Temporal reasoning usually requires assumptions to be "boxed" where a formula F is boxed if it is equivalent with []F. The PM performs a syntactic approximation of when a formula is boxed, but this could be improved. In particular, a formula F is boxed if
- it is of the form
[]G, for some formula G,
- it is of the form
G /\ H where both G and H are boxed,
- it is of the form
\A x : G or \A x \in S : G if G is boxed (and S is a constant),
- it is of the form
<>G and G is boxed,
- it is of the form
WF_x(A) or SF_x(A).
The following is an example of a context that is not recognized as being boxed.
CONSTANT S
VARIABLE x
A(m,n) == TRUE
THEOREM
ASSUME \A m,n \in S : WF_x(A(m,n))
PROVE [](x=x)
<1>1. x = x
OBVIOUS
<1>. QED
BY <1>1, PTL \* fails
Temporal reasoning usually requires assumptions to be "boxed" where a formula
Fis boxed if it is equivalent with[]F. The PM performs a syntactic approximation of when a formula is boxed, but this could be improved. In particular, a formulaFis boxed if[]G, for some formulaG,G /\ Hwhere bothGandHare boxed,\A x : Gor\A x \in S : GifGis boxed (andSis a constant),<>GandGis boxed,WF_x(A)orSF_x(A).The following is an example of a context that is not recognized as being boxed.