Science

Simplest applications of Power Geometry to Differential Equations

Alexander Bruno

RUSSIA

The Power Geometry studies the properties of solutions to an equation through the power exponents of its monomials [2]. Below we explain this approach for a partial differential equation with two independent variables x1, x2, one unknown function x3 and without parameters. Let X = (x1, x2, x3). We define a differential monomial a(X) as a product of powers of coordinates X and derivatives  k+l x3/ xk1 xl2. The point Q = Q(a) ï ÷ 3 corresponds to the monomial a(X) by the following manner: the vector Q = (q1, q2, q3) corresponds to the product const xq11, xq22, xq33, the vector Q = (-k, -l, 1) corresponds to the derivative k+l x3/xk1xl2, the sum of the vectors Q(a) + Q(b) corresponds to the product of monomials ab. A differential polynomial f(X) is a sum of the differential monomials; to the polynomial f in ÷ 3 there corresponds a set S = S(f) of points Q of its monomials. The set S(f) is called the support of the polynomial f.

Power Geometry studies properties of solution to the equation f(X) = 0 and gives algorithms for finding the solutions basing on the structure of the set S(f) in ÷ 3. For instance, let the polynomial f(X) is quasihomogeneous, i.e. its support S(f) lies ina plane L, ortogonal to vector L = (l 1, l 2, l 3). Then the equation f(X) = 0 admits the Lie operator Åli xi /xi and has the self-similar solution of the form x3 = x1a y (x1-b , x2), where the support of the solution is in another plane L1 ortogonal to vector L . It means that points (0,0,1) and (a -b , 1,0) belong to the plane L1, i.e. if the plane L1 is defined by the equation l1q1 + l2q2 + l3q3 = c, then l3 = c, l1a = l3, l1b = l2. The function y satisfies an ordinary differential equation.

Let logX = (logx1, logx2, logx3)*, where the asterisk denotes the transposition. After the power transformation logY = AlogX, where A is real nonsingular square matrix, we have f(X) = g(Y)/h(Y), where g(Y) and h(Y) are differential polynomials and the support S(h) consist of one point Q = 0.So, the ratio g(Y)/h(Y) has the support S(g/h) = S(g). Moreover, S(f) = A* S(g), where A* is the transposed matrix A. That property allows to simplify the quasigomogeneous differential equation f(X) = 0 by means of a power transformation and a logarithmic transformation of the form log y1 = z1.

For the arbitrary differential polynimial f(X), we can find all its quasihomogeneous first approximations f̂ (X) using its support S(f) . Here the first approximation x3 = j ̂(x1, x2) of a solution x3 = j (x1, x2) to the equation f(X) = 0 is a solution to the corresponding first approximation f̂ (X) = 0 of the equation, i.e. if f(x1, x2,j ) = 0, then f̂ (x1, x2, j ̂)=0. Combining algorithms of the selection of first approximations of the equation, of power transformations and of logarithmic transformations in many problems we can resolve a singularity and obtain the asymptotic expansions of solutions. In [2] these algorithms are given for problems with any number of variables, parameters and equations.

In exapmles the equations describing combustion without a source and with a source are considered. For the last one, a new class of self similar solutions with blow up is found.

More detaled presentation see in Proceedings of the International Conference on Differential and Functional Differential Equations, Moscow, Russia, August 16-21, 1999.



[Contents]

homeKazanUniversitywhat's newsearchlevel upfeedback

© 1995-2008 Kazan State University