Iterative improvement

Iterative improvement of the solution of systems of linear equations... 

Crystallographic refinement is a procedure which iteratively improves the agreement between structure factors derived from X-ray intensities and those derived from a model structure. For macro molecular refinement, the limited diffraction data have to be complemented by additional information in order to improve the parameter-to-observation ratio. This additional information consists of restraints on bond lengths, bond angles, aromatic planes, chiralities, and temperature factors. 

It is a common experience in science when we try to solve a new problem to find that an exact solution proves elusive. If no closed-form solution to the problem seems to be available, some process of iterative improvement may be needed to move from the best approximate solution to one of acceptable quality. With persistence and perhaps a little luck, a suitable solution will emerge through this process, although many cycles of refinement may be required (Figure 5.1). 

The process of iterative improvement of an initially modest solution to a problem. 

After each iterative improvement all the gradient methods compute the gradient VR (p). The way they use the gradient is different for the variants (steepest descent, Marquardt-method,...) of the gradient method. 

For reasons discussed above, we needed a complementary, ancillary tool for comparison of the mass spectra of components from multiple urine samples. We desired that the procedure have several characteristics (1) requires little if any manual data entry by the operator (2) utilizes data automatically generated by ChemStation and organized into Microsoft Excel spreadsheets (3) displays both retention times and mass spectral data in the same window (4) minimizes subjective operator judgments and (5) is simple and rapid to use. What emerged after several iterative improvements are the FindPeak macros discussed below. These are largely due to the expertise of Y. Aubut, with valuable input from J. Eggert. 

Non-linear regression calculations are extensively used in most sciences. The goals are very similar to the ones discussed in the previous chapter on Linear Regression. Now, however, the function describing the measured data is non-linear and as a consequence, instead of an explicit equation for the computation of the best parameters, we have to develop iterative procedures. Starting from initial guesses for the parameters, these are iteratively improved or fitted, i.e. those parameters are determined that result in the optimal fit, or, in other words, that result in the minimal sum of squares of the residuals. 

Currently, this is a major application of protein crystallography in most of the major drug companies. One of the best examples of this approach is the design of inhibitors for HIV protease (Dash et al., 2003). In brief, once the 3-D structure of HIV protease was determined, the active site was identified and used to screen small molecule libraries for potential compounds that could bind to HIV protease. These compounds were then tested for their ability to inhibit the protease. Lead compounds were then used to iteratively improve the inhibitors, using crystallographic studies, computational modeling, and biochemical tests. 

UPS hires people that will fit into its culture and who will fit into this iterative improvement and measurement work place. Remember, the people that UPS avoids are those wanting a fast track to the top. UPS is looking for those who want to be part of a team that is the best at what it does and who love the blocking and tackling of business. The payoff for a job well done is the opportunity to... 

While the residual r can be considerable reduced by iterative improvement, in many problems this does not mean that the residual error e will be also small. To relate e to r, define the norm IIAII of the square matrix A by... 

Solution of the equations is a process in which the coefficients of Eq. (14.28) are iteratively improved. To start, estimates must be made of the flow rates of all components in every stage. One procedure is to assume complete removal of a light key into the extract and of the heavy key into the raffinate, and to keep the solvent in the extract phase throughout the system. The distribution of the keys in the intermediate stages is assumed to vary linearly, and they must be made consistent with the overall balance, Eq. (14.27), for each component. With these estimated flowrates, the values of xik and yik are evaluated and may be used to find the activity coefficients and distribution ratios, Kik. This procedure is used in Example 14.9. 

I will discuss the iterative improvement of phases and electron-density maps in Chapter 7. For now just take note that obtaining the final structure entails both calculating p(x,y,z) from structure factors and calculating structure factors from some preliminary form of p(x,y,z). Note further that when we compute structure factors from a known or assumed model, the results include the phases. In other words, the computed results give all the information needed for a "full-color" diffraction pattern, like that shown in Plate 3d, whereas experimentally obtained diffraction patterns lack the phases and are merely black and white, like Plate 3e. 

In order to bypass this problem, a clever idea has been introduced the laboratory feedback control technique [20]. The optimization procedure is based on the feedback from the observed experimental signal (e.g., a branching ratio) and an optimization algorithm that iteratively improves the applied femtosecond-laser pulse. This iterative optimum-seeking process has been termed training lasers to be chemists [21]. 

Coincident with the development of sampling procedures were the constant iterative improvements in extraction, separation, identification and quantitation of organic compounds. Special emphasis was placed on selected compound classes such as the polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs), chlorinated benzenes, and chlorinated dibenzo-p-dioxins (dioxins). The best available procedures were used to determine these components because they have known acute or chronic effects and previous studies suggested that they might be present in effluents from the combustion of coal alone and combination coal/RDF. 

To date, a number of well-established strategies exist to select protein-ligand interactions. These can be exploited to identify binding molecules (combinatorial approach) or iteratively improve them (evolutionary approach). Additionally, if the binding interaction is restricted to the native... 

The process of evaluating and reporting the quality of information used in an exposure assessment is likely to be an iterative process as analysts identify and communicate key sources of uncertainties in the assessment results, they may be asked to take steps to incorporate new data and methods, to improve characterizations of variability and/or to reduce uncertainty. Again, however, the need for iterative improvement of an assessment depends on the needs of the decision for which the assessment is designed. 

Molecular structures obtained from least-squares fitting of experimental rotational parameters (or isotopic differences of such) to the corresponding quantities calculated from an approximate, but iteratively improved structure, have been called... 

The aim of the optimisation is to determine the spatial pattern of retrofit flue gas desulphurisation FGD which, for a given total installed capacity of abatement, minimises the magnitude of the difference between the deposition loads and the critical loads for total sulphur deposition at the receptor sites. Such differences between deposition loads and ciritical loads are termed critical load exceedences. For a near continuous distribution of emission controls at about 50 power stations and 11 receptor sites, there are a large number of possible strategies to work through in an exhaustive analysis. The problem was solved using optimisation by simulated annealing [8-10], a specialised iterative improvement technique. 

The algorithm starts with a (very bad) estimate of the density of states g E) = 1 for all energies which is iteratively improved by a modification factor / in the following loop ... 

The separate optimization of the internal orbital rotations at the beginning of each macro-iteration improves convergence considerably. However, this treatment so far neglects the coupling to the internal-external rotations. This coupling creates additional rotations R,y between the internal orbitals when the non-linear equations (53) are solved. Convergence can be further improved... 

Finally, we have also developed an improved method for the study of photolonlzatlon cross sections which Is based on the direct use of Schwlnger-type variational expressions for the electric dipole matrix elements themselves (16). These more general variational expressions can also be Iteratively improved. The analysis of these approaches revealed that the Iterative Schwinger method which we have outlined above leads to variationally stable photolonlzatlon cross sections. 

---