On global identifiability for arbitrary model parametrizations
Automatica (Journal of IFAC)
Numerical Data Fitting in Dynamical Systems: A Practical Introduction with Applications and Software
Numerical Data Fitting in Dynamical Systems: A Practical Introduction with Applications and Software
Hi-index | 0.00 |
We consider parameter estimation in ordinary differential equations (ODEs) from completely observed systems, and describe an improved version of our previously reported heuristic algorithm (IET Syst. Biol. , 2007). Basically, in that method, estimation based on decomposing the problem to simulation of one ODE, is followed by estimation based on simulation of all ODEs of the system. The main algorithmic improvement compared to the original version, is that we decompose not only to single ODEs, but also to arbitrary subsets of ODEs, as a complementary intermediate step. The subsets are selected based on an analysis of the interaction between the variables and possible common parameters. We evaluate our algorithm on a number of well-known hard test problems from the biological literature. The results show that our approach is more accurate and considerably faster compared to other reported methods on these problems. Additionally, we find that the algorithm scales favourably with problem size.