Chapter 3
Regulation and Control
3.1
The Concept of Machine
“Machine” is used formally to describe the embodiment of a transformation (e.g.,
Eq. (3.1); cf. the automata in Sect. 12.1.1). In this formal sense, it does not have
any particular connotation of animate or inanimate. The essential feature is that the
internal state of the machine, together with the state of its surroundings, uniquely
defines the next state to which it will go. It is, therefore, a suitable abstraction of a
biological entity qua an information processor. A determinate machine is canonically
represented by a closed, single-valued transformation (3.1) and (3.2); a Markovian
machine is indeterminate insofar as the transitions are governed by a stochastic matrix
(e.g., (3.3)); the determinate machine is clearly a special case of the more general
Markovian machine.
If there are several possible transformations and a parameter governs which trans-
formation shall be applied to the internal states of the machine, then we can speak
of a machine with input, the input being the parameter. The machine with input is
therefore a transducer (cf. Sect. 7.3).
A Markovian machine with input would be represented by a set of stochastic
matrices together with a parameter to indicate which matrix is to be applied at any
particular step. If these parameters are themselves controlled by a stochastic matrix,
then we have a so-called hidden Markov model (Sect. 17.5.2).
3.2
Regulation
Regulation may be considered in abstract terms common to any mechanism, whether
living or not. The essential elements of a regulatory system are shown in Fig. 3.1.
The lines connecting the components indicate communication channels. The dotted
lines indicate the paths along which the regulator can receive information about the
disturbance. By way of illustration, consider the operation of a simple thermostatted
© Springer Nature Switzerland AG 2023
J. Ramsden, Bioinformatics, Computational Biology,
https://doi.org/10.1007/978-3-030-45607-8_3
19