How could I add and update data in random order in a two-dimension matrix in Java

Question

for example the input is: (row, column, weight), such as (0,2,10.0),(1,2,3.0), (2,1,11.3), (1,2, 15.0)

the result should be:

0 1 2

0. . 10.0

1. . 15.0

2. 11.3.

(the data would be both inserted and updated )

I have noticed that it could not be done by simply using add(index i, elements e) or set() method provided by Arraylist in Java. Because it would throw :IndexOutOfBoundsException - if the index is out of range (index < 0 || index > size()) for instance, when I try to insert (0,2,5.0) after inserting (0,0,3.0).

Answer 1

The simplest thing to do with an ArrayList is to add null elements as needed to increase the size before trying to set an element at a given index.

private final List<List<Double>> matrix = new ArrayList<>();

public OptionalDouble set(int i, int j, double value) {
    while (i >= matrix.size()) matrix.add(new ArrayList<>());
    List<Double> row = matrix.get(i);
    while (j >= row.size()) row.add(null);
    Double old = row.set(j, value);
    return (old == null) ? OptionalDouble.empty() : OptionaDouble.of(old);
}

public OptionalDouble get(int i, int j) {
    List<Double> row = (i < matrix.size()) ? matrix.get(i) : Collections.emptyList();
    Double value = (j < row.size()) ? row.get(j) : null;
    return (value == null) ? OptionalDouble.empty() : OptionalDouble.of(value);
}

This is likely to perform quite poorly on large matrices, because of all the object creation performed here, and the poor locality of reference.

If possible, a contiguous array of primitive double elements will be better for large, dense data sets.