SUNY Geneseo Department of Computer Science

Experimentation

Previous Lecture

Misc

Hand out Lab 2

Beware of pseudocode vs Java in lectures

Questions?

Mini-assignments: be prepared to answer or ask questions when you come to class

Reading Summary

Sections 4.1 - 4.7

How to design experiments

Use experiments to compare theory to reality

Designing experiment

• Making prediction and hypothesis
• Choosing dependent and independent variables
• Choosing instruments with right accuracy and precision
• Using controls
• Vary independent variable
• Replicate experiment
• Minimize experimental error

Choose data analysis before starting experiment

Draw conclusion about whether experiment supported hypothesis, based on analysis

Controls and comparison to bio

• Typical hypothesis/experiment = "time to run this algorithm is proportional to f(n)"

Example Experiment

(Mini-assignment)

Recall the algorithm to draw a line n tiles long

    // In some subclass of Robot:
    drawLine( int n, Color c ) {
        for ( i = 0; i < n; i++ ) {
            this.paint( c )
            this.move()
        }
    }

We determined that this sends 2n robot messages to draw a line n tiles long

So actual implementations should run in time proportional to n, i.e., running times should be of the form

• time = c n

for some constant c

Do an experiment to test this idea

• Start with a Program that defines the algorithm as a Java method and provides a robot that can execute it.
• Add timing code to that program
• Run the program to get times for various values of n
• Note that program may (actually did) need to be modified to allow data to be collected consistently over a wide range of n values
• Some data and constants of proportionality

• We should now gather data for more n values, and gather additional measurements for the n values we already looked at.

Next

Logic and conditional statements

Read Sections 5.1 - 5.3

Next Lecture