您的模型在这里是错误的:
TRANS
case
arr[x] <= arr[y] : output[x] = low & output[y] = big & next(y) = y + 1 & next(x) = x + 1;
...
esac;
第一行表示如果arr[x] <= arr[y]
为真,则到下一个状态的转换关系output[x] = low & output[y] = big & next(y) = y + 1 & next(x) = x + 1
由 定义。但是,后一个表达式在除第一个状态(值的幸运匹配)之外的所有状态中都计算为false,因此不可能传出转换到另一个状态。
此外,请注意您正在尝试更改数组定义的值,这是无法做到的。要查看这一点,请比较此模型交换变量数组的值
MODULE main()
VAR
arr: array 0..1 of {1,2};
ASSIGN
init(arr[0]) := 1;
init(arr[1]) := 2;
TRANS
next(arr[0]) = arr[1] &
next(arr[1]) = arr[0];
具有以下输出
nuXmv > reset; read_model -i swap.smv; go; pick_state -v ; simulate -v
Trace Description: Simulation Trace
Trace Type: Simulation
-> State: 1.1 <-
arr[0] = 1
arr[1] = 2
******** Simulation Starting From State 1.1 ********
Trace Description: Simulation Trace
Trace Type: Simulation
-> State: 1.1 <-
arr[0] = 1
arr[1] = 2
-> State: 1.2 <-
arr[0] = 2
arr[1] = 1
...
与这个其他模型交换定义数组的值
MODULE main()
DEFINE
arr := [1, 2];
TRANS
next(arr[0]) = arr[1] &
next(arr[1]) = arr[0];
这导致
nuXmv > reset; read_model -i swap_def.smv; go; pick_state -v ; simulate -v
Trace Description: Simulation Trace
Trace Type: Simulation
-> State: 1.1 <-
arr[0] = 1
arr[1] = 2
******** Simulation Starting From State 1.1 ********
No future state exists: trace not built.
Simulation stopped at step 1.
您当前的冒泡排序模型需要进行许多修复才能更正,因此我决定使用维基百科作为参考从头开始编写一个新模型。我编写的模型可以在nuXmv中运行,它是一个扩展NuSMV的工具,具有一些有趣的新功能。
编辑:我(稍微)修改了原始答案中的模型,以便与 NuSMV 完全兼容
-- Bubblesort Algorithm model
--
-- author: Patrick Trentin
--
MODULE main
VAR
arr : array 0..4 of 1..5;
i : 0..4;
swapped : boolean;
DEFINE
do_swap := (i < 4) & arr[ (i + 0) mod 5 ] > arr[ (i + 1) mod 5 ];
do_ignore := (i < 4) & arr[ (i + 0) mod 5 ] <= arr[ (i + 1) mod 5 ];
do_rewind := (i = 4) & swapped = TRUE;
end_state := (i = 4) & swapped = FALSE;
ASSIGN
init(arr[0]) := 4; init(arr[1]) := 1; init(arr[2]) := 3;
init(arr[3]) := 2; init(arr[4]) := 5;
init(i) := 0;
next(i) := case
end_state : i; -- end state
TRUE : (i + 1) mod 5;
esac;
init(swapped) := FALSE;
next(swapped) := case
(i < 4) & arr[(i+0) mod 5] > arr[(i+1) mod 5] : TRUE;
do_rewind : FALSE;
TRUE : swapped;
esac;
-- swap two consequent elements if they are not
-- correctly sorted relative to one another
TRANS
do_swap -> (
next(arr[ (i + 4) mod 5 ]) = arr[ (i + 1) mod 5 ] &
next(arr[ (i + 0) mod 5 ]) = arr[ (i + 0) mod 5 ] &
next(arr[ (i + 1) mod 5 ]) = arr[ (i + 2) mod 5 ] &
next(arr[ (i + 2) mod 5 ]) = arr[ (i + 3) mod 5 ] &
next(arr[ (i + 3) mod 5 ]) = arr[ (i + 4) mod 5 ]
);
-- perform no action if two consequent elements are already
-- sorted
TRANS
(do_ignore|do_rewind) -> (
next(arr[ (i + 4) mod 5 ]) = arr[ (i + 0) mod 5 ] &
next(arr[ (i + 0) mod 5 ]) = arr[ (i + 1) mod 5 ] &
next(arr[ (i + 1) mod 5 ]) = arr[ (i + 2) mod 5 ] &
next(arr[ (i + 2) mod 5 ]) = arr[ (i + 3) mod 5 ] &
next(arr[ (i + 3) mod 5 ]) = arr[ (i + 4) mod 5 ]
);
-- perform no action if algorithm has finished
TRANS
(end_state) -> (
next(arr[ (i + 0) mod 5 ]) = arr[ (i + 0) mod 5 ] &
next(arr[ (i + 1) mod 5 ]) = arr[ (i + 1) mod 5 ] &
next(arr[ (i + 2) mod 5 ]) = arr[ (i + 2) mod 5 ] &
next(arr[ (i + 3) mod 5 ]) = arr[ (i + 3) mod 5 ] &
next(arr[ (i + 4) mod 5 ]) = arr[ (i + 4) mod 5 ]
);
-- There exists no path in which the algorithm ends
LTLSPEC ! F end_state
-- There exists no path in which the algorithm ends
-- with a sorted array
LTLSPEC ! F G (arr[0] <= arr[1] &
arr[1] <= arr[2] &
arr[2] <= arr[3] &
arr[3] <= arr[4] &
end_state)
您可以在nuXmv上使用以下命令验证模型,这些命令依赖于底层的MathSAT5 SMT Solver来执行验证步骤:
~$ nuXmv -int
nuXmv> read_model -i bubblesort.smv
nuXmv> go_msat;
nuXmv> msat_check_ltlspec_bmc -k 15
或者你可以简单地使用NuSMV
~$ NuSMV -int
NuSMV> read_model -i bubblesort.smv
NuSMV> go;
NuSMV> check_property
求解器找到的解决方案如下:
-- specification !( F ( G ((((arr[0] <= arr[1] & arr[1] <= arr[2]) & arr[2] <= arr[3]) & arr[3] <= arr[4]) & end_state))) is false
-- as demonstrated by the following execution sequence
Trace Description: MSAT BMC counterexample
Trace Type: Counterexample
-> State: 2.1 <-
arr[0] = 4
arr[1] = 1
arr[2] = 3
arr[3] = 2
arr[4] = 5
i = 0
swapped = FALSE
end_state = FALSE
do_rewind = FALSE
do_ignore = FALSE
do_swap = TRUE
-> State: 2.2 <-
arr[0] = 1
arr[1] = 4
i = 1
swapped = TRUE
-> State: 2.3 <-
arr[1] = 3
arr[2] = 4
i = 2
-> State: 2.4 <-
arr[2] = 2
arr[3] = 4
i = 3
do_ignore = TRUE
do_swap = FALSE
-> State: 2.5 <-
i = 4
do_rewind = TRUE
do_ignore = FALSE
-> State: 2.6 <-
i = 0
swapped = FALSE
do_rewind = FALSE
do_ignore = TRUE
-> State: 2.7 <-
i = 1
do_ignore = FALSE
do_swap = TRUE
-> State: 2.8 <-
arr[1] = 2
arr[2] = 3
i = 2
swapped = TRUE
do_ignore = TRUE
do_swap = FALSE
-> State: 2.9 <-
i = 3
-> State: 2.10 <-
i = 4
do_rewind = TRUE
do_ignore = FALSE
-> State: 2.11 <-
i = 0
swapped = FALSE
do_rewind = FALSE
do_ignore = TRUE
-> State: 2.12 <-
i = 1
-> State: 2.13 <-
i = 2
-> State: 2.14 <-
i = 3
-- Loop starts here
-> State: 2.15 <-
i = 4
end_state = TRUE
do_ignore = FALSE
-> State: 2.16 <-
我希望您会发现我的回答对您有所帮助,尽管很晚;)。