Red-Black Trees

by Thomas A. Anastasio

A Red-Black Tree with NULLs shown Black-Height of the tree = 4

Red-Black Trees
Definition: A red-black tree is a binary search tree where:
Every node is either red or black. Each NULL pointer is considered to be a black node If a node is red, then both of its children are black. Every path from a node to a leaf contains the same number of black nodes.

Definition: The black-height of a node, n, in a red-black tree is the number of black nodes on any path to a leaf, not counting n.
3

A valid Red-Black Tree Black-Height = 2

Theorem 1 Any red-black tree with root x, has at least n = 2bh(x) 1 internal nodes, where bh(x) is the black height of node x. Proof: by induction on height of x.

Theorem 2 In a red-black tree, at least half the nodes on any path from the root to a leaf must be black.
Proof If there is a red node on the path, there must be a corresponding black node.

Theorem 3 In a red-black tree, no path from any node, N, to a leaf is more than twice as long as any other path from N to any other leaf.
Proof: By definition, every path from a node to any leaf contains the same number of black nodes. By Theorem 2, a least the nodes on any such path are black. Therefore, there can no more than twice as many nodes on any path from N to a leaf as on any other path. Therefore the length of every path is no more than twice as long as any other path
9

Theorem 4 A red-black tree with n internal nodes has height h <= 2 lg(n + 1). Proof: Let h be the height of the red-black tree with root x. By Theorem 2, bh(x) >= h/2 From Theorem 1, n >= 2bh(x) - 1 Therefore n >= 2 h/2 1 n + 1 >= 2h/2 lg(n + 1) >= h/2 2lg(n + 1) >= h
10

Bottom Up Insertion
Insert node as usual in BST Color the Node RED What Red-Black property may be violated?
Every node is Red or Black Leaf nodes are Black NULLS If node is Red, both children must be Black Every path from node to descendant leaf must contain the same number of Blacks
11

Bottom Up Insertion
Insert node; Color it RED; X is pointer to it Cases
0: X is the root -- color it black 1: Both parent and uncle are red -- color parent and uncle black, color grandparent red, point X to grandparent, check new situation 2 (zig-zag): Parent is red, but uncle is black. X and its parent are opposite type children -- color grandparent red, color X black, rotate left on parent, rotate right on grandparent 3 (zig-zig): Parent is red, but uncle is black. X and its parent are both left or both right children -- color parent black, color grandparent red, rotate right on grandparent
12

P
X

U
G P

Case 1 U is Red Just Recolor and move up

13

P
S X

U
X P G

Case 2 Zig-Zag Double Rotate X around P; X around G Recolor G and X

S U
14

P
X S

U
P X G

Case 3 Zig-Zig Single Rotate P around G Recolor P and G

S
15

Insert 4 into this R-B Tree 1

11

2
7 5 8

14
15

Black node

Red node
16

Insertion Practice
Insert the values 2, 1, 4, 5, 9, 3, 6, 7 into an initially empty Red-Black Tree

17

Asymptotic Cost of Insertion

O(lg n) to descend to insertion point O(1) to do insertion O(lg n) to ascend and readjust == worst case only for case 1 Total: O(log n)
18

Red-Black Trees
Bottom-Up Deletion

Recall ordinary BST Delete

1. If vertex to be deleted is a leaf, just delete it. 2. If vertex to be deleted has just one child, replace it with that child 3. If vertex to be deleted has two children, replace the value of by its in-order predecessors value then delete the inorder predecessor (a recursive step)

Bottom-Up Deletion
1. Do ordinary BST deletion. Eventually a case 1 or case 2 will be done (leaf or just one child). If deleted node, U, is a leaf, think of deletion as replacing with the NULL pointer, V. If U had one child, V, think of deletion as replacing U with V. 2. What can go wrong??

Which RB Property may be violated after deletion?

1. If U is red?
Not a problem no RB properties violated 2. If U is black? If U is not the root, deleting it will change the black-height along some path

Fixing the problem

Think of V as having an extra unit of blackness. This extra blackness must be absorbed into the tree (by a red node), or propagated up to the root and out of the tree. There are four cases our examples and rules assume that V is a left child. There are symmetric cases for V as a right child

Terminology
The node just deleted was U The node that replaces it is V, which has an extra unit of blackness The parent of V is P The sibling of V is S
Black Node Red Node Red or Black and dont care

Bottom-Up Deletion Case 1

Vs sibling, S, is Red
Rotate S around P and recolor S & P

NOT a terminal case One of the other cases will now apply All other cases apply when S is Black

Case 1 Diagram
P V+ S V+ S
Rotate

S P

Recolor

P
V+

Bottom-Up Deletion Case 2

Vs sibling, S, is black and has two black children.
Recolor S to be Red P absorbs Vs extra blackness
If P is Red, were done If P is Black, it now has extra blackness and problem has been propagated up the tree

Case 2 diagram
P
V+
Recolor and absorb

P+
S

Either extra black absorbed by P or

P now has extra blackness

Bottom-Up Deletion Case 3

S is black Ss RIGHT child is RED (Left child either color)
Rotate S around P Swap colors of S and P, and color Ss Right child Black

This is the terminal case were done

Case 3 diagrams
P V+ S V S
Rotate

P
V

Recolor

Bottom-Up Deletion Case 4

S is Black, Ss right child is Black and Ss left child is Red
Rotate Ss left child around S Swap color of S and Ss left child Now in case 3

Case 4 Diagrams
P V+ S P

P
Rotate

V+ S

V+

Recolor

65
50 10 70

80 60 90

62 Perform the following deletions, in the order specified Delete 90, Delete 80, Delete 70

Red Black Trees

Top-Down Insertion

Review of Bottom-Up Insertion

In B-Up insertion, ordinary BST insertion was used, followed by correction of the tree on the way back up to the root This is most easily done recursively
Insert winds up the recursion on the way down the tree to the insertion point Fixing the tree occurs as the recursion unwinds

Top-Down Insertion Strategy

In T-Down insertion, the corrections are done while traversing down the tree to the insertion point. When the actual insertion is done, no further corrections are needed, so no need to traverse back up the tree. So, T-Down insertion can be done iteratively which is generally faster

Goal of T-D Insertion

Insertion is always done as a leaf (as in ordinary BST insertion) Recall from the B-Up flow chart that if the uncle of a newly inserted node is black, we restore the RB tree properties by one or two local rotations and recoloring we do not need to make changes further up the tree

Goal (2)
Therefore, the goal of T-D insertion is to traverse from the root to the insertion point in such a way that RB properties are maintained, and at the insertion point, the uncle is Black. That way we may have to rotate and recolor, but not propagate back up the tree

Possible insertion configurations

X (Red or Black)

If a new node is inserted as a child of Y or Z, there is no problem since the new nodes parent is black

Possible insertion configurations

X

If new node is child of Z, no problem since Z is black.

If new node is child of Y, no problem since the new nodes uncle (Z) is black do a few rotations and recolor. done

Possible insertion configurations

X

If new node is inserted as child of Y or Z, its uncle will be red and we will have to go back up the tree. This is the only case we need to avoid.

Top-Down Traversal
X Y Z
As we traverse down the tree and encounter this case, we recolor and possible do some rotations.
There are 3 cases.

Remember the goal to create an insertion point at which the parent of the new node is Black, or the uncle of the new node is black.

Case 1 Xs Parent is Black

P P

X
Y Z Y

X
Z

Just recolor and continue down the tree

Case 2
Xs Parent is Red (so Grandparent is Black) and X and P are both left/right children
Rotate P around G Color P black Color G red

Note that Xs uncle, U, must be black because it (a) was initially black, or (b) would have been made black when we encountered G (which would have had two red children -- Xs Parent and Xs uncle)

Case 2 diagrams
G
P X Y Z S Y U

P X
Z S U G

Rotate P around G. Recolor X, Y, Z, P and G

Case 3
Xs Parent is Red (so Grandparent is Black) and X and P are opposite children
Rotate P around G Color P black Color G red

Again note that Xs uncle, U, must be black because it (a) was initially black, or (b) would have been made black when we encountered G (which would have had two red children -- Xs Parent and Xs uncle)

Case 3 Diagrams (1 of 2)
G
P S Y X Z S U P Y X Z

G
U

Step 1 recolor X, Y and Z. Rotate X around P.

Case 3 Diagrams (2 of 2)
G X P Z U P Z

X
G

Step 2 Rotate X around G. Recolor X and G

An exercise insert F
D
T

J E K

Top-Down Insert Summary

Case 1 P is Black Just Recolor X P Recolor X,Y,Z X P

Case 2 P is Red X & P both left/right P X Y Case 3 P is Red X and P are opposite children X Y

Recolor X,Y,Z X Z Y

G P

Rotate P around G Recolor P,G Y Z

P G X Z

G
P Recolor X,Y,Z Rotate X around P Y Z

G
X P Z Rotate X around G Recolor X, G P

X
G

Red Black Trees

Top-Down Deletion

Recall the rules for BST deletion

1. If vertex to be deleted is a leaf, just delete it. 2. If vertex to be deleted has just one child, replace it with that child 3. If vertex to be deleted has two children, replace the value of by its in-order predecessors value then delete the inorder predecessor (a recursive step)

What can go wrong?

1. If the delete node is red?
Not a problem no RB properties violated 2. If the deleted node is black? If the node is not the root, deleting it will change the black-height along some path

The goal of T-D Deletion

To delete a red leaf
How do we ensure thats what happens?
As we traverse the tree looking for the leaf to delete, we change every node we encounter to red. If this causes a violation of the RB properties, we fix it

Bottom-Up vs. Top-Down

Bottom-Up is recursive
BST deletion going down the tree (winding up the recursion) Fixing the RB properties coming back up the tree (unwinding the recursion)

Top-Down is iterative
Restructure the tree on the way down so we dont have to go back up

Terminology
Matching Weiss text section 12.2
X is the node being examined T is Xs sibling P is Xs (and Ts) parent R is Ts right child L is Ts left child

This discussion assumes X is the left child of P. As usual, there are left-right symmetric cases.

Basic Strategy
As we traverse the tree, we change every node we visit, X, to Red. When we change X to Red, we know
P is also Red (we just came from there) T is black (since P is Red, its children are Black)

Step 1 Examine the root

1. If both of the roots children are Black
a. Make the root Red b. Move X to the appropriate child of the root c. Proceed to step 2

2. Otherwise designate the root as X and proceed to step 2B.

Step 2 the main case

As we traverse down the tree, we continually encounter this situation until we reach the node to be deleted X is Black, P is Red, T is Black We are going to color X Red, then recolor other nodes and possibly do rotation(s) based on the color of Xs and Ts children 2A. X has 2 Black children 2B. X has at least one Red child

Case 2A X has two Black Children

P
2A1. T has 2 Black Children 2A2. Ts left child is Red 2A3. Ts right child is Red ** if both of Ts children are Red, we can do either 2A2 or 2A3

Case 2A1 X and T have 2 Black Children

P P

Just recolor X, P and T and move down the tree

Case 2A2
X has 2 Black Children and Ts Left Child is Red

Rotate L around T, then L around P Recolor X and P then continue down the tree

P
X T P X
L2 L1 L2

L
L1

Case 2A3
X has 2 Black Children and Ts Right Child is Red Rotate T around P Recolor X, P, T and R then continue down the tree

P X T R X L
R1 R2

T P
R1

R
R2

Case 2B X has at least one Red child

Continue down the tree to the next level If the new X is Red, continue down again If the new X is Black (T is Red, P is Black) Rotate T around P Recolor P and T Back to main case step 2

Case 2B Diagram
P
X T

Move down the tree. P P

If move to Black child (2B2) If move to the Red child (2B1) Rotate T around P; Recolor P and T Move down again Back to step 2, the main case

Step 3
Eventually, find the node to be deleted a leaf or a node with one non-null child that is a leaf.
Delete the appropriate node as a Red leaf

Step 4
Color the Root Black

Example 1 Delete 10 from this RB Tree

15 6 17

3
10 7

12
13

16
18

20 23

Step 1 Root has 2 Black children. Color Root Red Descend the tree, moving X to 6

Example 1 (contd)
15

X
3

17

12
10 7 13

16
18

20 23

One of Xs children is Red (case 2B). Descend down the tree, arriving at 12. Since the new X (12) is also Red (2B1), continue down the tree, arriving at 10.

Example 1 (contd)
15 6 17

3
10 7

12

16
13 18

20 23

Step 3 -Since 10 is the node to be deleted, replace its value with the value of its only child (7) and delete 7s red node

Example 1 (contd)
15 6 3 7 12 16 17 20

13

18

23

The final tree after 7 has replaced 10 and 7s red node deleted and (step 4) the root has been colored Black.

Example 2 Delete 10 from this RB Tree

15 6 17

3
2 4 10

12
13

16

20

Step 1 the root does not have 2 Black children. Color the root red, Set X = root and proceed to step 2

Example 2 (contd)
X
6 15 17

3
2 4 10

12
13

16

20

X has at least one Red child (case 2B). Proceed down the tree, arriving at 6. Since 6 is also Red (case 2B1), continue down the tree, arriving at 12.

Example 2 (contd)
15 6

17

3
2

T
4

X
10

12
13

16

20

X has 2 Black children. Xs sibling (3) also has 2 black children. Case 2A1 recolor X, P, and T and continue down the tree, arriving at 10.

Example 2 (contd)
15 6 3 2 4 17

P X
10

12

16 13

20

X is now the leaf to be deleted, but its Black, so back to step 2. X has 2 Black children and T has 2 Black children case 2A1 Recolor X, P and T. Step 3 -- Now delete 10 as a red leaf. Step 4 -- Recolor the root black

Example 2 Solution
15 6 3 2 4 12 13 16 17 20

Example 3 Delete 11 from this RB Tree

15 10

5
3 2 4 7 6 9 11

12
13

Valid and unaffected Right subtree

Step 1 root has 2 Black children. Color Root red. Set X to appropriate child of root (10)

Example 3 (contd)
15 10

X
12

5
3 2 4 7 6 9 11

13

X has one Red child (case 2B) Traverse down the tree, arriving at 12.

Example 3 (contd)
15 10

5
3 2 4

T
7 6 9

X
11

12
13

Since we arrived at a black node (case 2B2) assuring T is red and P is black), rotate T around P, recolor T and P Back to step 2

Example 3 (contd)
15 5

P
10

3 2 4

7 6

T
9 11

X
12 13

Now X is Black with Red parent and Black sibling. X and T both have 2 Black children (case 2A1) Just recolor X, P and T and continue traversal

Example 3 (contd)
15 5 10 3 2 4 7 6 9 11

P
12

13

Having traversed down the tree, we arrive at 11, the leaf to be deleted, but its Black, so back to step 2. X and T both have two Black children. Recolor X, P and T. Step 3 -- delete 11 as a red leaf. Step 4 -- Recolor the root black

Example 3 Solution
15 5 10 3 2 4 7 6 9 12 13