算法设计与分析课程

Chapter 1 Algorithm Analysis

• Complexity
  • time/space
  • Best/Worst/Average

Searching

• Linear
  • Best: $\Omega (1)$
  • Worst: $O(n)$
  • Space: $O(1)$
  • Average: $E(x)=\sum Pr(x)\cdot cost(x)={\sum }_{i=1}^n\frac{1}{n}\cdot i=\frac{1}{n}\sum i=\frac{1}{n}\cdot \frac{n(n+1)}{2}=\frac{n+1}{2}=O(n)$
• Binary（sorted array as input）
  • Best：$\Omega (1)$
  • Worst：$O(\log n)$
  • Space：$O(1)$
  • Average：$E(x)=\sum Pr(x)\cdot cost(x)={\sum }_{i=1}^{\log n}\frac{2^{i-1}}{n}\cdot i=\frac{1}{n}({\sum }_{i=1}^{\log n}{2}^{i-1}\cdot i)=\frac{1}{n}(2^{\log n}\cdot \log n-2^{\log n}+1)=O(\log n)$

Sorting

• Linear
  • Selection：every iteration, select the $i$th smallest number and locates $x$ at $i$th slot.
    • Best=Worst=Average=${\sum }_{i=1}^{n-1}i=\frac{n(n-1)}{2}=\Theta (n^2)$
    • Space：$O(1)$
  • Bubble：repeated swap the adjacent elements if they are in wrong order.
    • Best=Worst=Average=${\sum }_{i=1}^{n-1}i=\frac{n(n-1)}{2}=\Theta (n^2)$
    • Space：$O(1)$
  • Insertion：each time takes first element in the unsorted part and inserts it to the right place of the sorted one($\forall A[j]>x,A[j+1]\leftarrow A[j],j\leftarrow j-1$)
    • Best：$\Omega (n)$
    • Worst：${\sum }_{i=2}^n=\frac{n(n-1)}{2}=O(n^2)$
    • Space：$O(1)$
    • Average：
      • suppose $A[i]$ should be inserted at position $j$，$E(x)=\sum Pr(x)\cdot cost(x)=\frac{1}{i}\cdot (i-1)+{\sum }_{j=2}^i\frac{1}{i}\cdot (i-j+1)=\frac{i}{2}-\frac{1}{i}+\frac{1}{2}$，
      • ${\sum }_{i=2}^n(\frac{i}{2}-\frac{1}{i}+\frac{1}{2})=\frac{n^2}{4}+\frac{3n}{4}-{\sum }_{i=1}^n\frac{1}{i}=O(n^2)$
• Recursive
  • MergeSort：select the mid element, MergeSort(A[left, mid]), MergeSort(A[mid+1, right]), Merge(A, left, mid, right)
    • Best=Worst=Average=$\Theta (n\log n)$
    • Space：$O(n)$
  • QuickSort： Randomly Choose a $pivot$ and partition the array by smaller and larger halves, QucikSort(A[1, i-1]), QuickSort(A[i+1, n])
    • Best：$\Omega (n\log n)$
    • Worst：$O(n^2)$，退化到linear
    • Space：$O(\log n)$
    • Average：
      $T(n)=E_m[T(m-1)+T(n-m)]+cn$
      $=E_m[T(m-1)]+E_m[T(n-m)]+cn$
      $=$