This paper aims at obtaining, by means of integral transforms, analytical approximations in short times of solutions to boundary value problems for the one-dimensional reaction-diffusion equation with constant coefficients. The general form of the equation is considered on a bounded generic interval and the three classical types of boundary conditions, i.e., Dirichlet as well as Neumann and mixed boundary conditions are considered in a unified way. The Fourier and Laplace integral transforms are successively applied and an exact solution is obtained in the Laplace domain. This operational solution is proven to be the accurate Laplace transform of the infinite series obtained by the Fourier decomposition method and presented in the literature as solutions to this type of problem. On the basis of this unified operational solution, four cases are distinguished where innovative formulas expressing consistent analytical approximations in short time limits are derived with respect to the behavior of the solution at the boundaries. Compared to the infinite series solutions, the analytical approximations may open new perspectives and applications, among which can be noted the improvement of numerical efficiency in simulations of one-dimensional moving boundary problems, such as in Stefan models.
Uniqueness from pointwise observations in a multi-parameter inverse problem
本文证明了一维反应扩散方程确定多个非常数系数反问题的一个唯一性结果。这种反应扩散方程包括Kolmogorov、Petrovsky和Piskunov的经典模型以及来自生物学的更复杂的模型。当反应项中含有一个非常系数为$μ_k(x),$的$N,$次的未知多项式部分时,我们的结果给出了该多项式部分的唯一性的一个充分条件.这个充分条件只涉及逐点测量的反应扩散方程的解决方案$u$和它的空间导数$\partial u / \partial x$在一个单一的点$x_0,$在一个时间间隔$(0,ε).$。除了这个唯一性的结果,我们给了几个反例的唯一性,强调我们的假设的最优性。最后,在特定的情况下,N=2和$N=3,$,我们表明,这种逐点测量可以允许一个有效的数值确定未知的多项式反应项。