[r-t] 23-spliced TB major

Fri Apr 30 19:45:39 UTC 2010

This is similar to an approach I used in the mid '80s to generate random 
multi-parts of spliced, although I picked the calling first and then 
found methods to fit. I had two programs running back to back on a BBC 
micro:

- the first did a random sort of the possible changes for each row, then 
a systematic search to find the first method in order with a given lead 
end, for each of the leads of the peal that had not yet had a method chosen.

- the second proved the resultant composition, and if there was any 
falseness determined a minimal set of leads that could be removed 
leaving the remaining leads mutually true, chosen at random from amongst 
possible sets.

Eventually these would converge to a true composition within a few 
hours. I produced a couple of peals of 67 atw Surprise Major this way 
when Simon Linford first mooted the idea of ringing this, the first 
starting from the 45 atw extension of Norman Smith's, and the second 
from a blank sheet.

This was also used to generate the "Triples" Surprise Major peal. It is 
a bit of a sledgehammer approach, as in general there will be only a 
limited number of e.g. first sections that are true - it is best to find 
these first. As a quick hack when looking for a composition to build up 
to the "Triples" I checked only the first two sections to come up with a 
set that worked before finding 3rd and 4th sections to go with them. 
Unfortunately for one attempt I forgot to turn them back on. Patrick 
Brooke discovered this when checking the composition, but we decided to 
go for the false peal for practice in any case.

The latest implementation along these lines is SCAMP 
<http://myweb.tiscali.co.uk/saddleton/software/scamp.htm>. This is much 
more sophisticated and interactive. Looking at the code I see I have a 
Randomize function to vary the order in which methods are found, but it 
doesn't appear to be called.

Most of the work on the Cambridge Colleges peal was done when the only 
computing resource I had was the University mainframe: I generated the 
methods by hand, and had to wait for the operators to collect the output 
hard-copy after each attempt. There is more on this at 
<http://myweb.tiscali.co.uk/saddleton/ss8.htm>.

Philip

Philip Earis said  on 30/04/2010 15:07:
> 
> 
> I’ve been thinking a bit more about generating random methods for an 
> arbitrary peal of 23-spliced.
> 
>  
> 
> The simplest way to do this seemed to be with excel. In column A, I just 
> entered the different possible changes for row 1. Cells A2:A13 therefore 
> read:
> 
>  
> 
> X
> 
> 34
> 
> 36
> 
> 38
> 
> 56
> 
> 58
> 
> 78
> 
> 3456
> 
> 3478
> 
> 3458
> 
> 5678
> 
> 345678
> 
>  
> 
> To select one of these at random, I then just enter the formula below.
> 
>  
> 
> =INDIRECT("A"&RANDBETWEEN(2,13))
> 
>  
> 
> Columns B onwards are used for subsequent changes in the lead, and the 
> notation for a lead brought together using the concatenate function.
> 
>  
> 
> Of course, the danger here is that methods are trivially false, eg with 
> the same bit of place notation appearing in adjacent changes.  To screen 
> out these, I ran the methods through Richard Smith’s method filter, 
> which produced just a list of methods with true 7-lead plain courses.
> 
>  
> 
> The first 23 such methods generated were: * *
> 
> *   *
> 
> 567.7.5.45.25.256.5.236.2-4.3.2.6.23456.2367.36.23456.36.23.4.234.4.236-25.2.6.34567.567-27 
> 
> 
> 347.345.3.6.256.567.56.367.47.2.2347.345.6.56.34.2345.36.56.6.345-27.2347.1.2567.56.27.45.347.34567.3456.4 
> 
> 
> 34.36.567.1-27.5.1.4.347.4.2345.4.6-45.2-2.45.2347.34.4.23.7.567.7.67.345.347.7.456 
> 
> 
> 56.345.34567.4.2.7-236.7.27.4.2345.3456.34.234.367.56.236.234.25.347.27.34.2367-567.27.6.34567.56.34.67 
> 
> 
> 3456.7.345.47.56.25.27.6.2-7.23.6.256.34.567.34.36.2.1.2347-7.367.25.2567.567.1.34.5.3456.4567 
> 
> 
> 3.56.34567.6-567.5.3.347.4.2347.345.56.236.36.2347.36.236.6.23.234.34.4.36.2567.25.56.6.5.3.567.2 
> 
> 
> 345.567.56.1.5.2.56.6.34.27.347.345.6.256.34.27-6.4.45-34.7.6.7.567.25.47.5.34567.36.27 
> 
> 
> 34567.567-4567.2.5.7.23.7.47.347.2345.34.3456.34.5.34.4.234.1.234.34.27.3.256.5.25.67.34567.567.3.456 
> 
> 
> 3-3456.4567.256.27.5.1.4.34.234.25.2.56.3456.3.2.256-25.34.2.47.36.256.567.25.4.345.347-27 
> 
> 
> 34.7.34567.45-567.7.1.47.27-23.256.23456.234.3.234.6.23456.2345-47.2.23.2567-256.45-7.347.2367 
> 
> 
> 3456-567.6.5.27.256.2367.4-347.5.4.2.4.345.36-56.3.234.2.47.36.27.56.5.47.36-36.1 
> 
> 
> 3.3456.345.1-25.567.2367.2-34.25.23456.236.4.1.256.234.6.2345-27.47.367.2567.2-47.345.3456.7.2347 
> 
> 
> -36.34567.6.7.5.56.23.47.4.47.3.6.2.56.345.256.36.234.25.47.4.2347.236.7.27-1-347.34567.6 
> 
> 
> 56.345.34567.4567.56.5.27.1.347-2347.25.23456.4.2.5.36.256.236.45.7.27.7.23.2567.2.25.45.567.3.5.67 
> 
> 
> 3456.5.3456.47.56.256.567.2367.234.7.2.345.2.3456.236.23.6.23456.36.23.4.27.2347.6-5-47.345.347.345.2367 
> 
> 
> -36.345.456.7.5.567.236.4.7.234.3.36.23456.2.347.4.34.36.345.7.47.2.6.25.567.2567.4.34.5.567.2367 
> 
> 
> 36.345.347.6.56.25.27.6.47-4.23.6.2.234.23.3456.23456.234.45.347.4.234.36.2-256.4567.347.34567.7.67 
> 
> 
> 34567.5.347.6.256.27.256.3.34.27-25.4-4.23.36.234.3456.45.347.27.2.1.5.27.2567.67.3.3456.7.23 
> 
> 
> 347-345.4.2567.56.5.236.2347.47.27.45.34.2.234.45.236.3456-3.4.2.27.36.2567.27.7.1.567.3456.3.6 
> 
> 
> 7-36.4567.25.56.2.6.347.47.347.345.6.34.2.34567.34.6.56.23.2347.234-3.5.7.25.4.3.56.36.4 
> 
> 
> 7.34.5.456.256-2.236.4.7.347.5.256.6.56.25.4.6.56.23.27.347.2347.236.256.2567.25.45.3456.3-6 
> 
> 
> 3.5.34567.4.56.2567.27.36.47.7.47.23.236.56.6.45.56.36.34.3.4-47.23.7.56.2567.4.5.7.347.6 
> 
> 
> 567.36.3456.47.256.27.5.36.234.347-2345.236.234-1-56.3456.1.34.2.34.36.2.2567.25.47.56.3456.347.47 
> 
> 
> *   *
> 
> I like the complete lack of subjectivity in the method selection in this 
> approach.
> 
> *   *
> 
> All that’s needed now is a composition…
> 