We discuss the long-time behavior of Andreoli's differential equation for genetic algebras and for Bernstein algebras and show convergence to an equilibrium in both cases. For a class of Bernstein algebras this equilibrium is determined explicitly.
Continuous time modelsAndreoli [1] introduced the differential equation ~ : X 2 --X, x ( 0 ) = y in the standard simplex of a genetic algebra with genetic realization to model the time dependence of the genotype frequencies of a population in the limiting case of continuously overlapping generations. While Andreoli restricted his attention to three-dimensional systems, Heuch [5] later considered genetic algebras of arbitrary (finite) dimension and showed that this equation can be solved by elementary functions; see also W6rz-Busekros [9].The long-time behavior of the solutions seems to have been unknown so far in the general case, although it was known for special classes of algebras [9]. We show in this note that for any genetic algebra with genetic realization every solution in the standard simplex tends to a stationary point for t--* oo.Using different methods, we show that the same holds for Bernstein algebras (which are not necessarily genetic).In the remainder of this section we derive some preliminary results about the differential equation 2 = x 2 -x in a real or complex commutative algebra A. The first one is proved in [7]:Lemma 0.1 Let G(y, t) resp. S(y, t) be the solution of 2 = x 2 resp. 2 = x 2 -x with x(O) = y. Then S(y, t) = e -" G(y, 1 -e -9 , wherever both sides are defined. Now let G(y, t) = ~k>~o tkgk(Y) be the Taylor series expansion about t = 0. From [7] it is known that the gk satisfy the recursion