C-Sharp | Java | Python | Swift | GO | WPF | Ruby | Scala | F# | JavaScript | SQL | PHP | Angular | HTML
Some have spatial relationships on a two-dimensional plane, a grid. We use 2D arrays to represent this.
Initializers. The syntax of 2D arrays is often complex. Two integer indexes locate a point in a 2D array. And with initializers, we quickly create these arrays.
Array example. We introduce a two-dimensional array of width 4 and height 4—a little square. I assign some elements with the array at indexes (two are required, an X and a Y).
Loop: In the nested for-loops we iterate over each element. Each row contains a subarray.
And: We use the subarrays like a one-dimensional array containing int values. We have an array of equal-length arrays.
Based on: Java 7 Java program that uses 2D array public class Program { public static void main(String[] args) { // Create new 2-dimensional array. int[][] values = new int[4][4]; // Assign three elements within it. values[0][0] = 1; values[1][1] = 2; values[3][2] = 3; // Loop over top-level arrays. for (int i = 0; i < values.length; i++) { // Loop and display sub-arrays. int[] sub = values[i]; for (int x = 0; x < sub.length; x++) { System.out.print(sub[x] + " "); } System.out.println(); } } } Output 1 0 0 0 0 2 0 0 0 0 0 0 0 0 3 0
Array initializers. With these, we fill a newly-declared array with constant values. Here I create another little square, this time 2 by 2, with the values one through 4 in it.
Note: This syntax requires just one line. It is easier to type and review for humans.
Java program that uses 2D array initializer public class Program { public static void main(String[] args) { // Use initialize syntax. int[][] values = { { 1, 2 }, { 3, 4 } }; System.out.println(values[0][0]); System.out.println(values[1][0]); System.out.println(values[0][1]); System.out.println(values[1][1]); } } Output 1 3 2 4
Jagged array. This is like a 2D array but with irregular subarray lengths. Its lengths are not even—they are jagged. Here we create a jagged array of two String arrays.
And: The first subarray has two elements. The second one has three String elements. We loop over all these values.
Java program that uses jagged array public class Program { public static void main(String[] args) { // Create an array of String arrays: a jagged array. String[][] values = new String[2][]; // Fill first row with 2-element array. values[0] = new String[2]; values[0][0] = "cat"; values[0][1] = "dog"; // Use 3-element array for second row. values[1] = new String[3]; values[1][0] = "fish"; values[1][1] = "bird"; values[1][2] = "lizard"; // Display rows and elements. for (String[] array : values) { for (String element : array) { System.out.print(element); System.out.print(" "); } System.out.println(); } } } Output cat dog fish bird lizard
3D array. Usually two dimensions is enough, but we can use three dimensions in an array. And even more are supported. It is rare to need a 3D array.
Tip: Three-dimensional arrays can often be replaced with lesser-dimensioned arrays. This would likely be faster.
Java program that creates 3D array public class Program { public static void main(String[] args) { // Create space cube with 9 points. byte[][][] space = new byte[3][3][3]; space[0][0][0] = 10; space[1][1][1] = 20; // Middle of the cube. space[2][2][2] = 30; // Display points in our space-cube. System.out.println(space[0][0][0]); System.out.println(space[1][1][1]); System.out.println(space[2][2][2]); } } Output 10 20 30
Flattened 2D arrays. A compiler transforms a 2D array indexing operation into multiplication and addition. So to a compiler, a 2D array is similar to a 1D array with an indexing function.
Here: I develop a flattened 2D array, which uses a multiply and add to locate elements in a 1D array.
Indexes: We add the X coordinate to the Y coordinate multiplied by the number of rows. Two indexes are combined into one.
Java program that uses flattened 2D array public class Program { public static void main(String[] args) { // Create 4 by 4 one-dimension array. // Can access 0, 0 through 3, 3 with multiplication. int[] array = new int[4 * 4]; // Assign locations in the flattened array. array[0 + (1 * 4)] = 10; // 0, 1 array[3 + (3 * 4)] = 100; // 3, 3 array[2 + (2 * 4)] = 1000; // 2, 2 // Read those locations. System.out.println(array[0 + (1 * 4)]); System.out.println(array[3 + (3 * 4)]); System.out.println(array[2 + (2 * 4)]); // Read an empty location. System.out.println(array[3 + (2 * 4)]); // 3, 2 } } Output 10 100 1000 0
Flat array benchmark. Two-dimensional arrays often come with performance losses. Here I benchmark a flattened array (a 2D array stored as a 1D array) against a real 2D array.
Description: The benchmark sets up two 10 by 10 arrays. It then does a load and a store in the two arrays at two positions.
Result: The program shows the performance advantage of 1D arrays over 2D arrays. A flat array is nearly twice as fast.
Java program that benchmarks flat, 2D arrays public class Program { public static void main(String[] args) { int[] flat = new int[10 * 10]; // Flattened 10 x 10 array. flat[0 + (4 * 10)] = 1; int[][] two = new int[10][10]; // 10 x 10 two-dimensional array. two[0][4] = 1; long t1 = System.currentTimeMillis(); // ... Read, assign flattened array. for (int i = 0; i < 100000000; i++) { int value = flat[0 + (4 * 10)]; flat[9 + (9 * 10)] = value; } long t2 = System.currentTimeMillis(); // ... Read, assign 2D array. for (int i = 0; i < 100000000; i++) { int value = two[0][4]; two[9][9] = value; } long t3 = System.currentTimeMillis(); // ... Times. System.out.println(t2 - t1); System.out.println(t3 - t2); } } Output 131 ms, Flattened 1D array, load and store 256 ms, 2D array, load and store
Alternatives. We can use alternative data structures to replace a 2D array. A HashMap can store ArrayLists in its keys: we can look up collections by any index value.
A syntax form. 2D arrays use a convenient syntax form for storing data with 2D spatial relationships. We index a 2D array with a pair of coordinates.
Drawbacks. Performance suffers with 2D arrays—using 1D arrays may prove worthwhile. And often programs can be written in a clearer way with simpler collections.