有没有办法获取随机数生成器的状态? [英] Is there a way to obtain the state of the random number generator?

问题描述

假设我使用 srand(123) 播种 123，然后运行 ​​rand() X 次.稍后，我希望能够重新启动 Julia 并播种一个数字(或状态)，这样当我再次运行 rand() 时，我会得到如果我有种子 123 并运行会生成的数字rand() X + 1 次.有什么办法可以做到这一点，还是我真的必须运行 rand() X 次才能获得我想要的状态?

Say I seed 123 with srand(123), and run rand() X times. Later, I want to be able to restart Julia and seed a number (or state) such that when I run rand() again I get the number that would have been generated if I had seed 123 and run rand() X + 1 times. Is there any way I can do that, or do I really have to run rand() X times to obtain the state I want?

推荐答案

如果在 在julia中检索RNG种子对你来说是不可行的，我能想出的最好办法是复制全局随机数生成器的整个结构:

If the solution with custom random number generator presented in Retrieve RNG seed in julia is not feasible for you the best I can come up with is to copy the whole structure of global random number generator:

function reset_global_rng(rng_state)
    Base.Random.GLOBAL_RNG.seed = rng_state.seed
    Base.Random.GLOBAL_RNG.state = rng_state.state
    Base.Random.GLOBAL_RNG.vals = rng_state.vals
    Base.Random.GLOBAL_RNG.idx = rng_state.idx
end

rs = deepcopy(Base.Random.GLOBAL_RNG)
println(rand(5))
# [0.301558,0.602108,0.220952,0.0338732,0.553414]
reset_global_rng(rs)
println(rand(5))
# [0.301558,0.602108,0.220952,0.0338732,0.553414]

虽然我不是 100% 确定它是如何不与 random.jl 中的 dsfmt_gv_srand() 交互的.

although I am not 100% sure how it does not come into interaction with dsfmt_gv_srand() in random.jl.

这篇关于有没有办法获取随机数生成器的状态?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！