> | geo:= proc(u,b) you may easily check that geo(u-1,u)=geo(u,u) for all u ...!
local R; R:=2^((u-b)/2); return(evalf[20](1/(1-2^(-b-1))/R^2*product(1+R/2^(i/2),i=0..300) )); end; george:= proc(u) return(geo(u,ceil(u))); end; geoN:= proc(u)=geo(u,u) return(evalf[20](1/(1-2^(-u-1))*product(1+1/2^(i/2),i=0..300) )); end; geoM:= proc(u)=geo(u,u+1) return(evalf[25](2/(1-2^(-u-2))*product(1+1/2^(i/2),i=1..301) )); end; |
> |
> | geoN(0); geoN(10); geoN(20); geoN(30); geoN(40); geoN(50); geoN(60); geoN(70); geoN(80); geoN(90); geoN(100); |
> | plot([george(x),geoN(x)],x=0..20); actual curve in RED, and upper bound in GREEN... |
> | plot(geoN(x),x=0..30); |
> | plot3d(geo(x,x+e),x=4..30,e=0..1); for each x, the max is reached at b=u+1. Remember geo(u,u+1)=geo(u+1,u+1). So the curse is essentially symmetric for large enough values of u (u>30). |
> |