A Synthesis Method for
Quaternary Quantum Logic Circuits

Quaternary Galois field GF(4) is an algebraic structure consisting the set of elements Q={0, 1, 2, 3}. The addition ($+$) and multiplication (.) operations over GF(4) are shown in the Table 1.

The logical NOT in quaternary quantum system is defined as $NOT(a)=a+1$, where ${}^{\prime}+^{\prime}$ denotes the modulo 4 addition, and $a$ ={0, 1, 2, 3}.

We present nine projection operations, grouped into three types $L_{i}$, $J_{i}$, and $P_{i}$, where $i$ = {$0$, $1$, $2$, $3$}. While $L_{i}$ and $J_{i}$ types were defined earlier [11, 12] as
$L_{i}(a)$ = $1$ if $a$ = $i$ and $0$ otherwise, and
$J_{i}(a)$ = $2$ if $a$ = $i$ and $0$ otherwise.
We introduce the new $P_{i}$ type operations, which are defined as
$P_{i}(a)$ = 3 if a = $i$ and a = $0$ otherwise.
Table 2 presents the truth tables for $L_{i}$, $J_{i}$, and $P_{i}$ types of operators as well as for the derived operators $L^{\prime}_{i}$, $J^{\prime}_{i}$ and $P^{\prime}_{i}$. The $L_{i}$, $J_{i}$ and $P_{i}$ operations are commutative, associative and distributive over AND and OR logic.

The $2$-qudit Quaternary Feynman gate [8] is defined as:
Feynman($A$, $B$) = $A$ + $B$, where ’+’ operator is addition over GF(4)(Figure 1.a).
The quaternary $3$-qudit Toffoli gate [8], shown in Figure 1.b is defined as:
Toffoli$(A,B,C)=A.B+C$, where $A$ and $B$ are the control inputs and $C$ the target input, ’+’ and ’.’ operators are addition and multiplication over GF(4). We use the quaternary $MAX$ and $MIN$ gates [13] respectively to replace the OR and the AND gates. These two gates are defined as
MAX($A_{1},A_{2},..,A_{n}$, $B$)=$\left\{\begin{array}[]{ll}\mbox{$A_{i}$}&\mbox{if $A_{i}$ $\geq$ $A_{j}$, $i\neq j$ and $A_{i}$ $\geq$ $B$};\\ \mbox{$B$}&\mbox{if $\forall i,B\geq A_{i}$};\end{array}\right.$
MIN($A_{1},A_{2},...A_{n}$, $B$) =$\left\{\begin{array}[]{ll}\mbox{$A_{i}$}&\mbox{ if $A_{i}$ $\leq$ $A_{j}$, $i\neq j$ and $A_{i}\leq B$};\\ \mbox{$B$}&\mbox{if $\forall i,B\leq A_{i}$};\end{array}\right.$
where $i$ = $\{1,2,3...n\}$ (Figures 2.a, 2.b). The quaternary Feynman and Toffoli gates can be realized using quaternary M-S gates [7, 8], defined as (Figure 2.c)
M-S($A$, $B$) = $Z$ operation on $B$ if $A$ = $3$, otherwise $B$, where $Z$ is one of the $24$ shift operations [10] shown in Table 3.

A new generalized quaternary gate (GQG) is required to realize the 24 shift operations [10] given in the Table 3. It is a multi-qudit gate shown in Figure 3.a. The controlling input of GQG can be used to select the $1$-qudit shift operation on the target input. The GQG is formally defined as
$GQG(A_{1},A_{2},..,A_{n},B)$=$\left\{\begin{array}[]{llll}\mbox{$B$ shift $X$}&\mbox{if $A_{1},A_{2},..,A_{n}=0$};\\ \mbox{$B$ shift $Y$}&\mbox{if $A_{1},A_{2},..,A_{n}=1$};\\ \mbox{$B$ shift $Z$}&\mbox{if $A_{1},A_{2},..,A_{n}=2$};\\ \mbox{$B$ shift $W$}&\mbox{if $A_{1},A_{2},..,A_{n}=3$};\\ \mbox{$B$ }&\mbox{otherwise};\end{array}\right.$
The realization of a GQG gate using quaternary M-S gates, is shown in Figure 3.b.

We propose a new $3$-qudit $C^{2}CS$ gate, used for realizing the simplification rules for quaternary minterms, as
$C^{2}CS(A,B,C)=\left\{\begin{array}[]{ll}\mbox{$C^{0123}$}&\mbox{if $A\neq B$};\\ \mbox{$C$}&\mbox{otherwise};\\ \end{array}\right.$ where the values of the inputs are from the set {1, 2, 3} and $C^{0123}$ is the cyclic shift operation $x^{0123}$ defined in Table 3. The symbolic representation of $C^{2}CS$ is shown in Figure 5.a and an instance of this gate is shown in Figure 5.b, where $x^{0123}$ shift operation is applied on $C$ if $A=1$, $B=3$ or $A=3$, $B=1$. The realization of the $C^{2}CS$ gate using M-S gates is shown in Figure 6.

We introduce the new $2$-qudit quaternary Modulo$4$ Addition gate as: ADD(A, B)= $A$ $\bigoplus$ $B$, where $\bigoplus$ represent the modulo $4$ addition. This gate can be used as a template for simplification of adder circuits. The symbolic representation of ADD gate is shown in Figure 7.a and the realization using quaternary M-S gate is shown in Figure 7.b

Consider an $m$-variable quaternary quantum logic function
$f(a_{1},a_{2},\ldots,a_{m})$ = $\sum^{n}_{i=0}$(minterms for one )_i + $\sum^{p}_{j=0}$(minterms for two)_j + $\sum^{s}_{k=0}$(minterms for three)_k,
where $\sum$ implies logical quaternary OR, n, p and s are respectively the number of input vectors for which $f$ is $1$, $2$ and $3$. Thus, $f$ is 0 for $(4^{m}-n-p-s)$ of the input vectors. We express the minterms for one, two and three using the $L_{i}$, $J_{i}$ and $P_{i}$ operations respectively. From Table II, we can state that
$\prod^{m}_{i=1}L_{0}(a_{i})$ = $\left\{\begin{array}[]{ll}\mbox{1}&\mbox{if $\forall i$ $a_{i}$ = 0};\\ \mbox{0}&\mbox{if $\exists$ i $a_{i}$ = 1, 2 or 3};\end{array}\right.$ $\prod^{m}_{i=1}L_{1}(a_{i})$ = $\left\{\begin{array}[]{ll}\mbox{1}&\mbox{if $\forall i$ $a_{i}$ = 1};\\ \mbox{0}&\mbox{if $\exists$ i $a_{i}$ = 0, 2 or 3};\end{array}\right.$
$\prod^{m}_{i=1}L_{2}(a_{i})$ =$\left\{\begin{array}[]{ll}\mbox{1}&\mbox{if $\forall i$ $a_{i}$ = 2};\\ \mbox{0}&\mbox{if $\exists$ i $a_{i}$ = 0, 1 or 3};\end{array}\right.$ $\prod^{m}_{i=1}L_{3}(a_{i})$ =$\left\{\begin{array}[]{ll}\mbox{1}&\mbox{if $\forall i$ $a_{i}$ = 3};\\ \mbox{0}&\mbox{if $\exists$ i $a_{i}$ = 0, 1 or 2};\end{array}\right.$
$\prod^{m}_{i,p,k,s=1}L_{0}(a_{i}).L_{1}(a_{p}).L_{2}(a_{k}).L_{3}(a_{s})$ =$\left\{\begin{array}[]{lll}\mbox{1}&\mbox{if $\forall i,p,k,s$ $a_{i}$ = 0, $a_{p}$ = 1, $a_{k}$ = 2, $a_{s}$ = 3};\\ \mbox{0}&\mbox{if $\exists$ i, p, k, s $a_{i}$ = 1, 2 or 3, $a_{p}$ = 0, 2 or 3,};\\ \mbox{ }&\mbox{$a_{k}$ = 0, 1 or 2, $a_{s}$ = 0, 1 or 2};\end{array}\right.$
where $i+p+k+s=m$.

Hence, the minterms for which $f=1$ are
1. $\prod^{m}_{i=1}L_{0}(a_{i}=0)$, 2. $\prod^{m}_{i=1}L_{1}(a_{i}=1)$, 3. $\prod^{m}_{i=1}L_{2}(a_{i}=2)$,
4. $\prod^{m}_{i=1}L_{3}(a_{i}=3)$, 5. $\prod^{m}_{i,p,k,s=1}L_{0}(a_{i}=0).L_{1}(a_{p}=1).L_{2}(a_{k}=2).L_{3}(a_{s}=3).$

Similarly from Table 2, the minterms for which $f=2$ are
1. $\prod^{m}_{i=1}J_{0}(a_{i}=0)$, 2. $\prod^{m}_{i=1}J_{1}(a_{i}=1)$, 3. $\prod^{m}_{i=1}J_{2}(a_{i}=2)$,
4. $\prod^{m}_{i=1}J_{3}(a_{i}=3)$, 5. $\prod^{m}_{i,p,k,s=1}J_{0}(a_{i}=0).J_{1}(a_{p}=1).J_{2}(a_{k}=2).J_{3}(a_{s}=3).$

The minterms for which $f=3$ are
1. $\prod^{m}_{i=1}P_{0}(a_{i}=0)$, 2. $\prod^{m}_{i=1}P_{1}(a_{i}=1)$, 3. $\prod^{m}_{i=1}P_{2}(a_{i}=2)$,
4. $\prod^{m}_{i=1}P_{3}(a_{i}=3)$, 5. $\prod^{m}_{i,p,k,s=1}P_{0}(a_{i}=0).P_{1}(a_{p}=1).P_{2}(a_{k}=2).P_{3}(a_{s}=3)$.

Next, we define six simplification rules derived from Table 2
1. $L_{i}(a).0=0$, $J_{i}(a).0=0$ and $P_{i}(a).0=0$
2. $L_{i}(a).1=L_{i}(a)$, $J_{i}(a).2=J_{i}(a)$ and $P_{i}(a).3=P_{i}(a)$
3. $L_{i}(a)+0=L_{i}(a)$, $J_{i}(a)+0=J_{i}(a)$ and $P_{i}(a)+0=P_{i}(a)$
4. $L_{i}(a)+1=1$, $J_{i}(a)+2=2$ and $P_{i}(a)+3=3$
5. $L_{i}(a).L^{\prime}_{i}(a)=0$, $J_{i}(a).J^{\prime}_{i}(a)=0$ and $P_{i}(a).P^{\prime}_{i}(a)=0$
6. $L_{i}(a)+L^{\prime}_{i}(a)=1$, $J_{i}(a)+J^{\prime}_{i}(a)=2$ and $P_{i}(a)+P^{\prime}_{i}(a)=3$

The truth table for the $2$-qudit arbitrary function $f(a,b)$ is given in Table 4.

We re-write the function $f(a,b)$ using our proposed methodology as
$f(a,b)$ = $L_{0}(a)L_{2}(b)+L_{2}(a)L_{0}(b)+L_{2}(a)L_{2}(b)+L_{3}(a)L_{1}(b)+J_{0}(a)J_{3}(b)+J_{1}(a)J_{2}(b)+J_{2}(a)J_{1}(b)+J_{3}(a)J_{0}(b)++J_{3}(a)J_{3}(b)P_{0}(a)P_{1}(b)+P_{1}(a)P_{0}(b)+P_{1}(a)P_{1}(b)+P_{2}(a)P_{3}(b)$+$P_{3}(a)P_{2}(b)$
By using simplification rules $7$ and $9$, we get
$f(a,b)$=$C^{2}CS(a_{02},b_{02},0)+L_{2}(a,b)+L_{3}(a)L_{1}(b)+C^{2}CS(a_{03},b_{03},1)+C^{2}CS(a_{12},b_{12},1)+J_{3}(a,b)+C^{2}CS(a_{10},b_{10},2)+C^{2}CS(a_{23},b_{23},0)+P_{1}(a,b)$

We synthesize the $2$-qudit half and full adder circuits using our proposed methodology and these circuits are simplified with the use of Modulo4 Addition gate. But the details are not provided due to limitation of space. The gate level implementation of $2$-qudit half and full adder are shown in Figure $8$ and $9$ respectively.

The quantum cost of the half and the full adder circuits, as shown in Figures 10 and 11, and of the benchmark circuits such as $sum2$, $rd53$, $rd73$, $xor5$ are evaluated in terms of the number of M-S gates used. The number of M-S gates required to realize a $2$-qudit Feynman gate, a $3$-qudit Toffoli, $MAX$ and $MIN$ gates are 5, 17, 6, 6 respectively [7, 13]. To realize the new GQG, $C^{2}CS$, ADD gates we need 8 M-S gates for each of them. The M-S gate count cost comparison of our circuits with that in [14] are given in Table 6. While the second column of the table indicates the maximum number of ancilla qudits required to synthesize the above mentioned circuits, the third column shows the number of ancilla qudits required after using our simplification rules in Section 4.2. The columns 4, 5 and 6 establish that although our design has a small increase in cost for the circuits $rd53$, $rd73$, $xor5$, the results for the $2$-qudit quaternary half and full adder shows that there is more than $50\%$ reduction in the M-S gate count cost compared to [14].