Explicit Convergence Methods

For some systems of equations it is possible to guess a value of a variable, and then use one of the equations to solve explicitly for a new calculated value of the same variable, Then the calculated value and the original guess are compared and a new guess is made. 

The new guess can be simply the calculated value (this is called successive substitution). Convergence may be very slow because of (1) a very slow rate of approach of to or (2) an oscillation of back and forth around The loop can even diverge. 

Therefore a convergence factor p can be used to speed up or slow down the rate at which X( , is permitted to change from iteration to iteration. 

Note that letting P — I corresponds to successive substitution. This method is illustrated in the following example. 

We have four equations and four variables 0. (AT)t, T 2, and 7 2- The iterative procedure is  

---

The development of these explicit-rjj methods has yielded a database of benchmark results for small polyatomic molecules. These calculations are listed as MP2-R12 and CCSD(T)-R12 in our tables. We have selected the version called MP2-R12/A as a benchmark reference for our study of the convergence to the MP2 limit. This is the version that Klopper et al. found to agree best with our interference effect. The close agreement with extrapolations of one-electron basis set expansions justifies this choice. 

Another potential advancement is permitted in the ASPEN system. Tear streams can be designated as desired, so that a user might define blocks or series of blocks and simulate these sets as quasi-linear blocks. The convergence method could utilize this information and solve the material (and energy) balances explicitly. In this way, a simultaneous modular architecture could be utilized. Implementation of these programs will be for later enhancements of ASPEN, not the initial version. 

Iteration and convergence method explicit equations Monotone sequences and secant method Newton- Raphson Free ion molali-ties by difference Newton- Raphson conti nued fraction Newton- Raphson Newton-Raphson conti nued fraction conti nued fraction for anions only conti nued fraction conti nued fraction conti nued fraction brute force... 

The formula is known both as the improved Euler formula and as Heun s method [170]. This method has roughly the stability of the explicit Euler method. Unfortunately, the stability may not be improved by iterating the corrector because this iteration procedure converges to the trapezoid rule solution only if At is small enough. 

The slow X 3 convergence cannot easily be avoided. To improve on it, one can try to describe electron correlation not only by means of virtual orbital excitations but also by means of spatial two-electron basis functions that depend explicitly on the electron-electron distances. This is the idea behind the explicitly correlated methods [69]. Indeed, in this manner, it is possible to accelerate the convergence from X 3 to X 7, greatly reducing the basis set requirements for high accuracy. Nevertheless, such calculations are complex and cannot yet be used routinely on large molecules. 

The simulation is performed by using the explicit Euler method as predictor and the implicit Euler method as corrector (with fixed point iteration iterated %ntil convergence , see Sec. 4 1-1)- I 4- l he results for the rotation of the chassis obtained with h = 10 are plotted. The thickness of the band around this solution curve corresponds to the estimated global error increment magnified by 5 10. This figure reflects clearly the influence of the second derivative of the solution on the size of the error. 

Another option is a q,p) = p and b q,p) = VU q). This guarantees that we are discretizing a pure index-2 DAE for which A is well-defined. But for this choice we observed severe difficulties with Newton s method, where a step-size smaller even than what is required by explicit methods is needed to obtain convergence. In fact, it can be shown that when the linear harmonic oscillator is cast into such a projected DAE, the linearized problem can easily become unstable for k > . Another way is to check the conditions of the Newton-Kantorovich Theorem, which guarantees convergence of the Newton method. These conditions are also found to be satisfied only for a very small step size k, if is small. 

The discrete protonation states methods have been tested in pKa calculations for several small molecules and peptides, including succinic acid [4, 25], acetic acid [93], a heptapeptide derived from ovomucoid third domain [27], and decalysine [61], However, these methods have sofar been tested on only one protein, the hen egg lysozyme [16, 61, 71], While the method using explicit solvent for both MD and MC sampling did not give quantitative agreement with experiment due to convergence difficulty [16], the results using a GB model [71] and the mixed PB/explicit... 

Some equations such as/(x)=0 cannot be explicitly solved for x. If multiple solutions are not expected in a narrow range, Newton s method is often simple to implement and has faster convergence than the natural method of interval splitting. The method is recursive and uses the first-order expansion off (x) in the vicinity of the fcth guess... 

Initially, we develop Matlab code and Excel spreadsheets for relatively simple systems that have explicit analytical solutions. The main thrust of this chapter is the development of a toolbox of methods for modelling equilibrium and kinetic systems of any complexity. The computations are all iterative processes where, starting from initial guesses, the algorithms converge toward the correct solutions. Computations of this nature are beyond the limits of straightforward Excel calculations. Matlab, on the other hand, is ideally suited for these tasks, as most of them can be formulated as matrix operations. Many readers will be surprised at the simplicity and compactness of well-written Matlab functions that resolve equilibrium systems of any complexity. 

W1/W2 theory and their variants would appear to represent a valuable addition to the computational chemist s toolbox, both for applications that require high-accuracy energetics for small molecules and as a potential source of parameterization data for more approximate methods. The extra cost of W2 theory (compared to W1 theory) does appear to translate into better results for heats of formation and electron affinities, but does not appear to be justified for ionization potentials and proton affinities, for which the W1 approach yields basically converged results. Explicit calculation of anharmonic zero-point energies (as opposed to scaling of harmonic ones) does lead to a further improvement in the quality of W2 heats of formation at the W1 level, the improvement is not sufficiently noticeable to justify the extra expense and difficulty. 

---