Tatjana Petrov

ETH Zurich

Title: Approximate reductions of rule-base models

Abstract: In recent works, general algorithms for the exact reductions
of rule-based models were established. However, in the stochastic
setting, the reduced state space often remains combinatorially large.
This is because preserving exactness demands complete independence
between the dynamics over groups of proteins' domains (sites). We extend
this line of research by introducing  an error measure which allows us
to quantitatively study the effect of approximate reductions of
rule-based models, which can be seen as approximate probabilistic
bisimulation. The error is introduced over point-wise probability
distributions, and it is extended to the trace density semantics. We
perform the analysis on a rule-based formalism in terms of
site-graph-rewrite rules.