From: "Mark D. Niemiec" <mniemiec@interserv.com>

In the J programming language, this is denoted by G &. F and is called "G under F".

(J (http://jsoftaware.com) is the descendent of APL, also designed by Kenneth Iverson, and it includes functional programming operators, as well as using ASCII characters for readability.)

Under can also be used dyadically: F^-1( G( F(x), F(y)) for example: x +&.*: y NB. Euclidean distance (*: is square) x +&.% y NB. Electrical resistance in parallel (% is reciprocal) x -:@+ y NB. Arithmetic mean (-: is halve; F@G(x,y) = F(G(x,y)) ) x -:@+&.% y NB. Harmonic mean

Lucky for us, J is still readable in that brisk, bracing APL sense (-: (-: is smiley) An Amazon review of the book "Bypasses" mentions exp( add( log(x), log(y) ) ) In all these examples except arithmetic mean, F is nonlinear. --Steve