Concurrent Cloud Classifier

Making use of the output of a simple weather simulation to perform an analysis to determine the prevailing wind direction and the types of clouds that might result.

The data corresponds to a single layer of air as it evolves over time. This air layer at a particular time is represented as a regular two-dimensional matrix with each matrix entry consisting of three floating point values. The first two elements represent the x- and ycoordinate components of a wind vector (w) and the third element is a lift value (u) representing the rate at which air shifts upwards or downwards. These terms are sometimes also called advection (w) and convection (u). Each time step will have a new matrix with changing data. Given a set of rules, this data was then used to determine the clouds that might occur and the prevailing wind direction.

Concurrency

The Map Pattern was applied to assign the type of cloud that is likely to form in that location based on a comparison of the local average wind direction and uplift value. This was suitable because we were able to operate on each location independently without having to combine results. The Reduce Pattern was used to calculate the prevailing wind over all of the locations and time steps. This pattern was suitable because we needed to produce a single answer (the sum of all x advection values and y advection values) from an array via addition (an associative operator).

For more details on the implementation and results, see Report.pdf

A few things to note when running the scripts:

1. After running the make command, all the necessary compilations of source files occurs.

2. Do not call make run because it does not allow you to pass command-line arguments in. Rather run the program by calling java CloudDataThreaded <input_file> <output_file> from the bin directory. You will then see the run times for all of the tests in the terminal.

   • <input_file> - The data related to the cloud classifications and prevailing winds. See "simplesample_input.txt" for an example.
   • <output_file> - The true cloud classifications and prevailing winds. See "simplesample_output.txt" for an example.

3. Ensure the input file is in the bin directory.

4. I have included a directory called "Other Java" which contains two classes:

   • CloudClassification.java - was used to compare the produced output to the given output.
   • SampleGenerator.java - used to generate input data files.