Motivated by a parameter identification problem for a fourth-order partial differential equation which models the design of car windscreens by the sagging process, we numerically study a parameter identification problem which is of second-order as an equation for the parameter. The error magnitude and structure for a regularized Landweber method for solving this nonlinear inverse problem with interior data depend markedly on the type of this second-order equation.

We consider heat transfer processes where radiation in a large number of frequency bands plays a dominant role. In the simulation of such processes, the radiative transfer equation has to be solved repeatedly. To obtain an efficient and accurate solution method, we propose a new hybrid algorithm which combines fast solvers for the radiative transfer equation in the low and high absorption regime, respectively. A key role is played by an exact, residual based error formula. The algorithm is applied to a cooling problem of high quality optical glass.

We present an exact residual based error formula in natural norms for a class of transport equations. The derivation of the error formula relies on an abstract formulation in a general Hilbert space setting. The key role is played by the validity of an inversion formula. Its verification is for particular radiative transfer equations equivalent to the identification of strong and weak traces. The residual based error formula can be used in the design of efficient and accurate simulations of the cooling process of high quality glass [5].

We present a coupled system of elliptic equations describing the steady state of the thermoelectrical behaviour of an aluminium electrolytic cell. The thermal model is a free boundary problem which consists of the heat equation with Joule heating as a source. We neglect the Joule heating in the ledge, and allow for temperature-dependent electrical conductivity. We also formulate a numerical approximation using a finite element method. An iterative algorithm and numerical results are presented. The existence of a weak solution is also proved.

We present a multi-parameter non-constant-invariant class of Abel ordinary differential equations with the following remarkable features. This one class is shown to unify, i.e. it contains as particular cases all the integrable classes presented by Abel, Liouville and Appell, as well as all those shown in Kamke's book and various other references. In addition, the class being presented includes other new and fully integrable subclasses, as well as the most general parameterized class of which we know whose members can systematically be mapped into Riccati equations. Finally, many integrable members of this class can be systematically mapped into an integrable member of a different class. We thus find new integrable classes from previously known ones.

We present an algorithm for solving first-order ordinary differential equations by systematically determining symmetries of the form $[\xi=F(x),\, \eta=P(x)\,y+Q(x)]$, where $\xi\; \pa/\pa x + \eta\; \pa/\pa y$ is the symmetry generator. To these linear symmetries one can associate an ordinary differential equation class which embraces all first-order equations mappable into separable ones through linear transformations $\{t=f(x),\,u=p(x)\,y+q(x)\}$. This single class includes as members, for instance, 429 of the 552 solvable first-order examples of Kamke's [12] book. Concerning the solution of this class, a restriction on the algorithm being presented exists, only in the case of Riccati equations, for which linear symmetries always exist, but the algorithm will only partially succeed in finding them.

We perform an analytical and numerical study of the crossover from the Josephson effect to that of bulk superconducting flow for a one-dimensional superconductor containing a sandwich layer of normal material. A generalized Ginzburg–Landau (GL) model, proposed in Chapman & Gunzburger [1] is used in modelling the whole structure. When the thickness of the normal layer is very small, the introduction of three effective $\delta$-function potentials of specified strength leads to an exact analytical solution of the modified stationary GL equation. The resulting current density-phase offset relation is analyzed numerically. We show that the critical Josephson current density $j_c$ corresponds to a bifurcation of the solutions of the nonlinear boundary value problem for the modified GL-equation. The influence of the second term in the Fourier-decomposition of the supercurrent density-phase relation is also investigated. We derive also a simple analytical formula for the critical Josephson current.