 04 06 09 11 12 15 17
03 04 08 10 12 14 15 16
03 05 06 07 08 15 16 17
03 05 07 09 10 11 12 17
03 05 07 09 10 11 14 17
03 05 09 10 11 12 13 16
04 05 06 09 10 11 14 15
04 05 07 08 12 13 14 17
04 08 09 11 12 13 16 17
06 07 09 10 11 12 13 15
08 09 10 11 13 14 15 17
..........

#2 invers from C(17,8,5,6) >> C(17,9,8,11)=53
previous record was C(17,9,8,11)=54

06 07 08 09 10 11 13 14 15
04 05 07 08 11 12 13 14 17
04 05 07 08 09 12 13 14 17
04 05 06 09 10 11 14 15 17
04 05 06 08 10 12 14 15 16
03 07 08 09 10 11 12 13 15
03 05 09 10 11 12 13 16 17
03 05 06 07 08 11 15 16 17
03 05 06 07 08 09 15 16 17
03 04 08 10 12 14 15 16 17
03 04 06 07 09 11 13 14 16
02 08 09 11 13 14 15 16 17
02 06 07 10 11 12 14 16 17
02 06 07 09 10 12 14 16 17
02 06 07 08 09 11 12 13 15
02 04 05 07 08 09 10 11 16
02 04 05 06 12 13 15 16 17
02 04 05 06 07 10 12 13 15
02 03 05 10 12 13 14 15 16
02 03 05 07 10 13 14 15 17
02 03 05 06 08 09 11 12 14
02 03 04 07 09 11 12 15 17
02 03 04 07 08 12 14 15 16
02 03 04 06 08 10 11 13 17
02 03 04 06 08 09 10 13 17
01 05 07 09 11 12 14 15 16
01 05 06 08 10 13 14 16 17
01 04 07 10 11 13 15 16 17
01 04 07 09 10 13 15 16 17
01 04 06 08 09 11 12 16 17
01 04 05 07 10 12 13 15 16
01 03 07 08 09 10 11 14 17
01 03 06 11 12 13 14 15 17
01 03 06 09 12 13 14 15 17
01 03 05 06 08 10 12 13 16
01 03 04 06 07 08 10 12 14
01 03 04 05 08 09 11 13 15
01 03 04 05 07 09 10 11 12
01 03 04 05 07 08 13 14 15
01 03 04 05 06 10 14 16 17
01 02 05 08 10 11 12 15 17
01 02 05 08 09 10 12 15 17
01 02 05 07 08 10 13 14 16
01 02 05 06 07 09 11 13 17
01 02 04 09 10 11 12 13 14
01 02 04 06 08 13 14 15 16
01 02 04 06 08 12 13 15 16
01 02 04 06 07 08 14 15 17
01 02 03 07 08 12 13 16 17
01 02 03 06 09 10 11 15 16
01 02 03 05 06 07 10 14 15
01 02 03 04 05 08 14 16 17
01 02 03 04 05 06 07 12 16
.........
Regards

Nice work edi, you should send these to WeEf's repository.