International Journal of Computer Applications (0975 – 8887)

Volume 57– No.11, November 2012

Iterative Method for Recreating a Binary Tree

from its Traversals
Nitin Arora Pradeep Kumar Kaushik Satendra Kumar
Assistant Professor Assistant Professor Assistant Professor
Dept. of Computer Sci. & Dept. of Computer Sci. & Dept. of Computer Sci. &
Engg. Engg. Engg.
Uttarakhand Technical Uttarakhand Technical Uttarakhand Technical
University University University
Women Institute of Technology Shivalik College of Engineering Seemant Institute of
(WIT), Dehradun (SCE), Dehradun Technology (SIT), Pithoragarh

ABSTRACT
Many reconstruction algorithms for binary tree have been
discussed in this paper. A particular focus of this paper is on
“A new Non-Recursive Algorithm for Reconstructing a Binary
Tree from its Traversals”. The computation time required for
executing the reconstruction algorithm are O(N) and space
complexity is O(NlogN) where N is the number of nodes in
the binary tree. This algorithm works well for most of the
input cases, but it has some drawbacks. There are some
sequences of pre-order and in-order for which no legitimate
tree can be constructed but the algorithm didn’t take these
cases into consideration and constructed a wrong tree for
these cases.
In this paper, we have proposed a solution to the problem in
the previous algorithm and designed an algorithm which is the
modified version of the previous algorithm for generating a
correct binary tree. The new modified algorithm is
implemented in C language and tested in GCC Compiler in
Linux, for all types of input cases. The New modified
algorithm works well for all types of input cases. We have
calculated the best case time complexity of modified
algorithm and show that a correct tree can be reported in O(N) Figure 1: Binary Tree containing 9 nodes
time in best case and O(NlogN) space where N is the number
of nodes in the tree. We have discussed some applications of Commonly there are three traversing methods: in-order, pre-
the new modified algorithm in Huffman Coding, compiler order and post-order traversal. In an in-order traversal, first the
design, text processing and searching algorithms. left child is processed recursively, and then process the current
node followed by the right child. The output of an in-order
Key words traversal algorithm of the binary tree shown in figure 1 is: B,
Non-recursive; tree traversals; binary tree. D, A, I, G, E, H, C, F and that of the pre-order is: A, B, D, C,
E, G, I. H, F.
1. INTRODUCTION
It is well known that given the in-order traverse of a binary
A tree is a fundamental structure in computer science. Almost
tree, along with one of its pre-order or post-order traversals,
all operating systems store files in trees or tree like structures.
the original binary tree can be uniquely identified. It is not
It is well known that given the in-order traverse of a binary
difficult to write a recursive algorithm to reconstruct the
tree, along with one of its pre-order or post-order traversals,
binary tree. Most text books and reference books present the
the original binary tree can be uniquely identified. It is not
recursive and non-recursive algorithms for traversing a binary
difficult to write a recursive algorithm to reconstruct the
tree in in-order, post-order and pre-order. The computation
binary tree. Most text books and reference books present the
time required is O(N2) where N is the number of nodes in the
recursive [1, 2, 3, 4, 5] and non-recursive algorithms [6, 7, 8,
tree. Many iterative methods for reconstructing a binary tree
9, 10] for traversing a binary tree in in-order, post-order and
has been proposed till date for which computation time
pre-order.
required is O(N). In this paper one of the proposed algorithms
Many iterative methods [11, 12, 13, 14, 15] for reconstructing
has been examined. The proposed algorithm works well for
a binary tree from its traversals have been proposed till date
most of the input cases, but it has a drawback. There are some
for which computation time required is O(N) where N is the
sequences of pre-order and in-order traversals for which no
number of nodes in the tree.
legitimate tree can be constructed but the proposed algorithm
A binary tree is different recursively as either empty or
didn’t take this case into consideration and constructed a
consists of a root, a left tree and a right tree. The left and
wrong tree for these cases. We have modified this algorithm
right trees may themselves be empty, thus a node with one
so that if for an in-order and pre-order sequence a correct tree
child could have a left or right child.

6
 International Journal of Computer Applications (0975 – 8887)
Volume 57– No.11, November 2012

cannot be drawn then it is reported in O(N) best case time and

O(NlogN) space.
This paper is organized as follows: Section 2 includes the
related reconstruction algorithm; Section 3 introduces our
modified algorithm and its complexity analysis, performance
and result comparison in Section 4 and conclusion and future
aspects with applications of our algorithm are discussed in
section 5.

2. BACKGROUND AND RELATED

WORK
In computer science, a data structure is a particular way of
storing and organizing data in a computer so that it can be Figure 2(a): Reconstruction of a binary tree.
used efficiently. Different kinds of data structures are suited to
different kinds of applications, and some are highly The next data in the pre-order list is 2 so the root node of the
specialized to specific tasks. For example, B-trees are left sub-tree is 2. Hence the data before 2 in the in-order list
particularly well-suited for implementation of databases, (i.e. 4, 7) forms the left sub-tree of the node that contains a
while compiler implementations usually use hash tables to value 2. The data that comes to the right of 2 in the in-order
look up identifiers [2, 16]. list (i.e. 8, 5) forms the right sub-tree of the node with value 2.
Data structures are used in almost every program or software Figure 2(b) shows structure of tree after expanding the left
system. Data structures provide a means to manage huge and right sub-tree of the node that contains a value 2.
amounts of data efficiently, such as large databases and
internet indexing services. Usually, efficient data structures
are a key to designing efficient algorithms. Some formal
design methods and programming languages emphasize data
structures, rather than algorithms, as the key organizing factor
in software design [2, 16].

2.1 Related reconstruction algorithm

It is well known that given the in-order traverse of a binary
tree, along with one of its pre-order or post-order traversals,
the original binary tree can be uniquely identified. Most
textbooks and reference books present the traversal
algorithm as a recursive algorithm [1, 2, 3, 4, 5, 8, 10]. It is
not difficult to write a recursive algorithm to reconstruct the
binary tree. The computation time required is O(N2) where N
is the number of nodes in the tree.
Figure 2(b): Reconstruction of a binary tree (contd.)
2.1.1 Recursive Algorithm
Now the next data in pre-order list is 4, so the root node of the
If we know the sequences of nodes obtained through in- left sub-tree of the node that contains a value 2 is 4. The data
order/pre-order/post-order traversal it may not be feasible to before 4 in the in-order list forms the left sub-tree of the node
reconstruct the binary tree. This is because two different that contains a value 4. But as there is no data present before 4
binary trees may yield same sequence of nodes when in in-order list, the left sub-tree of the node with value 4 is
traversed using post-order traversal. Similarly in-order or pre- empty. The data that comes to the right of 4 in the in-order list
order traversal of different binary trees may yield the same (i.e. 7) forms the right sub-tree of the node that contains a
sequence of nodes. However, a unique binary tree can be value 4. Figure 2(c) shows structure of tree after expanding
constructed if the result of in-order and pre-order traversal are the left and right sub-tree of the node that contains a value 4.
available. The recursive algorithm for reconstruct a binary tree
from its traversals is as follows:

In-order traversal: 4, 7, 2, 8, 5, 1, 6, 9, 3
Pre-order traversal: 1, 2, 4, 7, 5, 8, 3, 6, 9

The first value in the pre-order traversal gives us the root of

the binary tree. So the node with data 1 becomes the root of
the binary tree. In in-order traversal, initially the left sub-tree
is traversed then the root node and then the right sub-tree. So
the data before 1 in the in-order list (i.e. 4, 7, 2, 8, 5) forms
the left sub-tree and the data after 1 is the in-order list (i.e. 6,
9, 3) forms the right sub-tree. Figure 2(a) shows the structure
of the tree after separating the tree in left and right sub-trees.
Figure 2(c): Reconstruction of a binary tree (contd.)

7
 International Journal of Computer Applications (0975 – 8887)
Volume 57– No.11, November 2012

We are now left with only one value 7 in both the pre-order
and in-order form we simply represent it with a node as
shown in figure 2(d).

Figure 2(g): Reconstruction of a binary tree (contd.)

Figure 2(d): Reconstruction of a binary tree (contd.)

Figure 2(h): Reconstruction of a binary tree (contd.)

Figure 2(e): Reconstruction of a binary tree (contd.)

Figure 2(i): Reconstruction of a binary tree (contd.)

In the same way one by one all the data are picked from the
Figure 2(f): Reconstruction of a binary tree (contd.) pre-order list and are placed and their respective sub-trees are
constructed and the whole tree is constructed. Figure 2(e) to
2(i) shows each step of construction of nodes one by one.

8
 International Journal of Computer Applications (0975 – 8887)
Volume 57– No.11, November 2012

2.1.2 Parallel Algorithms L1[i] = i+1;

L2[i] = I+IP[i]-S[i]-1;
V. Kamakoti and C. Pandu Rangan [17] presents an R1[i] = if(L1[i] ≤ L2[i]) then
optimal algorithm for reconstructing a binary tree. They L2[i]+1
presents a parallel EREW PRAM algorithm using O(n/log log else
n) processors to reduce the reconstruction process to parallel i+1
merging. Their algorithms runs in O(log n) time. R2 [i] = if(P [i] is the right child of P[j]) then
R2[j]
Stephan Olariu and Michael Overstreet [18] present else
reconstructing algorithm with the best parallel merging L2[j]
algorithms, their algorithm can be implemented in O(log log This algorithm is meant for only binary trees implemented
n) time using O(n/log log n) processors on CREW PRAM or using arrays.
in O(log n) time using O(n/log n) processors in EREW The computations of L1[i], R1[i], and R2[i] are
PRAM. straightforward; the computation of L2[i] is based on the
following lemma [13].
2.1.3 Non-Recursive Algorithms Lemma 3: The number of left descendants of P[i] equals
IP[i]-S[i]-1.
H. A. Burgdorff [12] presented a non-recursive algorithm for The nodes that precede P[i] in the in-order sequence fall into
reconstructing a binary tree from its in-order and pre-order two classes. The left descendants of P[i] belong to the first
sequences (in short i-p sequences) and their algorithm takes class. The others belong to the second class. Each node in the
O(N2) computation time. first class follows P[i] in the pre-order sequence; each node in
the second class precedes P[i] in the pre-order sequence.
G. H. Chen [13] has also proposed a non-recursive algorithm Thus, the number of left descendants of P[i] equals the
for reconstructing a binary tree from its traversals from its i-p number of nodes preceding P[i] in the in-order sequence
sequence array of time complexity O(N) and inefficient space. minus the number of nodes preceding P[i] in both the in-order
and the pre-order sequences, which is IP[i]-S[i]-1.
In the algorithm proposed by G H Chen [13] the non-recursive In the following, G. H. Chen presents the reconstructing
algorithm for reconstructing the original binary tree from its algorithm.
in-order and pre-order traversal is done in two stages [13].
First, the i-p sequence is constructed from the in-order and L1[1]:=2; L2[1]:=IP[1];
pre-order traversals. Then the original binary tree is R1[1]:=if L1≤L2[1] then L2[1] +1 else 2;
reconstructed from the i-p sequence array. Let I[1...n] and R2[1]:= N; S[1]:=0;
P[1…n] be the two sequences that represents the in-order and if L1[1]≤L2[1]
pre-order traversal respectively of the given binary tree with n then P[L1[1]] is the left child of P[1];
nodes. Also, let IP[1…N] represent the corresponding i-p if R1[1]≤R2[1]
sequence. In fact IP[i] = I-1 [P[i]], 1≤ i ≤N. then P[R1[1]] is the right child of P[1];
G H Chen has proposed a non-recursive algorithm by for i:= 2 to N do
proving following two lemmas [13]. begin {Assume P[j] is the parent of P[i]
Lemma 1: Let P[i] be a non-leaf node and s (s > 0) be the S[i]:= if P[i] is the left child of P[j] then S[j] else
number of its descendants. Then, IP[i] > IP[j] for i < j <k IP[j];
and IP[i] < IP[j] for k ≤ j ≤ i+s iff P[i+1],…P[K-1] are the L1[i]:= i+1;
left descendants of P[i] and P[k],…P[i+s] are the right L2[i]:= i+IP[i] –S[i] -1;
descendants of P[i]. Moreover, P[i + 1] is the left child of P[i] R1[i]:= if L1[i]≤L2[i] then L2[i] + 1 else i+1;
and P[k] is the right child of P[i]. R2[i]:= if P[i] is the right child of P[j] then R2[j]
else L2[j];
According to Lemma 1, the binary tree can be reconstructed if L1[i]≤L2[i] then P[L1[i]] is the left child of
by sequentially examining array P. For each non leaf node P[i];
P[i], if s is known, then left and right child can find (if they if L1[i]≤L2[i] then P[L1[i]] is the left child of
exist) as stated by lemma 1. Let S[i], 1 ≤ i ≤ n, denote the P[i];
cardinality of the set { j | j < i and IP[j] < IP[i] }. S[i] if R1[i]≤R2[i] then P[R1[i]] is the right child of
is the number of nodes that precede P[i] in both the in-order P[i];
and the pre-order sequences. Arrays S will be used in the end
reconstruction algorithm and can be computed according to The computation time and space required for executing the
the following lemma, with S[1] = 0 initially. reconstructing algorithm are O(N).
Lemma 2: Let P[j] be the parent of P[i]. If P[i] is the left child Vinu V Das [11] presented a non-recursive algorithm for
of P[j], then S[i] = S[j]. If P[i] is the right child of P[j], reconstructing a binary tree from its in-order and pre-order
then S[i] = IP[j]. traversals and their algorithm takes O(N) computation time
Let [L1[i]……..L2[i]] be the range of array P where the left and space complexity is O(NlogN).
descendants of P[i] are located and [R1[i]….R2[i]] be the
range of array P where the right descendants of P[i] are The proposed non-recursive algorithm by Vinu V Das [11] is
located. If L1[i] > L2[i] (R1[i] > R2[i]), then P[i] has no left to reconstruct the binary tree, implemented using linked list,
(right) descendants. Initially, L1[1]=2, L2[i]=IP[1], and from its in-order and pre-order traversals [11]. Let P[1….N]
R2[1]=n. Also, R1[1]=L2[1] + 1, if L1[1] ≤ L2[i], R1[1]=2 and I[1….N] be the pre-order and in-order traversal of the
otherwise. Let P[j] (j<1) be the parent of P[i], L1[i], L2[i], given binary tree with N nodes. From this sequence a binary
R1[i] and R2[i] are computed as follows: tree will be reconstructed using linked list and root node

9
 International Journal of Computer Applications (0975 – 8887)
Volume 57– No.11, November 2012

address will be returned for any further use. This non- if(root == NULL)
recursive algorithm works on the following lemma [11]: root = NewNode
while (I[count2] is present in the LL)
Lemma: Let P[j] be the parent of P[i], (where j < i) then Lp[j] {
be the corresponding index in the in-order sequence of node PresentNode= address of
P[j] and Lp[i] be the corresponding index in the in-order I[count2];
sequence of node P[i]. If Lp[j] is greater than Lp[i] then P[i] is remove the I[count2] from the
the left child of P[j] otherwise it is the right child. The binary LL;
tree can be reconstructed by exploiting the following three right = 1;Left=0;count2++;
properties of the above lemma [11]. }
1. First node in the pre-order sequence will be the root of if( data != I[count2] )
the binary tree. {
2. The node(s) which comes to the left of root node in the Add the “data” into the linear list LL
in-order sequence will be the nodes in the left sub tree If( Left == 1)
and node(s) in the right are the nodes in the right sub tree Place the NewNode as a left child of
of the reconstructed root node. PresentNode
3. Apply the second property recursively to the left and else if(Right == 1)
right sub tree of the root node to obtain the complete Place the NewNode as a right child of
binary tree. PresentNode
For example in Figure 3, the first node of the pre-order Left=1; Right=0;
sequence is A, which is a root node of the reconstructed tree. }
The nodes B and D are the only two nodes to the left of A in else
in-order sequence, and they are the nodes in the left sub tree {
of the root node A. If( Left == 1)
Place the NewNode as a left child of
PresentNode
else if(Right == 1)
Place the NewNode as a right child of
PresentNode
Left=0; Right=1; count2++;
}
data = P[count1];
PresentNode=NewNode;
}
The computation time required for executing the
reconstruction algorithm are O(N) and space complexity is
O(NlogN) [11].

3. OUR MODIFIED NON-RECURSIVE

ALGORITHM
Figure 3: Binary Tree
The algorithm proposed by Vinu V Das [11] is implemented
in C language and checked for some different input
The new Non-Recursive Algorithm for Reconstructing a
sequences. The algorithm works correctly for some
Binary Tree from its Traversals presented by Vinu V Das is
sequences but on the other side it generated a wrong binary
based on the following [11]:
tree for some other input sequences. We made some
modification in the algorithm and proposed the solution. The
The left and right are the two variables and any of them will
proposed solution is based on the following checks:
be set to one when a new node needs to be placed to the left or
1. First construct the tree and if the height of the tree is less
right of the present node. If left = 1 and right = 0 then the new
than N, where N is the number of nodes then the tree is
node will be left child of the present node. Let LL be a linear
always correct.
linked list where all visited nodes are stored pre-order and
2. Otherwise, if the height of the tree is equal to n, then
unvisited nodes in in-order sequence. Whenever a node is
there is a chance that the constructed tree may be wrong.
visited in in-order it will be removed from the LL.
3. In this case where is the height of the tree is equal to the
PresentNode is a pointer variable that holds the address of the
number of nodes, the tree is traversed in an in-order
present node.
fashion and it is compared with the given in-order
Following is the reconstruction algorithm [11]:
sequence.
4. If the sequence matches then the tree formed is correct.
Left = 0; Right = 0; count1=0; count2=0; data=P[0];
5. Otherwise, the in-order and pre-order sequences cannot
root=NULL;
combine to form a legitimate tree.
for(count1=0;count1<n;count1++)
The left and right are the two variables and any of them will
{
be set to one when a new node needs to be placed to the left or
Create a NewNode
right of the present node. If left = 1 and right = 0 then the new
NewNode->info = data
node will be left child of the present node. Let LL be a linear
NewNode->lChild = NULL
linked list where all visited nodes are stored pre-order and
NewNode->rChild = NULL

10
 International Journal of Computer Applications (0975 – 8887)
Volume 57– No.11, November 2012

unvisited nodes in in-order sequence. Whenever a node is The best case time complexity of our modified algorithm
visited in in-order it will be removed from the LL. is O(N) where N is the number of nodes.
PresentNode is a pointer variable that holds the address of the
present node. 3.1 Complexity analysis
An extra variable m is used in our algorithm which keeps a By the performance analysis of a program, we mean the
count of the number of nodes having two children. amount of computer memory and time needed to run a
program.
Following is the modified Non-Recursive algorithm for
generating a tree from its in-order and pre-order traversals: On analysis of the algorithm we find out the recurrence
relation that can be formed is
/*Variable Declarations*/ (3.1)
left=0; right=0; count1=0; count2=0; data=P[0]; root=NULL;
m=0; /* New variable in our program which keeps a count of We can solve this recurrence relation by iterative method as
the number of nodes having two children */ follows:
for ( count1 = 0 ; count1<n ; count1++ )
{
/* create a new node*/ On adding the subsequent terms of (3.1), we get
newnode->info = data;
newnode->lchild = NULL;
newnode->rchild = NULL;
If (root == NULL)
… … …
root=newnode;
… … …
/*Intially the linked list is empty*/
while ( I[count2] is present in the linked list )
Summing up all these terms we get
{
____________________________________
presentnode = address of I[count2];
remove the I[count2] from the linked list;
right=1;left=0;count2++;
/*Here a small modification is done*/
m++;
}
(3.2)
If( data != I[count2] )
{
Add the data into the linked list; From the Sterling approximation [1, 19]
If (left == 1) (3.3)
Place the newnode as left child of
presentnode Taking log at base e on both the sides of (3.3)
If( right == 1)
Place the newnode as right child of (3.4)
presentnode
left=1;right=0;
From (3.2) and (3.4) we get,
}
else
{
If( left == 1)
So the time complexity of the algorithm is of order of
Place the newnode as a left child of
.
presentnode
elseif( right == 1)
For the best case scenario the loop runs for n times
Place the newnode as right child of
only, it does not have to search the link list for a parent
presentnode
address. Hence the time complexity will be n.
left=0;right=1;count2++;
}
data=P[count1+1];
presentnode = newnode;
} /*for loop end*/
… … …
… … …
If (m is greater than zero or check if the in-order traversal of
the tree formed matches with the given in-order sequence)
______________________________
{
Print the tree;
}
else
{ Hence is of the order of .
Print that a legitimate tree cannot be constructed with the
sequences;
}

11
 International Journal of Computer Applications (0975 – 8887)
Volume 57– No.11, November 2012

4. PERFORMANCE ANALYSIS AND

RESULTS COMPARISON
The new Non-Recursive Algorithm for Reconstructing a
Binary Tree from its Traversals presented by Vinu V Das [11]
is implemented in C language and test for some input in-order
and pre-order sequences. The algorithm works well for some
input cases but it has a drawback. This algorithm generates
wrong binary tree for some another input sequences. We have
modified this algorithm and generate a new Modified non-
recursive reconstructing algorithm. The new modified
algorithm is implemented in C language and test for some
input in-order and pre-order sequences. Our new modified
algorithm works well for all types of input cases. Fig. 5: The generated tree for input sequence Pre-order: 1, 2,
4, 8, 9, 5, 3, 6, 7 and In-order: 7, 3, 6, 1, 5, 2, 9, 4, 8 by the
algorithm presented by Vinu V Das
4.1 Performance comparison
The tree generated in figure 5 for the input sequence Pre-order
Both the algorithms presented by Vinu V Das [11] and our traversal: 1, 2, 4, 8, 9, 5, 3, 6, 7 and In-order traversal: 7, 3, 6,
new modified Non-Recursive Algorithm for Reconstructing a 1, 5, 2, 9, 4, 8 is an incorrect binary tree because we cannot
Binary Tree from its Traversals is implemented in C language generate the same pre-order and in-order traversals as
and test for different input cases. supplied.

4.1.1 Sample Input and Output

1. Supplied input sequence with total 9 nodes

Pre-order traversal: 1, 2, 4, 8, 9, 5, 3, 6, 7
In-order traversal: 8, 4, 9, 2, 5, 1, 6, 3, 7
Both the algorithm generated the same output tree shown in
the figure 4

Fig. 6: The generated output for input sequence Pre-order: 1,

2, 4, 8, 9, 5, 3, 6, 7 and In-order: 7, 3, 6, 1, 5, 2, 9, 4, 8 by our
new modified the algorithm

The output message that the tree cannot be constructed in

figure 6 for the input sequence Pre-order traversal: 1, 2, 4,
8, 9, 5, 3, 6, 7 and In-order traversal: 7, 3, 6, 1, 5, 2,
9, 4, 8 is generated.

Fig. 4: The generated tree for input sequence Pre-order: 1, 2, Some other inputs supplied to our modified algorithm.
4, 8, 9, 5, 3, 6, 7 and In-order: 8, 4, 9, 2, 5, 1, 6, 3, 7
3. For the input
The tree generated in figure 4 for the input sequence Pre-order
traversal: 1, 2, 4, 8, 9, 5, 3, 6, 7 and In-order traversal: 8, 4, 9, Pre-order Traversal: 1, 2, 3, 4, 5, 6
2, 5, 1, 6, 3, 7 is the correct binary tree because we can In-order Traversal: 1, 2, 3, 4, 5, 6
generate the same pre-order and in-order traversals as The correct generated binary tree is shown in figure 7.
supplied.

2. For the input sequence

Pre-order traversal: 1, 2, 4, 8, 9, 5, 3, 6, 7
In-order traversal: 7, 3, 6, 1, 5, 2, 9, 4, 8
The output tree generated by the algorithm presented by Vinu
V Das [11] is shown in figure 5 and the output generated by
our algorithm is shown in figure 4.3.

Fig. 7: The generated output for input sequence Pre-order: 1,

2, 3, 4, 5, 6 and In-order: 1, 2, 3, 4, 5, 6 by our new
modified the algorithm

12
 International Journal of Computer Applications (0975 – 8887)
Volume 57– No.11, November 2012

4. For the input sequence compression may go wrong. The modified algorithm
prevents any wrong tree from forming.
Pre-order Traversal: 1, 2, 3, 4, 5, 6, 7
In-order Traversal: 7, 6, 5, 4, 3, 2, 1 6. REFERENCES
The correct generated binary tree is shown in figure 8. [1] Vinu V Das, “Principles of Data Structures Using C and
C++”, New Age International Publishers, Reading,
Mass., 2005.
[2] M. Weiss, “Data Structures & Problem Solving Using
Java”, 2nd ed., Addison Wesley, 2002.
[3] D. E. Knuth, “The Art of Computer Programming”, Vol.
3 (2nd ed.): Sorting and Searching, Addison Wesley,
1998.
[4] R. Sedgewick, “Algorithms in Java”, 3d edition,
Addison Wesley, 2003.
[5] D. E. Kunth, “The Art of Computer Programming”, Vol.
1: Fundamental Algorithm, Addison-Wesley, Reading,
Fig. 8: The generated output for input sequence Pre-order: 1,
Mass., 1973.
2, 3, 4, 5, 6, 7 and In-order: 7, 6, 5, 4, 3, 2, 1 by our
new modified the algorithm [6] Mark Allen Weiss, “Data Structures and Algorithm
Analysis in C”, Vol. 3 (2nd ed.), Addison-Wesley, 1997.
5. CONCLUSION AND FUTURE WORK [7] Yedidyah Langsam, Moshe J. Augenstein, Aaron M.
A tree is a fundamental structure in computer science. Almost Tanenbaum: “Data Structures using C and C++”, 2nd
all operating systems store files in trees or tree like structures. Ed., Prentice-Hall India, July 2002.
It is well known that given the in-order traverse of a binary [8] Sartaj Sahni: “Data Structures, Algorithms and
tree, along with one of its pre-order or post-order traversals, Applications in JAVA”, 2nd Ed., University Press.
the original binary tree can be uniquely identified. It is not [9] A. Anderson and S. Carlsson: “Construction of a tree
difficult to write a recursive algorithm to reconstruct the from its traversals in optimal time and space”,
binary tree. Most text books and reference books present the Information Processing Letters, 34:21-25, 1990.
recursive and non-recursive algorithms for traversing a binary
tree in in-order, post-order and pre-order. The computation [10] N. Gabrani and P. Shankar: “A note on the reconstruction
time required is O(N2) where N is the number of nodes in the of a binary tree from its traversals”, Information
tree. Many iterative methods for reconstructing a binary tree Processing Letters, 42:117-119, 1992.
has been proposed till date for which computation time [11] Vinu V Das, “A new Non-Recursive Algorithm for
required is O(N). The Non- Recursive algorithm presented by Reconstructing a Binary Tree from its Traversals”, IEEE
Vinu V Das [11] is considered in this paper. This algorithm Comm., pp. 261-263, 2010.
works well for most of the input in-order and pre-order [12] H. A. Burgdorff, S. Jojodia, F.N. Springsteel, and
traversals, but generates wrong binary tree for some another Y.Zalcstein, “Alternative Methods for the Reconstruction
in-order and pre-order sequences. In this paper a new modified of Tree from their Traversals”, BIT, Vol. 27, No. 2, p.
Non Recursive algorithm has been proposed which works well 134, 1987.
for all the input cases.
[13] G. H. Chen, M. S. Yu, and L.T. Liu: “Non-recursive
Also in the future better algorithm can be found out than
Algorithms for Reconstructing a Binary Tree from its
the above one which would further reduce the complexity of
Traversals”, IEEE Comm., Vol. 88, pp. 490-492, 1988.
the algorithm. The algorithm can be implemented using
suitable simulation tools. [14] C. C. Lee, D. T. Lee, C. K. Wong: “Generating Binary
Trees of bounded height”, Acta Inf., 23, 529-544, 1986.
5.1 Applications [15] Seymour Lipschutz: “Theory and problem of Data
Structures”, International Edition, McGRAW-HILL,
The tree is fundamental structure in computer science. The 1986.
applications of new proposed algorithm in general are: [16] Glenn W. Rowe: “Introduction to Data Structure and
Algorithms with C++”, PHI, ISBN: 81-203-1277-5.
1. The optimality with the simplicity of the algorithm [17] Robert Sedgewick: “Algorithms in C++”, Ed. 3rd, 2001,
makes this method an alternative to other methods for ISBN: 81-7808-249-7.
temporary storage of a tree in secondary memory.
Almost all operating systems store files in trees or tree [18] C. P. Kruskal: “Searching, merging, and sorting in
like structures. parallel computation”, IEEE Transactions on Computers,
2. It may for instance be used on tree structures that are 32:924-946, 1983.
expensive to construct, such as optimum search trees and [19] Lindstrom, G. Scanning: “List structures without stacks
multidimensional trees. or tag bits”. Inform. Proc. Letters 2, 1973, pp. 47-51.
3. Trees are also used in compiler design, text processing [20] Arora N., Tamta V. and Kumar S: “Modified Non-
and searching algorithms. Recursive Algorithm for Reconstructing a Binary Tree
4. In Huffman Coding instead of sending the whole tree in from its Traversals”, IJCA, volume 43-No.10, April
transmission, redundant sequences of in-order and pre- 2012.
order are send from the sender to the receiver. If a wrong
tree is accidently formed then the whole process of

13