Exercise: Matrix Arithmetic

Inside the directory you created for this course on PHP, create a new folder called Exercise-21-Matrix-Arithmetic and put the .php solution files for this exercise within it.

Let's start with creating the simpler of the two functions — add().

Recall that the function add() takes in two arguments $a and $b, both matrices. Now what we assert (assume) here is that $a and $b both have the same dimensions and, in this way, skip the part of checking whether both of them could be validly added together.

So step one is to determine ::m:: and ::n:: from the given matrices $a and $b.

How to do so?

::m:: is the number of rows of $a, which is the same as that of $b, and could simply be determined by inspecting the length of $a. Similarly, ::n:: is the number of columns of $a (as well as that of $b) and could be determined by inspecting the length of any element of $a (or $b).

In the code below, we define two variables $m and $n to hold the values of ::m:: and ::n::, respectively:

<?php

function add($a, $b) {
   $m = count($a);
   $n = count($a[0]);
}

Alright, let's now see what ought to be done up next.

Let's start with a question...

How many elements are there in ::\text{A} + \text{B}::?

See, ::\text{A} + \text{B}:: is an ::m \times n:: matrix, likewise it would trivially have ::m \times n:: elements.

For instance, a ::2 \times 2:: matrix would have 4 elements; a ::3 \times 1:: matrix would have 3 elements; a ::10 \times 25:: matrix would have 250 elements; and so on and so forth.

This means that our add() function would have to compute a total of $m * $n elements. For each corresponding set of rows in $a and $b, add the corresponding elements in both of them, and then put the sum in the final matrix.

We'll accomplish this using a set of nested for loops — the outer loop taking care of iterating over the given rows while the inner loop taking care of iterating over the given columns.

First and foremost, let's define a variable $matrix to represent the sum matrix, ultimately returned by add(). We'll populate this array in our set of nested loops:

<?php

function add($a, $b) {
   $m = count($a);
   $n = count($a[0]);
   $matrix = []; // The matrix representing $a + $b;
}

Let's now take a look over the loops:

<?php

function add($a, $b) {
   $m = count($a);
   $n = count($a[0]);
   $matrix = [];

   for ($i = 0; $i < $m; $i++) {
      $matrix[$i] = [];
      for ($j = 0; $j < $n; $j++) {
         $matrix[$i][$j] = $a[$i][$j] + $b[$i][$j];
      }
   }
}

Time to understand what's happening over here...

Inside the outer loop, precisely at line 9, we create a new row in the $matrix array where we'll eventually put the sums computed in the inner loop. If we don't perform this step, the assignment of $matrix[$i][$j] would fail as $matrix[$i] would get resolved as NULL!

In the inner loop, which goes over each column in $a[$i], we compute the sum $a[$i][$j] + $b[$i][$j], in line 11, and then add the result to the matrix[$i] array, as its j^th element.

In the end, $matrix is returned.

Simple?

This completes our add() function.

Let's now move on to the second function — multiply().

The initial setup would be the same as that for add(), except for that now there is a third value to consider, i.e. $p — the number of columns of the matrix $b:

<?php

function multiply($a, $b) {
   $m = count($a);
   $p = count($b);
   $n = count($b[0]);
   $matrix = [];
}

Moving on, as before, let's put forward a simple question.

How many elements are there in ::\text{AB}::?

Given that ::A:: and ::B:: are ::m \times p:: and ::p \times n:: matrices, respectively, ::\text{AB}:: then becomes an ::m \times n:: matrix.

This just means that in our function, we need to compute $m * $n elements.

As before, a set of nested loops ought to be set up — the outer one iterating $m times while the inner one iterating $n times.

The way we'll be populating the $matrix array, which represents the product of $a and $b here, would also be the same as in the function add(), i.e. using $matrix[$i] = [] in the outer loop to add a new row to $matrix, and then assigning to $matrix[$i][$j] to add a new item to this row.

Below, we implement this minimal loop setup:

<?php

function multiply($a, $b) {
   $m = count($a);
   $p = count($b);
   $n = count($b[0]);
   $matrix = [];

   for ($i = 0; $i < $m; $i++) {
      $matrix[$i] = [];
      for ($j = 0; $j < $n; $j++) {
         // Code to go here...
      }
   }
   return $matrix;
}

As long as the computation itself is concerned, it's a bit thought-provoking. Let's see what's there in it...

Inside the nested loop, we compute the ::(i, j)^{th}:: element of ::\text{AB}::. Recall the following statement from the exercise's description above:

This means that we need to iterate over the entire ::i^{th}:: row of $a, multiplying each element with the corresponding element in the ::j^{th}:: column of $b, and then sum all these products to obtain $matrix[$i][$j].

This hints us at two things:

1. We need another nested loop to compute these sums.
2. We need a variable to hold the sum of the products while we aggregate more values into it and then ultimately dump it into $matrix[$i][$j].

Since this nested loop needs to iterate over the entire $i^th row of $a, which has a total of $p items, it'll need to iterate a total of $p times as well (computing the product of corresponding elements in the $i^th row of $a and the $j^th column of $b).

As for the variable, we'll call it $sum.

Altogether, we get to the following:

<?php

function multiply($a, $b) {
   $m = count($a);
   $p = count($b);
   $n = count($b[0]);
   $matrix = [];

   for ($i = 0; $i < $m; $i++) {
      $matrix[$i] = [];
      for ($j = 0; $j < $n; $j++) {
         $sum = 0;
         for ($k = 0; $k < $p; $k++) {
            $sum += $a[$i][$k] * $b[$k][$j];
         }
         $matrix[$i][$j] = $sum;
      }
   }
   return $matrix;
}

And this completes our exercise.