Alternating machine acceptance in bounded space, as a decision problem #
EXPTIME's complete problem: an alternating Turing machine is data in an instance, and
does this machine accept its input, with the tape indexed by the positions and no bound whatever on the length of a play?
is DescriptiveComplexity.ATMAcceptSpace, an ordinary iso-invariant problem of
the catalog. It stands to DescriptiveComplexity.ATMAccept exactly as
DescriptiveComplexity.DTMAcceptSpace stands to
DescriptiveComplexity.DTMAccept: the step bound is dropped and the space stays
bounded by construction.
Two differences from the polynomial-hierarchy problem, both of them relaxations:
- the vocabulary is
FirstOrder.Language.turingAlt 2– two marks, one per player – and the promise folded into the yes-instances is only that they split the states (DescriptiveComplexity.ATMData.BlocksSplit), notDescriptiveComplexity.ATMData.BlocksWellFormed, whose ordering clause is what bounds the number of alternations. Here the alternation is unbounded, which is the whole point; - acceptance is
DescriptiveComplexity.ATMData.AltAcceptsSpace, the least fixed point of the game operator rather than a budgeted recursion.
Isomorphism-invariance is proved from the two agreements an isomorphism
supplies – one per direction – so that only the forward transport lemmas of
DescriptiveComplexity.MachinesAltSpace are needed and no finiteness is
assumed.
The problem #
Alternating acceptance in bounded space: the machine is well formed, its two marks split the states, and the existential player wins the game on the configuration graph from some initial configuration.
Equations
- One or more equations did not get rendered due to their size.