next up previous
Next: The results of reduction Up: The reduction of chemical Previous: The chemical model

The method of reduction

The new method, developed by D. Semenov, allows us to select from the entire network only those species and reactions that are necessary to compute abundances of chosen (important) species with a reasonable accuracy. It differs from similar reduction technique, proposed by Ruffle et al. [7], by utilizing the possibility to reduce simultaneously the amount of species and reactions in the network. Probably, one may not be eager to reduce the amount of reactions in chemical networks since it doesn't speed up much the computations. However, it could be useful if one aims at the search of main destruction and formation pathways for the important species, where a smaller number of chemical reactions makes the analysis more easy.

Below we briefly outline the algorithm of our reduction technique. At first, one shall run chemistry with the full network in order to obtain abundances of all species within the network during the time evolution. At second, important species for which reduction will be made, must be specified. Using this information, weights of all species and all reactions are computed by the following iterative process:

The iterations are finished when all species and all time steps are passed. Then one can easily obtain a reduced network from the full network by choosing only those reactions, that have weights above a predefined cut-off value. The very applicability of this reduced network must be checked by comparing the calculated abundances of the important species with those obtained by the full network. If there is a significant difference in abundances of the important species computed with the full and reduced networks, then the cut-off must be readjusted to a new, smaller value, and last step must be repeated. This technique is discussed in more details in the paper by Wiebe et al. [14].
next up previous
Next: The results of reduction Up: The reduction of chemical Previous: The chemical model
Dimitri Semenov 2002-06-15