Observability, Reconstructibility and State Observers of Boolean Control Networks
read more
Citations
Optimal Control of Boolean Control Networks
Observability of Boolean networks: A graph-theoretic approach
On Pinning Controllability of Boolean Control Networks
On finite potential games
Identification of Boolean control networks
References
Metabolic stability and epigenesis in randomly constructed genetic nets
Positive Linear Systems: Theory and Applications
Combinatorial Matrix Theory
Related Papers (5)
Metabolic stability and epigenesis in randomly constructed genetic nets
Analysis and Control of Boolean Networks: A Semi-tensor Product Approach
Frequently Asked Questions (15)
Q2. What is the key ingredient of this observer?
It is worth noticing that the key ingredient of this observer is the state updating law, while the static output map, that provides the BCN state estimate, simply consists in selecting the first block of the observer state vector.
Q3. What is the reason why the output trajectories of two states are indistingu?
Since both state trajectories eventually become periodic of periods k` and kd, respectively, the corresponding output trajectories become periodic, too, and since condition ii) holds, the minimal periods of the outputs coincide with the minimal periods of the state trajectories.
Q4. What is the reason why the BN is a canonical vector?
The fact that all columns of L are canonical vectors implies that the set of all states of the BN can be partitioned into say r (disjoint) domains of attraction, each of them consisting of an elementary cycle (called equilibrium point, in case it consists of a single state) and a number of states that eventually converge to it.
Q5. What is the condition for the existence of a state observer?
a necessary condition for the existence of a state observer is that the BCN is reconstructible in some interval [0, T ].
Q6. What is the observability of a BN?
In the previous section the authors have seen that, as for linear statespace models, observability corresponds to the possibility of uniquely determining the system initial condition x(0) from the observation of the corresponding output evolution in some interval [0, T ].
Q7. What is the simplest way to describe the observer state?
in this case, the state observer could be described by means of a BCN, upon replacing each vector z(t) with the corresponding “canonical” representation ξ(t), thus getting a system of the following kindξ(t+ 1) = M n u(t) n y(t) n ξ(t), x̂(t) = Ĥξ(t),for suitable matrices11 M and Ĥ .
Q8. What is the arbitrary canonical vector in L4?
Ĥ takes the following form:Ĥ = [ δ44 δ 3 4 δ 4 4 δ 3 4 ∗ δ34 ∗ δ34 δ44 ∗ δ44 ∗ ∗ δ34 ∗ δ34 ] ,where the symbol ∗ denotes an arbitrary canonical vector in L4 (for instance δ14).
Q9. What is the simplest way to determine the observability of Boolean networks?
Φν ]; iii) if ` 6= d, the blocks H`c and Hdc are distinct and cannot be obtained one from the other by means of cyclic permutations of the columns, i.e. 6 ∃ a cyclic matrix C and h ∈ Z+ such that Hdc = H`c Ch.Proposition 3 provides a general characterization of observability for Boolean networks.
Q10. How do the authors prove that the smallest time instant T is able to identify the state ?
The authors want to prove that the smallest time instant T at which the authors are able to identify the state x(T ) from the output evolution y(t), t ∈ [0, T ], is just Tr + N̄ − 1 = 4.
Q11. What is the simplest way to prove that a BN is periodic?
every state trajectory of a BN takes only a finite number of distinct values and hence it eventually becomes periodic (possibly constant).
Q12. What is the state of the observer?
φNν−1,N nW.13estimation is represented by the static map Ĥ , that acts on the “canonical representation” z(t) of the observer state and realizes the map R.
Q13. what is the observability property of the h column?
Li1 the observability matrix in h steps corresponding to the input sequenceu(0) = δi0M , u(1) = δ i1 M , . . . ,u(h− 2) = δ ih−2 M .Clearly, the ith column of Ou,h provides the output sequence y(t), t ∈ [0, h−1], corresponding to the aforementioned input and the initial condition x(0) = δiN .
Q14. what is the smallest index in z(t)?
If not, they are simply neglected;(b) if the number dt of such distinct blocks in z(t) is smaller than the cardinality νt of the class {i ∈ [1, N ] : HδiN = y(t)}, then νt − dt blocks of the vector z(t + 1) are evaluated by making use of the remaining νt − dt canonical vectors δjN of {i ∈ [1, N ] : HδiN = y(t)}, through the formula zarr(t+1) = Ln u(t) n δjN ;(c) if νt < ν, the remaining blocks in z(t+1) are evaluated by making use of δiN , where i is the smallest index in {i ∈ [1, N ] : HδiN = y(t)}.
Q15. What is the value of the BCN state?
The blocks with lower index in the observer state are those that have proved to be compatible with a larger number of input/output samples, and for this reason once they prove to be compatible with the input and output for T +1 consecutive time instants, they provide the real value of the BCN state.