Algorithms Notes (1) - Analysis of Algorithms

转载请注明出处：http://blog.csdn.net/cxsydjn/article/details/70464287

普林斯顿 《算法》第四版， Coursera公开课，by Prof. Robert Sedgewick。

The note focuses on analyzing the performance of algorithms with the scientific method.

The note corresponds to Chapter 1.4 of Algorithms, (@Princeton) and Week 1 of Algorithms, Part I, (@Coursera).

Analysis of Algorithms Introduction

Scientific method

Insight. [Knuth 1970s] Use scientific method to understand performance.

• Scientific method, which is a framework for predicting performance and comparing algorithms.
  
  • Observe some feature of the natural world.
  • Hypothesize a model that is consistent with the observations.
  • Predict events using the hypothesis.
  • Verify the predictions by making further observations.
  • Validate by repeating until the hypothesis and observations agree.
• Principles
  
  • Experiments must be reproducible.
  • Hypotheses must be falsifiable.

Experimental Algorithmics

• Run the program for various input sizes and measure running time.

• Data analysis

  • Standard plot. Plot running time T (N) vs. input size N.
  • Log-log plot. Plot running time T (N) vs. input size N using log-log scale.
• Regression. Fit straight line through data points

• Hypothesis. The running time is about XXX seconds

• Doubling hypothesis. Quick way to estimate b in a power-law relationship. Run program, doubling the size of the input.

Mathematical Models

Mathematical models for running time.

• Total running time: sum of cost × frequency for all operations.

  • Need to analyze program to determine set of operations.
  • Cost depends on machine, compiler.- Frequency depends on algorithm, input data.

• Simplification 1: cost model

  • Cost model. Use some basic operation as a proxy for running time.
  • cost model = array accesses (we assume compiler/JVM do not optimize array accesses away!)

• Simplification 2: tilde notation

  • Estimate running time (or memory) as a function of input size N. Ignore lower order terms.

    • when N is large, terms are negligible
    • when N is small, we don’t care

  • Technical definition.

$$f(N) \sim g(N) \Longleftrightarrow \lim\limits_{N \to \infty }{\frac{f(N)}{g(N)}} =1$$

Order-of-Growth Classifications

• Determine the order of growth of the running time of a program as a function of the input size.
• Formulate a hypothesis for the running time of a program as a function of the input size by performing computational experiments.

Binary Search

Binary search: Java implementation

Binary search: mathematical analysis

Binary search application: 3-SUM

Theory of Algorithms

• Define tilde and order-of-growth notations.

Memory

• Calculate the amount of memory that a Java program uses a function of the input size.