The Boundary Value Approach

Furthermore, at least for bad initial guesses of initial values or parameters the initial value problem may even have no solution at all, while the original PI problem is stable and well conditioned. 

Example 7.2.2 This can be seen, when solving the truck example for different values of the adiabatic coefficient k. When k exceeds a value between 1,542 and 1.545, the relative translation p o changes its sign. This leads to the physically unrealistic situation shown in Fig. 7.2. Although technically irrelevant, these values of K can occur within the optimization loop as intermediate results. 

To overcome this difficulty one attempts to feed into the solution process already during integration information from measurements[Bock87]. This makes multiple 

One chooses a mesh covering the time interval under consideration 

The corresponding trajectory is in general discontinuous, see Fig. 7.3. For x to be a solution of the overall initial value problem we have to require continuity in the nodes  

---

The second limitation is harder to tackle rigorously, however as we discuss in Sects. 3.3 and 3.4, the boundary value approach can be used to suggest approximate solutions with much larger time steps. Increasing the time step is much harder to achieve with initial value formulation. 

There are important differences between the initial value formulation and the boundary value approach. Initial value solutions are based on interpolation forward in time one coordinate set after another. The boundary value approach is based on minimization of a target function of the whole trajectory. Minimization (and the study of a larger system) is more expensive in the boundary value formulation compared to initial value solver. However, the calculations of state to state trajectories and the abilities to use approximations (next section), make it a useful alternative for a large number of problems. 

Short-cut methods derived from conventional distillation have been foimd to be inappropriate for the estimation of reflux ratio for reactive case and new approaches have been adopted. Methods derived from the boundary-value approach (c/. 3.2.4) allow the determination of the minimum and real reflux ratio, which are key factors in the determination of vapor and liquid velocities. The reflux ratio can be calculated using an iterative procedure, described in Subawalla and Fair (1999) and summarized in figure 3.9. 

We use the Galerkin approach to prove the existence of the solution to the boundary value problem (2.9)-(2.11). It is well known that the eigenvalue functions... 

Earlier we solved the boundary value problem for the spheroid of rotation and found the potential of the gravitational field outside the masses provided that the outer surface is an equipotential surface. Bearing in mind that, we study the distribution of the normal part of the field on the earth s surface, where the position of points is often characterized by spherical coordinates, it is natural also to represent the potential of this field in terms of Legendre s functions. This task can be accomplished in two ways. The first one is based on a solution of the boundary value problem and its expansion into a series of Legendre s functions. We will use the second approach and proceed from the known formula, (Chapter 1) which in fact originated from Legendre s functions... 

For larger r values, x(r) is forced further from the constraint boundary. In contrast, as r approaches zero, xx r) and x2(r) converge to their optimal values of 1.5 and 2.5, respectively, and the constraint value approaches zero. The term — ln(g(x)) approaches infinity, but the weighted barrier term — rln (g(x)) approaches zero, and the value of B approaches the optimal objective value. 

The boundary value problem (Eqs. (10), (11)) is usually solved numerically. However, it is also possible to use another approach employing a linearization of this second-order, non-linear problem and a subsequent analytical treatment The analytical solution of the linearized boundary value problem in the film region is obtained in [15] ... 

One can describe the essence of current distributions resulting from uneven pattern density by the following approach. Firstly, the discrete pattern is viewed as a continuum. At any given location, the fractional area available for electrolysis (i.e., the fraction not covered by resist) is described by a, the active-area density. Thus, the active-area-density distribution is the only attribute of the resist pattern. This is to say that the sizes and shapes of individual features are ignored in this model. Secondly, one writes the boundary value problem that is normally solved to calculate the secondary current distribution. The problem domain, shown in Fig. 4 is the volume of electrolyte limited below by the electrode surface, at the sides by symmetry boundaries, and at the top by an iso-current-density plane. Within this domain, the potential is assumed to obey the Laplace equation. 

The advantage of this approach is in the simplicity of both the potential equations and the boundary-value conditions. Biro and Preis (1990) demonstrated that the Coulomb gauge can be enforced by the following boundary-value condition on the surface dV of the modeling region ... 

A complementary approach to the initial value formulation is the calculation of trajectories as a solution of a boundary value problem. In this approach the two end points are given as input to the calculations. It is therefore obvious that the trajectories will end at the pre-set interesting configurations. This simple construction solves the second limitation on initial value calculations that we mentioned above. Of course if the end points are not known and only the beginning configuration is available (e.g. the protein folding problem), then the initial value approach is the only viable option. 

The two equations of motion we discussed so far can be found in classical mechanics text books (the differential equations of motion as a function of the arch-length can be found in [5]). Amusingly, the usual derivation of the initial value equations starts from boundary value formulation while a numerical solution by the initial value approach is much more common. Shouldn t we try to solve the boundary value formulation first As discussed below the numerical solution for the boundary value representation is significantly more expensive, which explains the general preference to initial value solvers. Nevertheless, there is a subset of problems for which the boundary value formulation is more appropriate. For example boundary value formulation is likely to be efficient when we probe paths connecting two known end points. 

The coordinate transformation makes all the velocities coincide. The boundary layer approaches the core flow asymptotically and in principle stretches into infinity. The deviation of the velocity wx from that of the core flow is, however, negligibly small at a finite distance from the wall. Therefore the boundary layer thickness can be defined as the distance from the wall at which wx/wx is slightly different from one. As an example, if we choose the value of 0.99 for Wj/uico, the numerical calculation yields that this value will be reached at the point r]+ m 4.910. 

A number of numerical calculations was carried out for various Reynolds numbers up to 2000, and the duct s initial region length was found as a graphical function Lx = f(Re h, A) represented in Fig. 3.11. The dependence Lx vs A significantly differs for small and big Reynolds numbers Re. The value A = 0 corresponds to the case where the EPR is absent. It can be seen that, for A -> 0, all the curves arrive at the constant value c 0.03, but the curves associated with small Reynolds numbers 1, 5, and 10. Hence, it can be concluded that the principal Schlichting s formula (3.45) has been justified for large Re despite it was derived from the boundary-layer approach, rather than from the complete Navier - Stokes equations. 

In support of the philosophical appeal is the fact that Eq. (4) is an integral while in Eqs. (2) we use derivatives. Numerical estimates of integrals are, in general, more accurate and more stable compared to estimates of derivatives. On the other hand, computations of the whole path are more expensive than the calculation of one temporal slice of the trajectory at a time. The computational effort is larger in the boundary value formulation by at least a factor of N, where N is the number of time shces, compared to the calculation of a step in the initial value approach. To make the global approach computationally attractive (assuming that it does work), the gain in step size must be substantial. 

From a consideration of the colour solid it is apparent that there is a maximum possible luminous reflectance or transmission for any specific colour value. The gamut of possible colour values diminishes as the Y value approaches 1, and a few discontinuous steps in the relationship between chromaticity and maximum luminance with standard source Sc are shown in Fig. 26.18. An interesting feature is that the highly saturated reds, yellows, and oranges situated near the spectrum loci boundary in the... 

A computational approach of a very different nature, usually referred to as the boundary value method , introduces a multidimensional finite-difference mesh to directly solve the partial differential equations of reactive scattering This approach has been taken by Diestler and McKoy (1968) in work related to a previous one by Mortensen and Pitzer (1962). While the second authors used an iterative procedure to impose the physical boundary conditions of scattering, the more recent work constructs the wavefunction as a linear combination of independent functions Xj which satisfy the scattering equation for arbitrarily chosen boundary conditions. 

Based on this approach, the spatial relaxation of the electron component has been studied in plasmas of neon and molecular nitrogen and for some electric field strengths, again using the gas density N = 3.54 10 cm and applying the parameter values = 5 eV and (/ , = 2 eV to fix the boundary value /i(t/) according to Eq. (59). 

Note that even if the result Eej = E (h) can be obtained from experiments, a theoretical approach will in general not be consistent if material parameters depend on geometrical data, such as the film thickness h. In the present example the situation is different according to the extension of the theory by the parameter K, a consistent formulation is obtained. The material parameters are constant values but due to the enhancement the model is able to predict boundary layers as part of the solution. The introduction of the effective stiffness according to Eq. (26) is only a re-interpretation of the results, i.e., it is based on the solution of the boundary value problem. 

An alternative stress analysis approach, is based on the local stress field near a crack tip. Fig. 5. The solution of the boundary value problem for a semi-infinite, linear-elastic cracked body was found by Williams [48], and yields the Williams expansion of the stress field in a cracked body. The first term determines the local stress field in the vicinity of the crack tip ... 

This is a useful approach, leading to a large variety of methods. Note that nothing has yet been said about how to handle the boundary values that are not subject to diffusion in themselves. The outer value is normally constant at the bulk value. The value at the electrode, Co, is the problem, and will be dealt with in the next section in detail. For the moment, let us dwell on the Set (20). We need to find discrete expressions for the left-hand sides, the time derivatives. Here, we can draw on the practice in the field of ordinary differential equations (ode s). Taking the ith equation out of (20) and expressing it in the simple form... 

It turns out that under suitable conditions, the asymptotics for both the initial value problem and the boundary value problem in the critical case have the same form as in Sections II.B and III.A. In particular, the solution of the initial value problem approaches one of the solutions of the reduced equation in the limit as /x. 0. But the algorithm for constructing the asymptotic expansion undergoes some changes. 

---