[Nauty] geng with predefined colors

[Christoph Steinbeck]

Hi Brendan, 

thanks a lot for your response!
And yes, that does sound like a useful approach and your judgement on modifying geng was also very helpful. That was my conclusion from studying the code (OMG :)).
I’ll take a closer look at vcolg then.

Kind regards, 

Chris


> On 14. Apr 2021, at 07:59, Brendan McKay <Brendan.McKay at anu.edu.au> wrote:
> 
> Hi Chris,
> 
> geng is highly tuned to do what it does quickly and converting it to do something else would be difficult.  I wouldn't try that even though I understand the code pretty well.
> 
> If you are just colouring vertices and not edges, you should get faster performance by feeding graphs from geng into vcolg.
> 
> At the moment vcolg only allows you to specify the number of colours and the sum of the colours (where the colours are 0,1,2,...). It should be fairly easy to adjust it so that the precise number of vertices of each colour is specified. That would be far easier than rewriting geng.
> 
> Let me know if that sounds like a useful approach.
> 
> Cheers, Brendan.
> 
> On 14/4/21 3:03 am, Christoph Steinbeck wrote:
>> Dear all,
>> 
>> I would like to modify the geng graph generator to work with a predefined set of colours rather than just one.
>> 
>> But despite the old saying that no one needs documentation because the code is obvious, I am having trouble to see whether geng could do this at all and where I could inject this into the code, say, after parsing an array of colours as an argument in the form aXbYcZ which is X vertices of color a, Y vertices of color b, etc.
>> 
>> Any hint greatly appreciated :)
>> 
>> Kind regards,
>> 
>> Chris
>> 
>> 
>> 
>> _______________________________________________
>> Nauty mailing list
>> Nauty at anu.edu.au
>> https://mailman.anu.edu.au/mailman/listinfo/nauty
> _______________________________________________
> Nauty mailing list
> Nauty at anu.edu.au
> https://mailman.anu.edu.au/mailman/listinfo/nauty