freepoulite

Does max developers plans to integrate rational numbers to Max?

I'm a bit curious about it because as soon as you want to do some "complex" rythmn things you really need it to keep precision.

the awesome bach library support it in many ways but it is far much slow than native types in max and it become a problem when you need high speed calculation in real time music context

Does anyone know a faster alternative? (external, abstraction)

I know that the next update of bach will clearly increase speed but I'm a bit hurry...

Does max developers plans to integrate rational numbers to Max?
I'm a bit curious about it because as soon as you want to do some "complex" rythmn things you really need it to keep precision.
the awesome bach library support it in many ways but it is far much slow than native types in max and it become a problem when you need high speed calculation in real time music context
Does anyone know a faster alternative? (external, abstraction)
I know that the next update of bach will clearly increase speed but I'm a bit hurry...

native-rational-numbers-in-max

No clue about Cycling's plans of course... but would you mind posting an example in which you think bach's rationals are the bottleneck? This might be very useful for us to do some profiling.

On the other hand, it won't be difficult at all implementing a small library of abstractions for dealing with rational numbers - you can express your rationals as lists of numerator and denominator, and the only relatively tricky thing would be computing the lcm.

although that's indeed not an 'out-of-the-box' solution, you can deal with rational numbers (as Andrea already pointed out) if you represent them as lists of two numbers (numerator & denominator) quite easily. To compute the lcm and the gcd you could use the appropriate externals from my library as well (

), which are both quite well speed-optimized (see 

although that's indeed not an 'out-of-the-box' solution, you can deal with rational numbers (as Andrea already pointed out) if you represent them as lists of two numbers (numerator & denominator) quite easily. To compute the lcm and the gcd you could use the appropriate externals from my library as well ([sadam.gcd] and [sadam.lcm]), which are both quite well speed-optimized (see http://www.sadam.hu/software ).

I've tried to make my own rational abstraction but it is a bit slow too

Look at this patch
I've tried to make my own rational abstraction but it is a bit slow too


thanks a lot for your help and for your libraries!

Thank you for the patch, I'll look into it asap!

 & friends generate up A LOT of cpu usage. If you use simple lists instead of the symbol-based representation, you can make your 

actually the [fromsymbol] & friends generate up A LOT of cpu usage. If you use simple lists instead of the symbol-based representation, you can make your [p ratSimp] patch around 10 times faster:

i guess what uses the most CPU in bach is the the sublist formatting, not the actual process.

using custom abstractions should be fine, i see no need for an external.

btw, Timersintfloatratnumrat.maxpat simply chrashes my max 5 app.

btw2, i agree with the idea that the most effective solution is to use a list of 2 integers (as opposed to

formatting symbols such as "ratnum_4|3" or similar bs.)

btw2, i agree with the idea that the most effective solution is to use a list of 2 integers (as opposed to
formatting symbols such as "ratnum_4|3" or similar bs.)


how about this small set of abstractions?

This is tangential, but Clojure does support rational numbers and can be used inside Max. Haven't done it myself, but if you wanted to go all OpenMusic / PWGL on it...

This is tangential, but Clojure does support rational numbers and can be used inside Max.  Haven't done it myself, but if you wanted to go all OpenMusic / PWGL on it...


Thank you a lot your abstractions it looks really simple and fast!

Thanks to everyone for your contributions

@Adam
Thank you a lot your abstractions it looks really simple and fast!

native rational numbers in Max