Fitting the parameters

The SPC/E model approximates many-body effects m liquid water and corresponds to a molecular dipole moment of 2.35 Debye (D) compared to the actual dipole moment of 1.85 D for an isolated water molecule. The model reproduces the diflfiision coefficient and themiodynamics properties at ambient temperatures to within a few per cent, and the critical parameters (see below) are predicted to within 15%. The same model potential has been extended to include the interactions between ions and water by fitting the parameters to the hydration energies of small ion-water clusters. The parameters for the ion-water and water-water interactions in the SPC/E model are given in table A2.3.2. 

It may be easier to fit the parameters by forcing them to follow specified functional forms. In earhest attempts it was assumed that the forms should be normahzable (have the same shape whatever the size being broken). With complex ores containing minerals of different friability, the grinding functions S and B exhibit complex behavior near the grain size (Choi et al., Paiiiculate and Multiphase Processes Conference Proceedings, 1, 903-916.) B is not normalizable with respecl to feed size and S does not follow a simple power law. 

In principle, the velocities c and ti can be determined by taking a series of pictures at a very high frequency of the flow through a transparent plastic tube. Because of the particle size distribution, each particle moves at a different velocity, and this makes this method difficult to apply in practice. We have therefore used an indirect method, where we have measured the pressure losses of pneumatic conveying for two mixture ratios and then fit the parameters so that Eq, (14.126) coincides as accurately as possible with measured pressure losses. 

H = di(Z—iy di are the potential parameters I is the orbital quantum number 3 characterizes the spin direction Z is the nuclear charge). Our experience has show / that such a model potential is convenient to use for calculating physical characteristics of metals with a well know electronic structure. In this case, by fitting the parameters di, one reconstructs the electron spectrum estimated ab initio with is used for further calculations. 

The parameters were fitting according to Kiibler s [2] calculations performed non-relativistically so, while fitting the parameters, we had to use the non-relativistic version of the Creen function metod with spin polarization. Next, the two parts of the potential were defined as... 

The power, [P], in the fractal power-law regime gives as the fractal dimension, d(. P = —df for each level of the fit, the parameters obtained using the unified model are G, Rg, B, and P. P is the exponent of the power-law decay. When more than one level is fitted, numbered subscripts are used to indicate the level—i.e., G —level 1 Guinier pre-factor. The scattering analysis in the studies summarized here uses two-level fits, as they apply to scattering from the primary particles (level 1) and the aggregates (level 2). 

After fitting the parameters of the model to the data, the the best tuning constants were found. The cost functional to minimize was the integral of the absolute value of the error (lAE) ... 

A smooth function f(x) is used quite often to describe a set of data, (x y,). (x2,y2),. . .,(x j,yN). Fitting of a smooth curve, f(x), through these points is usually done for interpolation or visual purposes (Sutton and MacGregor, 1977). This is called curve fitting. The parameters defining the curve are calculated by minimizing a measure of the closeness of fit such as the function S(k)= [y j - f(x j ) 2. ... 

Subset of a population that is collected Frank and Todeschini [1994] in order to estimate the properties of the underlying population , e.g., the sample parameters mean x and standard deviation s. In the ideal case of representative sampling, the sample parameter fit the parameter of the population ji and a, respectively. 

If R is known, it is possible to fit the parameters k, ktCC0, A, At, fcp and fct using kinetic data from a single experiment. Thus, if the reaction diffusion parameter is known from the unsteady state after-effect experiments, the kinetic constant evolution can be determined as a function of free volume, and thus conversion. More details about this method will be published elsewhere (18). 

The purpose of this chapter is to develop a collection of methods that allow the determination of the best set of parameters for a particular given model and one or a collection of measurements. In other words we fit the parameter(s) to the measurement(s). 

Figure 4-60. The solver window set for the task of fitting the parameters in Figure 4-59.

In the second case, where LLE are lacking, VLE data are used to fit the parameters in the UNIQUAC model. These parameters, for the three binaries, were obtained from the literature in which VLE data were given at the following temperatures ... 

After fitting the parameters of a model to a set of measurement data, criteria for the good-ness-of-fit are required. There will always be some differences between the measured data and the values calculated from the model. These differences may be due to the following... 

The standard deviation has been determined as ct = j where v is the number of degrees of freedom in the fit. The parameters for the molecular interaction /3, the maximum adsorption Too, the equilibrium constant for adsorption of surfactant ions Ki, and the equilibrium constant for adsorption of counterions K2, are thus obtained. The non-linear equations for the Frumkin adsorption isotherm have been numerically solved by the bisection method. 

Results from these experimental runs were used as x, q data records to fit the parameters of six ANNs. In the experimental effort, a different feedforward ANN was used after each intermediate secondary measurement was obtained in the simulation-based effort, only one ANN accommodates all secondary measurements, and averaged dummy inputs are used for those secondary measurements not yet obtained. In addition in the experimental effort, a different ANN was used for final thickness and final void content predictions in the simulation-based effort, one ANN was used to predict both final thickness and final void content. The advantage of using one ANN to predict all values of q is that the parameters of only one ANN need be fitted. Fitting the parameters of an ANN for each variable in q is much more time-consuming. The disadvantage, however, is that the parameters A and abias are the same for each variable in q when just one ANN is used as an on-line model. When a different ANN is used for each variable in q, the parameters in A and abias are unique for each of those output variables, which results in increased on-line prediction accuracy. Similar speed-versus-accuracy arguments apply to the choice of one ANN for all secondary measurements versus an ANN for each secondary measurement. 

The failure of Set IV to yield reasonable unit cell parameters for conforma-tionally free helices is mostly due to the relatively small value of Fitting the parameters to the scaled AMI data tended to force s for C-C interactions to extremely low values. When this parameter was restrained to he at least 0.05 kcal/mol, the final value of for the F-F interactions was required to he larger (0.33 kcal/mol) as noted above, and the value to be correspondingly smaller. A dihedral term that properly describes the torsional behavior of perfluorocarbons may allow the vdW e parameters to have more typical values (0.01-0.1 kcal/mol). 

Of course, it is possible to be much more quantitative. One approach is to model S(t) or p(w) and fit the parameters to the experimental data. In the course of many such simulations we observed that the peak shift from zero time, t (T), was very similar to the S(t) function itself. 

The second way is to fit the parameters to vapour pressures and liquid or vapour densities. With the van der Waals equation, in which the parameters a and b are constants, this can be done only at one temperature. For the equations, for which the parameters are a function of temperature this is done over a range of temperatures, which is discussed below. 

To fit the parameters of a model, there must be at least as many data as there are parameters. There should be many more data. The case where the number of data equals the number of points can lead to exact but spurious fits. Even a perfect model cannot be expected to fit all the data because of experimental error. The residual sum-of-squares S2esidual is the value of S2 after the model has been fit to the data. It is used to calculate the residual standard deviation. ... 

Tachida (1991) characterized the behavior of a population on a neutral network based on the parameter 4Mcr, where M is the population size and a is the standard deviation of the effects of mutations on the fitness. The parameter a is related to the tolerance of a residue in a protein. If cr is large, then the effect of mutations is large. Three types of behavior are identified in this model. In the case 4M[Pg.151]

When solving equation (2.129) we assumed, like Davydov and Myasnikov,100 the self-energy E to be independent of k but unlike those authors, we included the complete self-energy Eop + Eac in the iteration process. The calculation was carried out on a spectrum of 1500 complex E values (resolution 1cm-1), which was recalculated at each step. Satisfactory convergence was reached within ten iterations or less. The spectrum of y w) = 2 Im X(ct>) is presented in Fig. 2.15 with the corresponding absorption spectra (2.125) at different temperatures. The model was completely determined by fitting the parameters Xac and /op to the lowest-temperature spectrum (3 K). Then we obtained... 

The expressions in equations (7.139) to (7.143) are exact to second order in perturbation theory. There are also higher-order terms of the same operator form as given in (7.137) but such contributions are much smaller as long as the interaction terms are small compared with the separation of the n and states this is usually the case. It is important to appreciate that the form of the yl-doubling operator is the same even when these higher order effects are included. This is a real advantage of the effective Hamiltonian approach. The correct form of the Hamiltonian can be established by a limited perturbation treatment. Thus, no approximation is made in fitting the parameters of this Hamiltonian to experimental data. The limitations, such as they are, arise only when the parameters so determined are compared with theoretical expectations. 

---