The situation with science is similar. Science is the activity of thought that creates functions. These functions, in contrast to concepts, are propositional (WP 117), and form the fragments from which science is able to piece together a kind of makeshift language, one which however, does not have any prior relation to truth, any more than philosophy does. Functions have meaning in creating a referential point of view, for Deleuze and Guattari, that is, in creating a basis from which things can be measured. As such, the first great functions are those such as absolute zero Kelvin, the speed of light etc., in relation to which a plane of reference is assumed. The plane of reference, again immanent to the functions that populate it, gains consistency through the strength and effectiveness of its functions. Also presupposed by science, in What is Philosophy?, are partial observers, the scientific counterpart of conceptual personae and artistic figures.
Beyond these similarities, Deleuze and Guattari distinguish between philosophy as the creation of concepts on a plane of immanence and science as the creation of functions on a plane of reference. Both relate to the virtual, the differential field of potential transformations of material systems, but in different ways. Philosophy gives consistency to the virtual, mapping the forces composing a system as pure potentials, what the system is capable of. Meanwhile, science gives it reference, determining the conditions by which systems behave the way they actually do. Philosophy is the “counter-effectuation of the event,” abstracting an event or change of pattern from bodies and states of affairs and thereby laying out the transformative potentials inherent in things, the roads not taken that coexist as compossibles or as inclusive disjunctions (differentiation, in the terms of Difference and Repetition), while science tracks the actualization of the virtual, explaining why this one road was chosen in a divergent series or exclusive disjunction (differenciation, according to Difference and Repetition). Functions predict the behavior of constituted systems, laying out their patterns and predicting change based on causal chains, while concepts “speak the event” (WP 21), mapping out the multiplicity structuring the possible patterns of behavior of a system—and the points at which the system can change its habits and develop new ones.
Function
Hjelmslev rightly signalizes the ambiguity that lies in the term "function" as it has hitherto been used in science, "where it designates both the dependence between two terminals and one or both of these terminals - the latter when the one terminal is said to be "a function of" the other" To avoid this ambiguity in modem terminology Hjelmslev introduces the technical term "functive" to denote the terminal of a function, reserving "function" for "the dependence between two terminals" only, no longer using it for those terminals themselves
L. Hjelmslev.