Ant tracking

We verified the accuracy of the predicted tracking error by comparing ex post daily tracking errors with ex ante daily tracking errors for the months of June and July 2002. The ex post exceeded the ex ante tracking error for 39.8% of the time, which is significantly higher than the expected 32% for a normal distribution. 

Fig. 1. A fault attribute is generated from seismic data. The ant tracking algorithm extracts surfaces from the attribute, and the human interpreter works with them in the structural editing tool. The surfaces are then exported as interpreted surfaces or written to a cube to produce an enhanced fault attribute.

Figure 3 and Figure 4 show some results from applying ant tracking to fault attributes. Note how noisy responses from chaotic layers in the attributes have been removed and how the fault surfaces have been made sharper and more continuous. 

Fig. 3. (a) Time slice of a fault attribute (variance) with (b) corresponding ant tracking results. 

D visualization is imperative to the analysis tool. Figure 5 shows an example of 3D visualization of extracted surfaces from a field offshore mid Norway. The horizontal slice is an enhanced attribute produced by the ant tracking, mapping all extracted surfaces, and a transparent seismic section. More about applying the fault interpretation workflow on this data can be found in [8]. 

EXHIBIT 25.8 Multifactor Model—Daily Returns versus ex ante Daily Tracking... 

In general, the adaptation mechanism is to keep track of the paths, which lead to the solution improving more often than others, and to increase their probabilities for the next generation of solutions. The scheme of the ant colony optimization algorithm allows to implement such an adaptation in a natural way. 

---