Derivative approximations 2-point

The Hammett equation is the best-known and most widely studied of the various linear free energy relations for correlating reaction rate and equilibrium constant data. It was first proposed to correlate the rate constants and equilibrium constants for the side chain reactions of para and meta substituted benzene derivatives. Hammett (37-39) noted that for a large number of reactions of these compounds plots of log k (or log K) for one reaction versus log k (or log K) for a second reaction of the corresponding member of a series of such derivatives was reasonably linear. Figure 7.5 is a plot of this type involving the ionization constants for phenylacetic acid derivatives and for benzoic acid derivatives. The point labeled p-Cl has for its ordinate log Ka for p-chlorophenylacetic acid and for its abscissa log Ka for p-chloroben-zoic acid. The points approximate a straight line, which can be expressed as... 

Therefore we now turn to the noise part of the S/N ratio. As we saw just above, the two-point derivative approximation can be put into the framework of the S-G convolution functions, and we will therefore not treat them as separate methods. 

When the user, whether working on stand-alone software or through a spreadsheet, supplies only the values of the problem functions at a proposed point, the NLP code computes the first partial derivatives by finite differences. Each function is evaluated at a base point and then at a perturbed point. The difference between the function values is then divided by the perturbation distance to obtain an approximation of the first derivative at the base point. If the perturbation is in the positive direction from the base point, we call the resulting approximation a forward difference approximation. For highly nonlinear functions, accuracy in the values of derivatives may be improved by using central differences here, the base point is perturbed both forward and backward, and the derivative approximation is formed from the difference of the function values at those points. The price for this increased accuracy is that central differences require twice as many function evaluations of forward differences. If the functions are inexpensive to evaluate, the additional effort may be modest, but for large problems with complex functions, the use of central differences may dramatically increase solution times. Most NLP codes possess options that enable the user to specify the use of central differences. Some codes attempt to assess derivative accuracy as the solution progresses and switch to central differences automatically if the switch seems warranted. 

This second order approximation is a centered-difference scheme, since it expresses the spatial derivative at point i by means of data from symmetrically distributed points. All the implied information is known at timestep n. 

The above has considered arbitrarily spaced grids, whereas in practice, the spacing is often that of Feldberg [231]. In terms of points, the special case is a sequence of positions given by an exponentially expanding series of spatial intervals. This will be detailed in Chap. 7. Here, it is sufficient to mention that this special case makes the derivation of the coefficients for various derivative approximations easier and the expressions themselves more compact, as was reported by Martfnez-Ortiz [385]. That author also found that there is a particular value for the expansion parameter, q = /2j for which the asymmetric four-point second derivative, referred to the second of the four points u"(2,4), is third-order accurate, rather than second-order as for other parameter values or arbitrary placement of points. This can be of use in simulation. The four-point approximation has some good properties besides this, as will be explained in Chap. 7. 

For those who prefer to keep the derivative approximation of G down to the two-point form, the above can perhaps be simplified a little the u-v device is not needed as such, as only the first substitution (6.6) is required. 

Finally, as mentioned earlier (Chap. 7, Sect. 7.2), Martinez-Ortiz [385] developed some rather simple formulae for derivative approximations for the special case of exponentially expanding grid spacings, and in the course of this work discovered that the four-point second-order derivative approximation u"(T), for the expansion factor 7 = /2 is third-order in accuracy, rather than second, as it is for other 7 values. This could be an easy and useful way to increase the accuracy, using the four-point formula. 

In 1924, the Russian astronomer Numerov (transliterating his own name as Noumerov), published a paper [421] in which he described some improvements in approximations to derivatives, to help with numerical simulations of the movement of bodies in the solar system. His device has been adapted to the solution of pdes, and was introduced to electrochemistry by Bieniasz in 2003 [108]. The method described by Bieniasz is also called the Douglas equation in some texts such as that of Smith [514], where a rather clear description of the method is found. With the help of the Numerov method, it is possible to attain fourth order accuracy in the spatial second derivative, while using only the usual three points. The first paper by Bieniasz on this method treated equally spaced grids, and was followed by another on unequally spaced grids [107], The method makes it practical to use higher-order time derivative approximations without the complications of, say, the (6,5)-point scheme described above, which makes the solution of the system of equations a little complicated (and computer time consuming). 

The two methods that stand out in terms of efficiency and convenience are BDF and extrapolation. Both require minimal programming effort, and can be extended to higher-order spatial derivatives. However, in the case of BDF, a limit is encountered. For the most convenient start-up methods such as the simple or the rational start, the accuracy from BDF is limited to 0(8T2). This means for one thing that one need not go beyond 3-point BDF (which is 0 8T2) in itself), but that no marked improvement can be gained from higher-order spatial derivative approximations, because there will then be a mismatch between the accuracy orders with respect to the time and spatial intervals. 

This program is again a Cottrell simulation using second-order extrapolation based on the Bl (Laasonen) method and unequal intervals, but in contrast with the above program C0TT EXTRAP, this one makes use of the four-point spatial derivative approximation, and the GU-function. It performs a little better than the above program, at little extra programming effort. 

Their primary feature is the coverage of all possible nodal arrangements according to the requirements of the interaction. Thus, unlike Yee s approach that employs only two mesh points for derivative approximation, these concepts involve a full set of nodes which results in coarse but very convergent grid topologies. A typical m-directed (for L = 3), depicted in... 

A more accurate estimate of the first derivative is obtained from a third-order approximation. Fit a cubic through the points T(R), T(R — Ar), T R —2 Ar), and T(R — 3 Ar) and differentiate to estimate the slope at point R. The derivative approximation is... 

In order to derive approximate analytical expressions of Ult we need to know the distribution of the four transition points in the complex z-plane. The transition points are zero points of the following quartic polynomials (see Eqs. (61)) ... 

The approximations that we have developed here for the first and second derivatives at point i consider only three points i itself, and the two adjacent points. It is of course possible to formulate finite difference equations over a greater number of points than this using the same general technique, and doing so usually results in a higher-order error (a more accurate approximation). However, assuming that AX is sufficiently small, the increase in accuracy is minimal and typically not worth the increase in complexity. 

Analogically to the representation of the wave-function in structural terms, there is a way to separate (hyper)polarizabilities into the individual contributions from individual atoms. A method for such separation was developed by Bredas [15, 16] and is called the real-space finite-field method. The approach can be easily implemented for a post-Hartree-Fock method in the r-electron approximation due to the simplicity of e calculation of the one-electron reduced density matrix (RDMl) elements. In our calculations we are using a simple munerical-derivative two-points formula for RDMl matrix elements (Z ) [88] (see also [48]) ... 

In dealing with the atmosphere we must be concerned with the discretization of variables in the vertical as well as in the horizontal. There are two ways to represent the vertical stmcture of the atmosphere. One is to use discrete points in the vertical with vertical derivatives approximated by finite differences. The other is to represent variables as a finite series of differentiable functions. However, the grid-point approach dominates the spectral... 

---