Local algorithm

A recently often used practical method is that of proposed by Pipek and Mezey [26], Their intrinsic localization is based on a special mathematical measure of localization. It uses no external criteria to. define a priori orbitals. The method is similar to the Edmiston-Ruedenberg s localization method in the a-n separation of the orbitals while it works as economically as the Boys procedure. For the application of their localization algorithm, the knowledge of atomic overlap integrals is sufficient. This feature allows the adoption of their algorithm for both ab initio and semiempirical methods. The implementation of die procedure in existing program systems is easy, and this property makes the Pipek-Mezey s method very attractive for practical use. 

A local algorithm for Metropolis Monte Carlo trajectories must be constructed carefully. Due to the finite probability of exactly repeated states in these paths, the corresponding transition probability includes a singular term [see Eq. (1.14)]. The generation algorithm for local path moves must take this singularity into account properly. Appropriate acceptance probabilities are given in [5]. (H. C. Andersen has drawn our attention to an omission in [5]. In Metropolis Monte Carlo trajectories sequences of multiple rejections can occur. Attempts to modify time slices in the interior... 

The following section starts with the description of automatic onset detection methods, which are extremely useful when dealing with large data sets, but are the most error-prone of the localization algorithms. 

Three dimensional localization requires the onset times from at least 4 sensors. The most common approach is to use an iterative localization algorithm, which requires the linearization of the problem. To do this, a first guess or trial h q)ocenter xo, yo, zo, to) is required. This first guess hypo-center must lie relatively close to the tme hypocenter, which is not known. The travel time residuals r, of the first guess h5qx)center are then a linear fimction of the correction in h q)ocentral distance (Havskov et al. 2002). 

This localization algorithm also provides the possibility to correct the body wave velocity iteratively. To do this, velocities are calculated from the travel time of all events (e. g. recorded during a certain period of the experiment) and the distance calculated between sensor and hypocenter. The linear extrapolation of all these calculated velocities gives a new average velocity for the localization. This procedure can be performed iteratively, in combination with the localization. 

Using the same theoretical example as mentioned above. Fig. 6.16 illustrates the effect of a systematically erroneous arrival time on the source localization. For an array of 40 by 40 AE-sources the theoretical arrival times at the four sensors were calculated. To introduce an error, 5 ps were added to the arrival times of Sensor 1. The iterative localization algorithm then yields AE-source locations that minimize the travel time residuals and thus distribute the error in arrival times over all sensors. Fig. 6.16 top depicts the difference between the actual and the calculated AE-source location, on the left side as error vectors and on the right side as a density function of the error value. Fig. 6.16 bottom left shows a density function of the minimized travel time residuals (mean value over all sensors) and bottom right the major axis of the error ellipsoid. In most cases the size and orientation of calculated location uncertainties (bottom right) corresponds well to the actual error vector (top left). 

For the scope of numerical applications, even in those cases where localization gives rise to definite symmetry properties (as in the diamond case), accurate symmetry equivalences are only reached under strong numerical conditions in the localization algorithm. This can be, in several cases, very computationally demanding and a more efficient shortcut to obtain accurate symmetry equivalence together with a good localization character is therefore desirable. 

Pdrez-Marm, D., Garrido-Varo, A., Guerrero-Ginel, J.E. and G6mez-Cabrera, A. Implementation of local algorithm with NIRS for compliance assurance in compound feedingstuffs. Applied Spectroscopy, 59 69-77 (2005). 

R.G Dambergs, D. Cozzolino, W.U. Cynkar, L. Janik, and M. Gishen, The determination of red grape quality parameters using the LOCAL algorithm. J. Near Infrared Spectrosc. 14(2), 71-79 (2006). 

Gfeller, B., 2009, Algorithmic Solutions for Transient Faults, in Communication Networks, On Swap Edges and Local Algorithms, Diss ETH Zurich. 

It must, however, be emphasized that all local algorithms such as MFC, DFD, and LB model compressible fluids, so that it takes time for the hydrodynamic interactions to propagate over longer distances. As a consequence, these methods become quite inefficient in the Stokes limit, where the Reynolds number approaches zero. Algorithms which incorporate an Oseen tensor do not share this shortcoming. 

In the literature, localization is approached using supervised and unsupervised communication paradigms. In the first case, a centralized supervisor (i.e., abase station) collects all the data coming from the robots and provides an estimate for the pose of the whole team. In the second approach, each robot runs a local algorithm to estimate its pose, using only its own sensors and shares data with its neighbors following the Mobile Ad hoc NETwork (MANET) model. 

Most approaches presented in the literature have been developed for homogenous teams in a probabilistic fiamewoik. To compute a good estimate, robots are assumed to be synchronized and to have the same sampling rate. Here we propose a probabilistic localization algorithm, which enables us to solve localization problems for a heterogeneous team. Time synchronization is not a requirement for such a procedure. 

This chapter deals with the design of a completely decentralized and distributed multi-robot localization algorithm. The localization procedure is based on nonlinear filtering. Specifically, it is an interlaced version of the EKF, able to produce accurate pose estimation for each robot of the team, thereby considerably reducing the computational load. 

---