问题1:卷首语是什么含义?它和随机算法 有什么关联? Randomized Algorithms “For him who seeks the truth, an error is nothing unknown." JOHANN WOLFGANG VON GOETHE
问题1: 卷首语是什么含义?它和随机算法 有什么关联?
确定性图灵机 一台图灵机是-个七元组(Q,∑,T,6,q0,Jaccept,9 reject)其中Q,∑,「都是有限集合,且满足 1. Q是状态集合; 2. 是输入字母表,其中不包含特殊的空白符口: 3.b∈为空白符 4.是带字母表,其中口∈且∑CT: 5. 0:Q×T→Q×T×{L,R}是转移函数,其中L,R表示读写头是向左移还是向右移; 6. q0∈Q是起始状态; 7. 9 accept∈Q是接受状态。9 reject∈Q是拒绝状态,且Areject卡9 accept 可以看出,转换函数决定了这么一个图灵机在每个格局(读写 头位置,当前带字母,当前状态)下,其转换新状态是唯一 的,计算是唯一的
确定性图灵机 可以看出,转换函数决定了这么一个图灵机在每个格局(读写 头位置,当前带字母,当前状态)下,其转换新状态是唯一 的,计算是唯一的
非确定图灵机 ·唯一的区别: 6:QXT→2Q×T×{L, 例如,设非确定型图灵机的当前状态为q,当前读写头所读的符号为x,若 6(9,x)={(q1,x1,d1),(q2,x2,d2),,(qk,xk,dk)} 则M将任意选择一个(q:,x:,d:),按其进行操作,然后进入下一步计算。 This means that a nondeterministic TM(algorithm)may have a lot of com- putations on an input x,while any deterministic TM (algorithm)has exactly one computation for every input
非确定图灵机 • 唯一的区别:
问题2:什么是随机算法 A randomized algorithm can be viewed as a nondeterministic algorithm that has a probability distribution for every noudeterministic choice. 问题3:这两句话是一致的吗? Another possibility is to consider a randomized algorithm as a deterministic algorithm with an additional in- put that consists of a sequence of random bits
问题3:这两句话是一致的吗? 问题2:什么是随机算法
问题4:你能根据这个描述画出随机算法的图灵机模 型的示意图吗? If one wants to formalize the notion of randomized algorithms,then one can take deterministic Turing machines A with an infinite additional tape.This additional tape is read-only;it contains an infinite random sequence of 0s and 1s,and A may move on it only from the left to the right.Another possibility 问题5:你能根据下个描述写出随机算法的图灵机模 型吗? Another possibility is to take a nondeterministic Turing machine with nondeterministic guesses over at most two possibilities and to assign the probability to every such possibility
问题4:你能根据这个描述画出随机算法的图灵机模 型的示意图吗? 问题5:你能根据下个描述写出随机算法的图灵机模 型吗?
随机算法的概率空间 ·(SAx,Prob): SA.x ={ClC is a computation of A on x; Prob is a distribution over Sax, Probax(C):it is 1/2 to the power of the number of random bits asked in C. Prob(A(x)=y),is the sum ofall Proba(C),where Coutputs y
随机算法的概率空间 • 𝑆𝐴,𝑥, 𝑃𝑟𝑜𝑏 : • 𝑆𝐴,𝑥 = 𝐶 𝐶 𝑖𝑠 𝑎 𝑐𝑜𝑚𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝐴 𝑜𝑛 𝑥 ; 𝑃𝑟𝑜𝑏 𝑖𝑠 𝑎 𝑑𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑜𝑣𝑒𝑟 𝑆𝐴,𝑥, • 𝑃𝑟𝑜𝑏𝐴,𝑥 𝐶 : 𝑖𝑡 𝑖𝑠 1 2 𝑡𝑜 𝑡ℎ𝑒 𝑝𝑜𝑤𝑒𝑟 𝑜𝑓 𝑡ℎ𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑟𝑎𝑛𝑑𝑜𝑚 𝑏𝑖𝑡𝑠 𝑎𝑠𝑘𝑒𝑑 𝑖𝑛 𝐶. • Prob(A(x) = y), is the sum of all ProbA,x (C), where C outputs y
如何评价随机算法:两个随机变量 算法的输出是随机变量 The probability that A outputs y for an input c,Prob(A()=y),is the sum of all Prb(C),where Coutputs y.Obviously,the aim of the randomized 算法的时间代价也是随机变量 Let Time(C)10 be the time complexity of the run C of A on x.Then the expected time complexity of A on x is Ezp-TimeA()=E(Time]=∑ProbA,z(C)·Tme(C)
如何评价随机算法:两个随机变量