Gradient-based approximation

This section discusses the theoretical methods adopted to explore the structure, energetics, and properties of sodium clusters. The scalability of ab initio methods and cumbersome computational efforts necessitate one to confine to Kohn-Sham DFT theoretical models. From the literature, it is clear that the local density approximation (LDA) is good for exploring relative energies of structures, while the gradient-based approximation (GGA) functionals work well for property evaluations [21]. As noted in our previous work [15], the trends in energetics and properties of... 

To improve upon this, from a chemical point of view rather crude assumption, the most widely employed corrections are based on using not only the density, but also its gradient. These corrections form the so-called generalized gradient approximation, GGA, or gradient expansion approximation (GEA) methods ... 

Model-based approaches allow fast derivative computation by relying on a process model, yet only approximate derivatives are obtained. In self-optimizing control [12,21], the idea is to use a plant model to select linear combinations of outputs, the tracking of which results in optimal performance, also in the presence of uncertainty in other words, these linear combinations of outputs approximate the process derivatives. Also, a way of calculating the gradient based on the theory of neighbouring extremals has been presented in [13] however, an important limitation of this approach is that it provides only a first-order approximation and that the accuracy of the derivatives depends strongly on the reliability of the plant model. 

The calculation of excited-state energies and of gradients based on the frozen ionic bonds approximation (as outlined in Ref. 45) is, from a computational point of view, considerably less demanding in comparison with other approaches such as RPA, CASSCF or Cl, and provides comparable accuracy. Therefore, this approach allows to carry out adiabatic molecular dynamics in the excited state, by calculating the forces on the fly (cf Ref. 45) applicable to relatively large systems. This is particularly convenient for the simulation of time-dependent transitions for which an ensemble of trajectories is needed. Moreover, the fast computation of nonadiabatic couplings on the fly allows one also to carry out nonadiabatic MD as outlined in Ref. 46. Of course, the application is limited to systems for which the frozen ionic bonds approximation offers an adequate description. 

Samples were clamped on its base on a copper water cooled table (Fig 13), so that in addition to the electric arc-induced thermal shock of about 500°C/s, a gradient of approximately 500°C/mm was present in the clamping area, reproducing the temperature gradient of point C in figures 1 and 13. After peak temperature of 1870°C was reached, samples cooled down... 

Gradient methods discussed above use a quadratic function (energy, gradient and approximate Hessian) to model the energy surface near the transition state. Distance-weighted interpolants provide a more flexible functional form that can interpolate arbitrarily spaced points with a smooth differentiable function. For a gradient-based optimization, the Shepard interpolation functions seem appropriate... 

Learning involves thus designating the miiumum of the error function. For this purpose, we usually apply gradient methods (conjugate gradients), based on the Hessian matrix (Newton, Levenberg-Marquardt methods), or on approximation of the inverse of the Hessian matrix (quasi-Newton methods) [2]. The MLP neural network is sensu stricte a static model, but it is possible to introduce dynamics... 

Fig. 5. RNA of plastid fraction (washed 1000g pellet) isolated from leaves maintained in darkness or illuminated for 3 hours after absorption of P-phosphate. RNA was extracted and centrifuged as described in legend for Fig. 3, but fractions here were collected manually and optical densities at 260 m/ were determined after dilution. (The two heavier RNA s, when obtained from maize chloro-plasts purified on a sucrose density gradient, are approximately 22 S and 17 S. The apparent base composition of these RNA s is approximately 25% adenine, 32% guanine, 20% uracil, and 23% cytosine these values are based on the distribution of radioactivity, after paper electrophoresis, among AMP, GMP, UMP, and CMP obtained by alkaline hydrolysis of plastid RNA s prepared from sucrose gradient purified plastids of maize leaves supplied P-phosphate.)...

---