I've never seen that approximation before but there must be some restriction on the allowable values of X since the expression diverges for large positive or negative values of x, while tanh(x) converges to +1 nad -1, respectively.
cheers Miller On Fri, Jul 03, 2015 at 01:23:29PM +0900, i go bananas wrote: > there's a good approximation for tanh that can be made using just a couple > of simple arithmetic objects: > > formula is: tanh(x) ~= x*(27+x*x) / (27+9*x*x) > > it's close enough for every application i've ever come across, it's cheap > cos it only uses simple arithmetic, and it doesn't degrade the resolution > of your signal like a table lookup does. > > > > On Fri, Jul 3, 2015 at 12:56 PM, Chris McCormick <ch...@mccormick.cx> wrote: > > > On 03/07/15 03:06, Pagano, Patrick wrote: > > > Is there a way i can replace something for tanh~ > > > and inits in Vanilla 46? > > > > Using a pre-computed table for speed: > > > > https://github.com/chr15m/blockhead/blob/master/e_tanh.pd > > https://github.com/chr15m/blockhead/blob/master/e_tanh-help.pd > > > > Or install "blockhead" using deken. :) > > > > Cheers, > > > > Chris. > > > > -- > > http://mccormick.cx/ > > > > _______________________________________________ > > Pd-list@lists.iem.at mailing list > > UNSUBSCRIBE and account-management -> > > http://lists.puredata.info/listinfo/pd-list > > > _______________________________________________ > Pd-list@lists.iem.at mailing list > UNSUBSCRIBE and account-management -> > http://lists.puredata.info/listinfo/pd-list _______________________________________________ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list