**I just announced the new *** Learn Spring * course, focused on the fundamentals of Spring 5 and Spring Boot 2:

*Learn Spring*course, focused on the fundamentals of Spring 5 and Spring Boot 2:

**>> CHECK OUT THE COURSE**

Last modified: July 25, 2019

In this quick tutorial, we’ll have a look at how to **compute the intersection between two Integer arrays ***‘a’* and* ‘b’*.

We’ll also focus on how to handle duplicate entries.

For the implementation, we’ll use *Streams.*

The intersection of two sets is by definition a set with all values from one, which are also part of the second set.

Therefore we need a *Function* or rather a *Predicate* to decide the membership in the second array. Since *List* provides such a method out of the box, we’ll transform this to a *List*:

Predicate isContainedInB = Arrays.asList(b)::contains;

To build up the resulting array, we’ll consider the elements of the first set sequentially and verify if they’re also contained in the second array.* *Then we’ll create a new array based on this.

The *Stream *API provides us with the needed methods. **First, we’ll create a Stream, then filter with the membership-Predicate and finally we’ll create a new array:**

public static Integer[] intersectionSimple(Integer[] a, Integer[] b){ return Stream.of(a) .filter(Arrays.asList(b)::contains) .toArray(Integer[]::new); }

Since arrays in Java are no *Set* implementation, we face the issue of duplicate entries in the input and then in the result. Notice that the number of occurrences in the result depends on the occurrences in the first parameter.

But for sets, elements must not occur multiple times. **We can archive this by using the distinct() method:**

public static Integer[] intersectionSet(Integer[] a, Integer[] b){ return Stream.of(a) .filter(Arrays.asList(b)::contain) .distinct() .toArray(Integer[]::new); }

So the length of the intersection no longer depends on the parameter order.

However, the intersection of an array with itself may not be the array again since we remove double entries.

A more general notion, which allows multiple equal entries, are multisets. For them, the intersection is then defined by the minimal number of input occurrences. So our membership-*Predicate* must keep score how often we add an element to the result.

The *remove()* method can be used for this, which returns the membership and consumes the elements. So after all equal elements in *‘b’* are consumed, no more equal elements are added to the result:

public static Integer[] intersectionSet(Integer[] a, Integer[] b){ return Stream.of(a) .filter(new LinkedList<>(Arrays.asList(b))::remove) .toArray(Integer[]::new); }

Since the *Arrays *API only returns an immutable *List,* we have to generate a dedicate mutable one.

In this article, we’ve seen how to use the *contains* and *remove *methods to implement an intersection for two arrays in Java.

All the implementation, code snippets, and tests can be found in our Github repository – this is a Maven-based project, so it should be easy to import and run as it is.

Isn’t contains for Lists an O(n) operation? So if array ‘a’ has n elements and array ‘b’ has m elements we are doing an O(m) operation n times. I believe this could be written as O(mn) operation. Wouldn’t it be better to index one of the arrays first in a hash set and then check the other one against that? I believe this is O(m+n) operation. (add and contains are O(1) for HashSets). The only time that this is worse than the O(mn) is if at least one of the n or m are equal to 1. Something like this:… Read more »

You are correct that more optimal solutions exist for this problem. This article offers a few different ways to solve the problem with Java Streams.

If this type of operation is a key performance bottleneck in your application, then extra care needs to be taken when choosing a solution.

Notice that this approach does not work for the Multiset solution.

For the other approaches we could simply use:

.filter(new HashSet(Arrays.asList(b))::contains)