/export/starexec/sandbox/solver/bin/starexec_run_HigherOrder /export/starexec/sandbox/benchmark/theBenchmark.xml /export/starexec/sandbox/output/output_files -------------------------------------------------------------------------------- YES We consider the system theBenchmark. Alphabet: 0 : [] --> a eq : [a * a] --> c false : [] --> c fork : [b * a * b] --> b if : [c * c * c] --> c lt : [a * a] --> c member : [a * b] --> c null : [] --> b s : [a] --> a true : [] --> c Rules: lt(s(x), s(y)) => lt(x, y) lt(0, s(x)) => true lt(x, 0) => false eq(x, x) => true eq(s(x), 0) => false eq(0, s(x)) => false member(x, null) => false member(x, fork(y, z, u)) => if(lt(x, z), member(x, y), if(eq(x, z), true, member(x, u))) This AFS is converted to an AFSM simply by replacing all free variables by meta-variables (with arity 0). We use rule removal, following [Kop12, Theorem 2.23]. This gives the following requirements (possibly using Theorems 2.25 and 2.26 in [Kop12]): lt(s(X), s(Y)) >? lt(X, Y) lt(0, s(X)) >? true lt(X, 0) >? false eq(X, X) >? true eq(s(X), 0) >? false eq(0, s(X)) >? false member(X, null) >? false member(X, fork(Y, Z, U)) >? if(lt(X, Z), member(X, Y), if(eq(X, Z), true, member(X, U))) about to try horpo We use a recursive path ordering as defined in [Kop12, Chapter 5]. Argument functions: [[false]] = _|_ [[true]] = _|_ We choose Lex = {} and Mul = {0, eq, fork, if, lt, member, null, s}, and the following precedence: fork > null > s > eq = lt = member > 0 > if Taking the argument function into account, and fixing the greater / greater equal choices, the constraints can be denoted as follows: lt(s(X), s(Y)) >= lt(X, Y) lt(0, s(X)) >= _|_ lt(X, 0) >= _|_ eq(X, X) > _|_ eq(s(X), 0) >= _|_ eq(0, s(X)) >= _|_ member(X, null) >= _|_ member(X, fork(Y, Z, U)) > if(lt(X, Z), member(X, Y), if(eq(X, Z), _|_, member(X, U))) With these choices, we have: 1] lt(s(X), s(Y)) >= lt(X, Y) because [2], by (Star) 2] lt*(s(X), s(Y)) >= lt(X, Y) because lt in Mul, [3] and [6], by (Stat) 3] s(X) >= X because [4], by (Star) 4] s*(X) >= X because [5], by (Select) 5] X >= X by (Meta) 6] s(Y) > Y because [7], by definition 7] s*(Y) >= Y because [8], by (Select) 8] Y >= Y by (Meta) 9] lt(0, s(X)) >= _|_ by (Bot) 10] lt(X, 0) >= _|_ by (Bot) 11] eq(X, X) > _|_ because [12], by definition 12] eq*(X, X) >= _|_ by (Bot) 13] eq(s(X), 0) >= _|_ by (Bot) 14] eq(0, s(X)) >= _|_ by (Bot) 15] member(X, null) >= _|_ by (Bot) 16] member(X, fork(Y, Z, U)) > if(lt(X, Z), member(X, Y), if(eq(X, Z), _|_, member(X, U))) because [17], by definition 17] member*(X, fork(Y, Z, U)) >= if(lt(X, Z), member(X, Y), if(eq(X, Z), _|_, member(X, U))) because member > if, [18], [23] and [27], by (Copy) 18] member*(X, fork(Y, Z, U)) >= lt(X, Z) because member = lt, member in Mul, [19] and [20], by (Stat) 19] X >= X by (Meta) 20] fork(Y, Z, U) > Z because [21], by definition 21] fork*(Y, Z, U) >= Z because [22], by (Select) 22] Z >= Z by (Meta) 23] member*(X, fork(Y, Z, U)) >= member(X, Y) because member in Mul, [19] and [24], by (Stat) 24] fork(Y, Z, U) > Y because [25], by definition 25] fork*(Y, Z, U) >= Y because [26], by (Select) 26] Y >= Y by (Meta) 27] member*(X, fork(Y, Z, U)) >= if(eq(X, Z), _|_, member(X, U)) because member > if, [28], [29] and [30], by (Copy) 28] member*(X, fork(Y, Z, U)) >= eq(X, Z) because member = eq, member in Mul, [19] and [20], by (Stat) 29] member*(X, fork(Y, Z, U)) >= _|_ by (Bot) 30] member*(X, fork(Y, Z, U)) >= member(X, U) because member in Mul, [19] and [31], by (Stat) 31] fork(Y, Z, U) > U because [32], by definition 32] fork*(Y, Z, U) >= U because [33], by (Select) 33] U >= U by (Meta) We can thus remove the following rules: eq(X, X) => true member(X, fork(Y, Z, U)) => if(lt(X, Z), member(X, Y), if(eq(X, Z), true, member(X, U))) We use rule removal, following [Kop12, Theorem 2.23]. This gives the following requirements (possibly using Theorems 2.25 and 2.26 in [Kop12]): lt(s(X), s(Y)) >? lt(X, Y) lt(0, s(X)) >? true lt(X, 0) >? false eq(s(X), 0) >? false eq(0, s(X)) >? false member(X, null) >? false We orient these requirements with a polynomial interpretation in the natural numbers. The following interpretation satisfies the requirements: 0 = 3 eq = \y0y1.3 + 3y0 + 3y1 false = 0 lt = \y0y1.3 + y0 + y1 member = \y0y1.3 + y0 + 3y1 null = 3 s = \y0.3 + 3y0 true = 0 Using this interpretation, the requirements translate to: [[lt(s(_x0), s(_x1))]] = 9 + 3x0 + 3x1 > 3 + x0 + x1 = [[lt(_x0, _x1)]] [[lt(0, s(_x0))]] = 9 + 3x0 > 0 = [[true]] [[lt(_x0, 0)]] = 6 + x0 > 0 = [[false]] [[eq(s(_x0), 0)]] = 21 + 9x0 > 0 = [[false]] [[eq(0, s(_x0))]] = 21 + 9x0 > 0 = [[false]] [[member(_x0, null)]] = 12 + x0 > 0 = [[false]] We can thus remove the following rules: lt(s(X), s(Y)) => lt(X, Y) lt(0, s(X)) => true lt(X, 0) => false eq(s(X), 0) => false eq(0, s(X)) => false member(X, null) => false All rules were succesfully removed. Thus, termination of the original system has been reduced to termination of the beta-rule, which is well-known to hold. +++ Citations +++ [Kop12] C. Kop. Higher Order Termination. PhD Thesis, 2012.