The resource DTIME is used to define complexity classes, sets of all of the decision problems which can be solved using a certain amount of computation time.
Many important complexity classes are defined in terms of DTIME, containing all of the problems that can be solved in a certain amount of deterministic time.
In general, we desire our complexity classes to be robust against changes in the computational model, and to be closed under composition of subroutines.
The well-known complexity class P comprises all of the problems which can be solved in a polynomial amount of DTIME.
For robust classes, such as P, the exact machine model used to define DTIME can vary without affecting the power of the resource.
The Computational Complexity literature often defines DTIME based on multitape Turing machines, particularly when discussing very small time classes.
[1] Due to the Linear speedup theorem for Turing machines, multiplicative constants in the time bound do not affect the extent of DTIME classes; a constant multiplicative speedup can always be obtained by increasing the number of states in the finite state control and the size of the tape alphabet.
In the statement of Papadimitriou,[2] for a language L, Using a model other than a deterministic Turing machine, there are various generalizations and restrictions of DTIME.