1

Cryptography in an Algebraic Alphabet

Dr. Htwe Htwe Oo
Associate Professor, Department of Mathematics, Taunggyi Technological University

[email protected]

Abstract – In this paper, we present a method of

encoding and decoding messages by using 26 alphabets Table 1
and their respective algebraic number (numerical
values). We use matrix and matrix with A B C D E F G H I J K L M N
their respective inverses to find ciphertext and 1 2 3 4 5 6 7 8 9
1
11
1 1
14
plaintext. Because of using 26 alphabets and their 0 2 3
respective algebraic numbers we use congruence
module 26 relation. Then we investigate the plaintext O P Q R S T U V W X Y Z
which starts with the given word and also investigate 1 1 1 1 1 2 2 2 2 2 2
0
5 6 7 8 9 0 1 2 3 4 5
the plaintext which ends with the given word.
Keywords – Matrices, Gaussian Elimination, Martix
In the simplest ciphers, successive pairs of plaintext are
Operations.
transformed into ciphertext by the following procedure:
I. INTRODUCTION Step 1 Choose a matrix with integer entries
The study of encoding and decoding secret
messages is called cryptography. Cryptography is a
to perform the encoding.
method of protecting information and communications
Step 2 Group successive plaintext letters into pairs,
through the use of codes, so that only those for whom the
adding an arbitrary “dummy” letter to fill out the
information is intended can read and process it. The prefix
last part if the plaintext has an odd number of
“crypt” means ‘hidden’ or ‘vault’ and the suffix ‘graphy’
letters, and replace each plaintext letter by its
stands for ‘writing’.
numerical value.
II. AIM Step 3 Successively convert each plaintext pair
To maintain the privacy of information
transmitted over public lines of communication.
into a column vector and form the
1. A Method of Encoding and Decoding product Ap. We will call p a plaintext vector and
Messages Ap the corresponding ciphertext vector.
Step 4 Convert each ciphertext vector into its alphabetic
1.1. Basic Concepts equivalent.
In the language of cryptography, codes are called From the above facts, we will do the plaintext message
ciphers, uncoded massages are called plaintext, and coded
messages are called ciphertext. The process of converting ROME WAS NOT BUILT IN A DAY
from plaintext to ciphertext is called enciphering, and the
reverse process of converting from ciphertext to plaintext into ciphertext message as follows:
is called deciphering. The simplest ciphers, called
substitution ciphers, are those that replace each letter of
alphabet by a different letter. To do this, we will use the matrix .
For example, in the substitution cipher, Because of having 21 letters in the plaintext message, we
will add the dummy letter Y to fill out the last pair.
Plain A B C D E F G H I J K L M N O P Q R S T U V W We will group the plaintext into pairs as
XYZ
RO ME WA SN OT BU IL TI NA DA YY.
Cipher D E F G H I J K L M N O P Q R S T U V W X Y
Z A B C. Or, equivalently, from Table 1,

The plaintext letter A is replaced by D, and so forth. 18 15 13 5 23 1 19 14 15 20 2 21 9 12 20 9 14 1

In this discussion, we assume that each plaintext 41 25 25.
and ciphertext letter except Z is assigned the numerical To encipher the pair RO, we perform the matrix
value that specifies its position in the standard alphabet.
For reasons that will become clear later, Z is assigned a
value of zero. product .
This is because, the remainder after making
division by 26 is one of the integers 0, 1, 2, …, 25, this
procedure will always yield an integer with an alphabet
equivalent.
 2

The remaining ciphertext vectors are

(mod 26)

(mod 26)

.
These correspond to the ciphertext pairs VS, WO,
YC, UP, CH, RK, GJ, LA, PC, FC, and WW, respectively.
So, the entire ciphertext message is

V S W O Y C U P C H R K G J L A P C F C W W.

It is obviously also possible to group the plaintext

in three-tuples and encipher by a matrix with integer
entries.

Using the matrix , we can obtain the

ciphertext message for the plaintext message
These correspond to the ciphertext pairs CER, FGL, EWO,
CONFERENCE ON SCIENCE AND TECHNOLOGY. EER, SDS, EWO, EQR, DIY, CXR, OEW, GVU
respectively.
Because of having 32 letters in the plaintext message we So the entire ciphertext message is
add dummy letter Y to fill out the last three-tuples. Then
we obtain CER FGL EWO EER SDS EWO EQR DIY CXR OEW
GVU.
CON FER ENC EON SCI ENC EAN DTE CHN OLO In general, we can do groups into sets of n letters
GYY by an matrix.

or, equivalently, from Table 1, 1.2. Definitions

If m is a positive integer and ‘a’ and ‘b’ are any
3 15 14 6 5 18 5 14 3 5 15 14 19 3 9 5 14 3 integers, then we say that ‘a’ is equivalent to ‘b’ modulo
5 1 14 4 20 5 3 8 14 15 12 15 7 25 25. m, written if m divides .
To encipher the pairs, we form the matrix products as For example,
follows:
 3

. 1.5. Theorem
For any modulus m, it can be proved that every
integer ‘a’ is equivalent, module m, to exactly one of the A square matrix A with entries in is
invertible modulo m if and only if the residue of det A
integers . We call this integer the residue ‘a’
modulo m has a reciprocal modulo m.
modulo m.
1.6. Corollary
We write to denote the set A square matrix A with entries in is
of residue module m. invertible modulo m if and only if m and the residue of det
A modulo m have no common prime factors.
1.3. Theorem
For any integer ‘a’ and modulus m, let 1.7. Corollary
A square matrix A with entries in is
. invertible modulo 26 if and only if the residue of det A
Then the residue r of a modulo m is given by modulo 26 is not divisible by 2 or 13.

If we have , where the entries are in

and the residue of modulo 26 is not
divisible by 2 or 13, then the inverse of A(mod 26) is
For example,

given by where

is the reciprocal of the residue of

.

If ‘a’ is a number in , then a number in 1.8. Example (Inverse of a matrix mod 26)

is called a reciprocal or multiplicative inverse of ‘a’

If we consider the matrix ,
modulo m if .
If ‘a’ and m have no common prime factors, then .
‘a’ has a unique reciprocal modulo m; So from Table 2, .
Conversely, if ‘a’ and m have a common prime
factor, then ‘a’ has no reciprocal modulo m.
In , and, 3 and 26 have no common Thus

prime factor. The reciprocal of 3 in is 9 since

.
. As a check,
The number 4 has no reciprocal modulo 26,
because 4 and 26 have 2 as a common prime factor.
We can provide the following table of reciprocal
modulo 26 as Table 2.
.
Table 2 Reciprocals modulo 26
Similarly, .
1
a 1 3 5 7 9 11 15 17 21 23 25
9 2. Decoding Ciphertext Message
1
1 9 21 15 3 19 7 23 5 17 25
1
2.1. Example
We will decode the previous ciphertext message
1.4. Deciphering
V S W O Y C U P C H R K G J L A P C F C W W,
If m is a positive integer, then the square matrix
A with entries in is said to be invertible modulo m if
which was enciphered by the matrix .
there is a matrix B with entries in such that The numerical equivalent of this ciphertext is
, where I represents identity matrix.
22 19 23 15 25 3 21 16 3 8 18 11 7 10 12 1 16 3 6 3 23 23.
If p is a plaintext vector, then is To obtain the plaintext pairs, we multiply each
the corresponding ciphertext vector and ciphertext vector by the inverse of the matrix A.
. So, and .
 4

The numerical equivalent of this ciphertext is

Then we get 3 5 18 6 7 12 5 23 15 5 5 18 19 4 19 5 23 15
5 17 18 4 9 25 3 24 18 15 5 23 7 22 21.
. To obtain the plaintext pair, we multiply each ciphertext
Then vector by the inverse of the matrix A.
So and .
Then we get

Then

(mod 26)

(mod 26)
From Table 1, the alphabet equivalents of these vectors
yields the message

ROME WAS NOT BUILT IN A DAY.

Similarly, we can deduce the message

SAKNOXAOJX

with enciphering matrix . Hence, we will get the

plaintext message

WE LOVE MATH.

We can decode also the ciphertext message

CER FGL EWO EER SDS EWO EQR DIY CXR OEW
GVU,

which was enciphered by the matrix .

From Table 1, the alphabet equivalents of these vectors
yields the message

. CONFERENCE ON SCIENCE AND TECHNOLOGYY

 5

which yields the message so the corresponding plaintext and ciphertext vectors are

CONFERENCE ON SCIENCE AND TECHNOLOGY.

We will conclude this section by discussing one

technique for breaking ciphers. We can deduce that the
.
message is a letter that begins with a word ends with a
We want to reduce
word. We will show with a small amount of such data, it
may be possible to determine the deciphering matrix of a
code and consequently obtain access to the rest of the
message. to I by elementary row operations and simultaneously
If we have a cipher and if are
plaintext vectors whose corresponding ciphertext vectors
apply these operations to to obtain
are known, then there is enough
information available to determine the matrix A and hence (the transpose of the deciphering matrix). This can
. be accomplished by adjoining P to the right of C and
The following theorem, provides a way to do this. applying row operations to the resulting matrix
until the left side is reduced to I. The final matrix will then
2.2. Theorem (Determining the Deciphering
Matrix)
have the form . The computations can be
Let be plaintext vectors, and carried out as follows:
be the corresponding ciphertext vectors in
n-cipher.
form the matrix

multiply the first row by

If is the matrix and if is replace 45 by its residue

the matrix, then the sequence of elementary row modulo 26

operations that reduces to I transforms p to .

The following example illustrates a simple add times the first
row to the second
algorithm for doing this.

3. Decoding Ciphertext Messages with the

Starting Word and Last Word replace the entries in the
second row by their
3.1. Example residues modulo 26
The following cipher is intercepted:

I O S B T G X E S P X H O P D E. multiply the second row

by
To decipher the message, given that it starts with the word
DEAR.
replace the entries in the
From Table 1, the numerical equivalent of the known second row by their
plaintext is residues modulo 26

D E A R
4 5 1 18 add times the
second row to the first
and the numerical equivalent of the corresponding
ciphertext is
replace the entries in the
I O S B first row by their residues
9 15 19 2 modulo 26.
 6

ATOM

Thus, , and so the deciphering matrix is and the corresponding ciphertext is

JYQO.
. We will do as the above procedure from example
To decipher the message, we first group the ciphertext into (3.1), but we have to take the letters from the end to
pairs and find the numerical equivalent of each letter: forward.

I O S B T G X E Then we get the resulting matrix as follows:

9 15 19 2 20 7 24 5

S P X H O P D E .
19 16 24 8 15 16 4 5 After the computations had been carried out, we get the
Next, we multiply successive ciphertext vectors on the left
by and find the alphabet equivalents of the resulting matrix and so the deciphering matrix
plaintext pairs:

is .
To decipher the message, we multiply the successive
ciphertext vectors from the end to forward.
Then we get as follows;

(mod 26)

(mod 26)

.
Finally, we construct the message from the plaintext pairs:

DE AR IK ES EN DT AN KS

as DEAR IKE SEND TANKS. .

Finally, we get the message
3.2. Example
To decode the ciphertext THEY SPLIT THE ATOM.
L N G I H G Y B V R E N J Y Q O, given that the last
four plaintext letters are known to be ATOM. ACKNOWLEDGEMENT
We draw the following table;
Foremost, I would like to express my sincere
L N G I H G Y B
gratitude to my advisor Prof. Dr. Thet Khaing (Rtd.) for
the continuous support of my paper for his patience,
1 1 2
7 9 8 7 2 motivation, enthusiasm and immerse knowledge. His
2 4 5
guidance helped me in the research and writing of this
V R E N J Y Q O paper.
2 1 1 1 2 1 1
5 III. CONCLUSION
2 8 4 0 5 7 5
We have observed how to do encoding and
From Table 1, the numerical equivalent of the known decoding secret messages with the aid of matrix
plaintext is operations. We have studied some examples of decoding
messages with the given beginning and last words. We
 7

have also expressed basic concepts of modular arithmetic

and reciprocal of a number modulo 26.

REFERENCES
[1] Lester S. Hill, “Cryptography in an Algebraic
Alphabet”, The American Mathematical Monthly,
Vol. 36, No. 6. (Jun. - Jul., 1929), pp. 306-312.
[2] Lester S. Hill, “Cryptography in an Algebraic
Alphabet”, The American Mathematical Monthly,
Vol. 38, No. 3 (Mar., 1931), pp. 135-154.
[3] Howard Anton and Chris Rorres, “Elementary
Linear Algebra: Applications Version, 10th,
2010.