Solution of the deconvolution problem in the general statement (Q1675336)

The following Volterra integral equation of the first kind is considered \[ Ag\equiv \int\limits_{0}^{t}q(t-\tau)g(\tau)d\tau=\Delta p(t), \quad t\in[0,T]. \] This equation arises in the description of well testing processes and this problem is known as the deconvolution problem. Here \(\Delta p(t)=p_0-p(t)\) and the unknown function \(g(t)\) depends on the functions \(p(t)\) and \(q(t)\) measured with large errors (from 5 to 15). The kernel \(q(t)\) is the flow rate and the right-hand side \(\Delta p(t)\) is the pressure change. The input data \(p(t)\), \(q(t)\) and the solution \(g(t)\) of this equation can have considerable variations on small time intervals (multiscale behavior). The functions \(p(t)\) and \(q(t)\) are bounded but, as a rule, discontinuous. To solve the problem, the authors use the variational regularization methods and construct a function basis (a system of exponents), which allows to take into account in the algorithm all a priori constraints known for the desired solution. They construct a family of approximate solutions that satisfies the conditions of smoothness and exactness required for the interpretation of well tests. For the constructed regularizing algorithms, the convergence theorems are formulated. Some details of numerical implementation are also described.