Code overview

This page describes, step by step, how the different components of Task 3 work together to generate input vectors, select an algorithm, compute the sum \(d = a\,x + y\), and write the result.

---

Generating Vectors

1. Invoke the executable

   From the terminal , use the run command associated with the generateVectors executable:

   run generateVectors <N> <filename_prefix>
   

   where

   • <N> is the number of elements in each vector and <filename_prefix> is the prefix for the output files.
   • <filename_prefix> is a path and file‐name prefix (for example, data/input/vector.txt).

2. What happens inside

   • The program allocates two std::vector<double> of length N.
   • It fills both vectors with a constant value (set in code).

   • It calls the write_vector helper, which:

     1. Examines the extension of <filename_prefix> (.txt, .dat, or .h5).
     2. Dispatches to the matching I/O routine (plain‐text, binary, or HDF5).
     3. Writes x to <filename_prefix>_x.<ext> and y to <filename_prefix>_y.<ext>.

3. Result

   • Two files are created: <filename_prefix>_x.<ext> and <filename_prefix>_y.<ext>, containing the vectors x and y respectively.

Preparing the configuration

Before running the sum, you must create or edit a YAML file (for example, config/config.yml) with these fields:

input:
  vector_x: "data/input/vector_N10_x.h5"
  vector_y: "data/input/vector_N10_y.h5"
N: 10
a: 3
output:
    path: "./data/output"
    prefix: "result"
    format: "txt" # or "dat" or "h5"
implementation: "gsl" # Or "default"

• input.vector_x, input.vector_y: paths to the two vectors.
• N: number of elements (must match how you generated the vectors).
• a: scalar multiplier in the sum.
• output.path: directory where the result will be saved.
• output.prefix: prefix for the output file.
• output.format: format for the output file (can be txt, dat, or h5).
• implementation: selects the algorithm (see next section).

Computing the sum

1. Invoke the executable

   From the terminal, use the run command associated with the vectorSum executable:

   run vectorSum config/config.yml
   

2. YAML parsing

   The program uses yaml-cpp to read all required parameters from config.yml.

3. Reading input vectors

   It calls the read_vector helper, which checks the file extension and reads the data into two std::vector<double> objects.

4. Choosing the implementation

   Based on the implementation field in the YAML config, it creates an instance of either VectorSumDefault or VectorSumGSL. Both derive from VectorSumInterface.

5. Computing the sum

   The chosen implementation's compute_sum method is called with the vectors x, y, scalar a, and an output vector d.

   • If implementation is default, it uses a simple loop to compute \(d = a \cdot x + y\).
   • If implementation is gsl, it uses gsl_vector_axpby for an in-place BLAS-optimized update.

6. Writing the result

   The program calls the write_vector helper to save the output vector d to the specified file in the chosen format.