4.4. Dynamic Array-Based Lists¶

The problem with a static array-based list is that it has a limited capacity. If we try to add new elements when the internal array is full, the method will throw an exception.

4.4.1. Resizing the internal array¶

How can we modify our class to allow for any number of elements? One solution is to create a larger internal array whenever the capacity is exceeded, and copy over all elements to the new one.

Java
Python

    private void resizeArray(int newCapacity) {
        if (newCapacity < MinCapacity) return;
        @SuppressWarnings("unchecked")
        E[] newArray = (E[]) new Object[newCapacity];
        for (int i = 0; i < listSize; i++)
            newArray[i] = internalArray[i];
        internalArray = newArray;
    }

    def _resizeArray(self, newCapacity):
        if newCapacity < self._minCapacity: return
        newArray = [None] * int(newCapacity)
        for i in range(self._listSize):
            newArray[i] = self._internalArray[i]
        self._internalArray = newArray

So, how large should the new internal array be? For now, let’s double the size of the internal array when we need to resize, which means that we add the following if-clause to the add method:

if listSize >= size of internalArray
    resizeArray(size of internalArray * 2)

That’s the only difference from the add method from StaticArrayList. So the dynamic add method will look like this.

Java
Python

    public void add(int i, E x) {
        if (!(0 <= i && i <= listSize)) throw new IndexOutOfBoundsException("list index out of range");
        if (listSize >= internalArray.length)
            resizeArray((int) (internalArray.length * CapacityMultiplier));
        listSize++;
        for (int k = listSize-1; k > i; k--)
            internalArray[k] = internalArray[k-1];
        internalArray[i] = x;
    }

    def add(self, i, x):
        if not (0 <= i <= self._listSize): raise IndexError("list index out of range")
        if self._listSize >= len(self._internalArray):
            self._resizeArray(len(self._internalArray) * self._capacityMultiplier)
        self._listSize += 1
        for k in reversed(range(i+1, self._listSize)):
            self._internalArray[k] = self._internalArray[k-1]
        self._internalArray[i] = x

As explained below, we don’t have to double the size, but we can multiply by 3 or 1.5 or 1.1. The important point is that we don’t add a constant number, but increase the size by a factor. This factor is the CapacityMultiplier in the code above.

Settings

Saving...

Server Error
Resubmit

4.4.2. How much to increase the array size¶

In the code above we doubled the size of the internal array whenever we needed to resize it. But we could have done something else, like:

Triple the size
Grow the size by 10%
Grow the size by 100 elements
Grow the size by 1 element

But which is best, and why?

There is a tradeoff: if we grow the array by a lot, we might waste memory. For example, immediately after we double the size, half of the array’s capacity is unused, so we use twice as much memory as needed. On the other hand, if we grow the array by a small amount, we need to resize it more often.

We will explore these tradeoffs by looking at the performance of the following small program under different resizing strategies:

list = new dynamic array list
for i in 1...n:
    list.add(i)

The program builds a list of length n by repeatedly calling \(add\). In this case, we could have used a static array-based list of capacity \(n\). So we would like the dynamic array-based list to have comparable performance to the static array-based list. This means that the program ought to take linear time.

4.4.2.1. Growing by a constant amount¶

What happens if we only grow the internal array by 1 element when we resize it?

if listSize >= size of internalArray
    resizeArray(size of internalArray + 1)

Every time we call add, the internal array will be resized. Resizing the array takes linear time, because if the internal array has size \(n\), it has to copy \(n\) elements from the internal array to the new array. To put it another way, the loop body newArray[i] = internalArray[i] will be executed \(n\) times.

Now suppose we run the program above to create a list of \(n\) elements. Adding up all the calls to resizeArray that happen, how many times does an array element get copied from the internal array to the new array (that is, how many times does the statement newArray[i] = internalArray[i] get executed)?

The array size is initially 1, so we get the following calls to resizeArray:

resizeArray(2), copying 1 element
resizeArray(3), copying 2 elements
resizeArray(4), copying 3 elements
…
resizeArray(n-2), copying \(n-3\) elements
resizeArray(n-1), copying \(n-2\) elements
resizeArray(n), copying \(n-1\) elements

In total, there are \(1+2+...+(n-1)\) element copy operations, which is equal to \(n(n-1)/2 = (n^2-n)/2\). This means that the program takes quadratic time, not linear!

Suppose for example that \(n = 1,000,000\). Using the formula above, the number of times an array element gets copied is \(999999 \times 1000000/2 = 499,999,500,000\). If copying one array element takes 1 ns, then the program spends nearly 10 minutes just resizing the array!

What happens if we instead grow the array by 100 elements every time? You can try the calculation yourself, for say \(n = 1,000,000\). What happens is that resizeArray gets called 100 times less often – so there 100 times fewer elements copied. But the runtime is still quadratic 1. When \(n = 1,000,000\), the total number of elements copied is about \(5,000,000,000\), still far too many.

In short, growing the array size by a constant amount is bad, because a loop that repeatedly adds to the array will take quadratic time.

4.4.2.2. Growing by a constant factor¶

One way to think about the problem is: as the array gets bigger, resizing it gets more expensive. So, to make up for that, when the array is bigger we need to grow it by more, so that we don’t have to resize as often. One way to do this is to always double the array size when it gets full. This turns out to work well!

Suppose that we run the example program with \(n = 1000\), i.e. we add 1000 elements to the list. As before, the internal array initially has a size of 1. What calls to resizeArray happen, and how many elements get copied each time?

resizeArray(2), copying 1 element
resizeArray(4), copying 2 elements
resizeArray(8), copying 4 elements
resizeArray(16), copying 8 elements
resizeArray(32), copying 16 elements
resizeArray(64), copying 32 elements
resizeArray(128), copying 64 elements
resizeArray(256), copying 128 elements
resizeArray(512), copying 256 elements
resizeArray(1024), copying 512 elements

You can see that the array gets resized a whole lot at the beginning – but as it gets bigger, it gets resized much less often. We can read off how many elements get copied: \(1+2+4+8+16+32+64+128+256+512 = 1023\).

Since the array starts from size 1 and always doubles, the array size is always a power of two. So to calculate the total number of elements copied, instead of adding up all the terms by hand, we can use the formula \(2^0+2^1+2^2+...+2^n = 2^{n+1}-1\) (with \(512=2^9\)).

Suppose that we now choose \(n=1,000,000\). How many elements get copied? In this case the final array size will be \(2^{20} = 1,048,576\). The array size will eventually grow from \(2^{18}\) to \(2^{19}\) to \(2^{20}\) elements, with the final call to resizeArray copying \(2^{19}\) elements. Using the formula above, the total number of elements copied is \(2^0+2^1+2^2+...+2^{19} = 2^{20}-1 = 1,048,575\).

Compared to when we grew the array by a fixed size of 1 element, this is \(500,000\) times fewer! So this in fact seems to be nice and efficient.

Let us now generalise to an arbitrary \(n\). The worst case is when the final call to add has to resize the array – that happens when \(n\) is one more than a power of two, \(n-1 = 2^k\). In that case, the final call to resizeArray grows the array from \(2^k\) to \(2^{k+1}\), copying \(2^k\) elements. The total number of elements copied is \(2^0+2^1+2^2+...+2^k = 2^{k+1} - 1 = 2 \cdot 2^k - 1 = 2(n-1) - 1 = 2n-3\). In fact, we have just proved the following result.

Theorem: When using the array-doubling strategy, calling add \(n\) times starting from an empty dynamic array list causes fewer than \(2n\) elements to be copied.

In short, the overhead of using a dynamic array list is at most two array elements copied per element that we add. But copying an array element is an extremely cheap operation, so dynamic array lists implemented using array doubling have almost no overhead, compared to static array lists. In particular, the complexity of our example program is linear, just as we wanted.

What happens if we instead grow the array by 50%? In fact, it still works out fine - the program takes linear time to run. To see this, you can use the same argument as above, but instead of using the formula \(2^0+2^1+...+2^k = 2^{k+1}\), you have to use the formula for a general geometric progression. What you get is an overhead of three elements copied per element added. In fact, Java ArrayLists grow the array by 50% on resizing.

In fact, growing the array by any constant factor works, because the same geometric progression reasoning applies. We can calculate the exact performance overhead of growing the array by any given factor:

Theorem: If we grow the array by a factor of \(k > 1\) when resizing it, then the overhead is at most \(1+1/(k-1)\) elements copied per add. For example, when growing by 20% (k = 1.2), the overhead is 6 elements copied per add.

In short, when resizing a dynamic array list, we should grow the array size by a constant factor. when adding many elements, this guarantees that we only have a constant factor of performance overhead due to occasional resizing. We can choose a large factor (such as 2) if we want fast performance, or a low factor (such as 1.2) if we want to save memory.

4.4.2.3. Constant amount vs constant factor¶

Here is a graph that shows just how big the performance difference is between the two resizing strategies: growing the array by a constant amount, and scaling it by a constant factor. The graph plots how many elements need to be copied, as a function of how many elements we add to the list.

Notice that although growing by 10000 seems pretty good at first, for largest lists it’s worse than growing by 10% (a factor of 1.1). We can see this more clearly if we zoom out the graph, making the x-axis go up to \(10,000,000\) instead of \(1,000,000\):

Though you can’t see it in the graph, at \(x=10,000,000\), growing by 10000 is 5000 times slower than growing by 10%! This is because the “growing by 10000” strategy takes quadratic time: if we do 10 times as many calls to add, it takes 100 times as long. Quadratic algorithms always lose to linear algorithms eventually!

4.4.3. Shrinking the internal array¶

We don’t have to change anything else in the code from StaticArrayList to have a working dynamic array list that has room for any number of elements.

But the problem is that if we first build a large list with 1000’s of elements, and then remove most of them, we will still have a large internal array where almost all cells are unused. So, let’s resize the array also when removing elements! When the array contains too many unused cells, we shrink it to half the size.

Now, it’s important that we don’t shrink the array when it’s half full. Why is that? Let’s consider the following sequence of additions and deletions:

append an element to the end
remove the last element
append another element to the end
remove it
append another one
remove it
…

If we’re unlucky and the initial list is full, then the first append will have to resize the array. Then when we remove that element, the list becomes less than half-full, and we have to resize again. Then the next append will resize, and the next remove will also resize. And so on… This will lead to a linear-time resize every time we append/remove, and so the final complexity will be linear (per operation). Which is not what we want.

How can we alleviate this? The solution is to wait even longer until we shrink the internal array! E.g., we can shrink the array (i.e., halve it), when it is only 1/3 full. So we can add the following lines to the end of the remove method:

if listSize <= size of internalArray * 1/3
    resizeArray(size of internalArray * 1/2)

That’s the only difference from the StaticArrayList.remove method.

Note that the factors 1/3 and 1/2 are not important, as explained before. The only thing that matters is that the minimum load factor (1/3) is smaller than the shrinking factor (1/2). So the dynamic remove method will look like this.

Java
Python

    public E remove(int i) {
        if (!(0 <= i && i < listSize)) throw new IndexOutOfBoundsException("list index out of range");
        E x = internalArray[i];
        for (int k = i+1; k < listSize; k++)
            internalArray[k-1] = internalArray[k];
        listSize--;
        internalArray[listSize] = null;   // For garbage collection
        if (listSize <= internalArray.length * MinLoadFactor)
            resizeArray((int) (internalArray.length / CapacityMultiplier));
        return x;
    }

    def remove(self, i):
        if not (0 <= i < self._listSize): raise IndexError("list index out of range")
        x = self._internalArray[i]
        for k in range(i+1, self._listSize):
            self._internalArray[k-1] = self._internalArray[k]
        self._listSize -= 1
        self._internalArray[self._listSize] = None   # For garbage collection
        if self._listSize <= len(self._internalArray) * self._minLoadFactor:
            self._resizeArray(len(self._internalArray) / self._capacityMultiplier)
        return x

Settings

Saving...

Server Error
Resubmit

4.4.4. Dynamic Array-based List: Full code¶

Finally, here is the full source code for the class DynamicArrayList. Note that now the constructor doesn’t take any capacity argument, since the internal array will automatically grow when needed.

In this example, we set the capacity multiplier to 1.5, meaning that we grow by 50% and shrink by 33% on every resize. The minimum load factor is set to 50% (which is smaller than 1/1.5 = 67%), and the minimum array capacity is 8. All these constants can be changed at will.

Java
Python

class DynamicArrayList<E> implements List<E> {
    private E[] internalArray;   // Internal array containing the list elements
    private int listSize;        // Size of list

    static int MinCapacity = 8;               // Minimum capacity of internalArray
    static double MinLoadFactor = 0.5;        // Must be smaller than 1/CapacityMultiplier
    static double CapacityMultiplier = 1.5;   // Factor to grow/shrink the capacity

    @SuppressWarnings("unchecked")
    public DynamicArrayList() {
        internalArray = (E[]) new Object[MinCapacity];
        listSize = 0;
    }

    public E get(int i) {
        if (!(0 <= i && i < listSize)) throw new IndexOutOfBoundsException("list index out of range");
        return internalArray[i];
    }

    public E set(int i, E x) {
        if (!(0 <= i && i < listSize)) throw new IndexOutOfBoundsException("list index out of range");
        E old = internalArray[i];
        internalArray[i] = x;
        return old;
    }

    public void add(int i, E x) {
        if (!(0 <= i && i <= listSize)) throw new IndexOutOfBoundsException("list index out of range");
        if (listSize >= internalArray.length)
            resizeArray((int) (internalArray.length * CapacityMultiplier));
        listSize++;
        for (int k = listSize-1; k > i; k--)
            internalArray[k] = internalArray[k-1];
        internalArray[i] = x;
    }

    public E remove(int i) {
        if (!(0 <= i && i < listSize)) throw new IndexOutOfBoundsException("list index out of range");
        E x = internalArray[i];
        for (int k = i+1; k < listSize; k++)
            internalArray[k-1] = internalArray[k];
        listSize--;
        internalArray[listSize] = null;   // For garbage collection
        if (listSize <= internalArray.length * MinLoadFactor)
            resizeArray((int) (internalArray.length / CapacityMultiplier));
        return x;
    }

    private void resizeArray(int newCapacity) {
        if (newCapacity < MinCapacity) return;
        @SuppressWarnings("unchecked")
        E[] newArray = (E[]) new Object[newCapacity];
        for (int i = 0; i < listSize; i++)
            newArray[i] = internalArray[i];
        internalArray = newArray;
    }

    public boolean isEmpty() {
        return listSize == 0;
    }

    public int size() {
        return listSize;
    }

    public Iterator<E> iterator() {
        return Arrays.stream(internalArray, 0, listSize).iterator();
    }
}

class DynamicArrayList(List):
    _minCapacity = 8            # Minimum capacity of internalArray
    _minLoadFactor = 0.5        # Must be smaller than 1/CapacityMultiplier
    _capacityMultiplier = 1.5   # Factor to grow/shrink the capacity

    def __init__(self):
        self._internalArray = [None] * self._minCapacity   # Internal array containing the list elements
        self._listSize = 0                                 # Size of list

    def get(self, i):
        if not (0 <= i < self._listSize): raise IndexError("list index out of range")
        return self._internalArray[i]

    def set(self, i, x):
        if not (0 <= i < self._listSize): raise IndexError("list index out of range")
        old = self._internalArray[i]
        self._internalArray[i] = x
        return old

    def add(self, i, x):
        if not (0 <= i <= self._listSize): raise IndexError("list index out of range")
        if self._listSize >= len(self._internalArray):
            self._resizeArray(len(self._internalArray) * self._capacityMultiplier)
        self._listSize += 1
        for k in reversed(range(i+1, self._listSize)):
            self._internalArray[k] = self._internalArray[k-1]
        self._internalArray[i] = x

    def remove(self, i):
        if not (0 <= i < self._listSize): raise IndexError("list index out of range")
        x = self._internalArray[i]
        for k in range(i+1, self._listSize):
            self._internalArray[k-1] = self._internalArray[k]
        self._listSize -= 1
        self._internalArray[self._listSize] = None   # For garbage collection
        if self._listSize <= len(self._internalArray) * self._minLoadFactor:
            self._resizeArray(len(self._internalArray) / self._capacityMultiplier)
        return x

    def _resizeArray(self, newCapacity):
        if newCapacity < self._minCapacity: return
        newArray = [None] * int(newCapacity)
        for i in range(self._listSize):
            newArray[i] = self._internalArray[i]
        self._internalArray = newArray

    def isEmpty(self):
        return self._listSize == 0

    def size(self):
        return self._listSize

    def __iter__(self):
        for i in range(self._listSize):
            yield self._internalArray[i]

1: You can get a precise number by using the formula for an arithmetic progression.

Data Structures and Algorithms

Chapter 4 Linear Structures