forward computation 
Block diagram implementation of the timerecursive algorithm is described, computational complexity with respect to increasing block size is compared with straight forward computation, and the proper timerecursive block size is suggested.


This approach avoids forward computation on the complex geometry of irregular grids.


A forward computation algorithm continuously computes dynamic slices as statements are executed.


As far as we know, no experimental data has been published for forward computation algorithm.


As mentioned above, the forward computation algorithms maintain the latest dynamic slice for each variable/location.


Dynamic slicing is memory intensive no matter whether we use forward computation or backward computation.


For forward computation of UXO response, they use s=4e6 S/m and mur = 100.


Forward computation of data slices for this test case is shown in Table 1.


Forward computation of full slices for execution in case of Error 2 is shown in Table 2.


It is not necessary to apply the Sboxes during the forward computation.


In section 2, we briefly describe the existing forward computation algorithm and highlight its costs.


In contrast, the forward computation algorithm only provides the set of statements in the slice.


In Algorithm 2, we present the forward computation algorithm for finding full slice.


Note that the structure of this forward computation algorithm is similar to the forward computation algorithm for data slicing.


Next we describe the forward computation algorithm for updating relevant slices.


So, a total of sixteen activation values are presented at the time of forward computation.


To avoid large hardware overhead, the complexity of the inverse process should be much smaller than that of the forward computation.


Thus, the main calculation costs of a forward computation are to be expected in the equation solution procedure.


Therefore it is apparent that the cost of forward computation of dynamic slices is very high.


Therefore we decided to use forward computation method in this work.

