Python Z3 和 concurrent.futures

Python Z3 and concurrent.futures

我想并行解决一组包含的问题,然后添加附加信息以解决新问题。

下面是用于解决问题的程序结构示例:

from z3 import *
import concurrent.futures


# solver test function
def add(a, b, solver, index):
    solver.append(a > b)
    assert solver.check()
    model = solver.model()
    return {
        'solver': solver,
        'av': model[a],
        'a': a,
        'b': b,
        'bv': model[b],
        'index': index
    }


with concurrent.futures.ThreadPoolExecutor(max_workers=5) as executor:
    # start solving the problems
    to_compute = []
    for i in range(3):
        sol = z3.Solver()
        to_compute.append(executor.submit(
            add,
            Int('a{}'.format(i)),
            Int('b{}'.format(i)),
            sol,
            i
        ))
    # wait for the solution to the computations
    next_to_solve = []
    for result_futures in concurrent.futures.as_completed(to_compute):
        results = result_futures.result()
        print(results)
        sol = results['solver']
        sol.append(results['a'] > results['av'])
        next_to_solve.append(
            executor.submit(
                add,
                results['a'],
                results['b'],
                sol,
                results['index']
            )
        )

每次程序的结果都不一样运行,结果包括:

我需要做什么才能使程序更可靠?

你看到这个例子了吗:http://github.com/Z3Prover/z3/blob/master/examples/python/parallel.py

我不是 z3py 中并发特性的专家,但似乎您需要非常小心地在您作为 运行 求解器的同一上下文中创建变量。该文件中有一些提示。