anyone? can i provide something else?

is this what you want? this patch quantises the input to an arbitrary list of numbers (ascending order)

i think this is s better version-- it compares the mean of the logs of two adjacent ratios, rather than the mean of the ratios themselves

thanks a lot -- this second version is a lot clearer even if i don't understand all the math yet. before your response i went through the tedious business of unpacking the long list into pairs and then individually running these through my math bit. while this worked, it mostly pointed to some other larger issues with my patch. your solution is a lot more elegant of course and is going to make these further refinements a lot less tedious. at this point in learning max the major business seems to be figuring out how to avoid long tedious patching processes. which sometimes ends up being tedious itself, ha ha. thanks again terry.

another way (much simpler, with adjustable precision, but w/o avg):

in answer to your question about learning max without going up a 'tedious patching' path, i guess if a method starts to get repetitive and tedious, it usually indicates there is a better way to do it, but the only way i know is to get to know as many objects as possible, and study other people's patches. from andrzej's example you can see there is _always_ more than one way of figuring out a solution. if it's not too pressing, you can leave the problem for a an hour or so (or a day or so) and try to get back to it with a different approach-- there's a certain amount of intuition involved in working out solutions in any kind of (programming) problem; perhaps more so in max...

andrzej, can you explain the adjustable precision in your patch? it doesn't seem to 'round up' for me. for example if my list contains '.8 and 1.', even with an input of .99 my final output is still .8 . in terry's patch, any input > .9 will quantize to 1.

thanks--