This work is licensed under the Creative Commons Attribution 4.0 International License.
Introduction and preliminary results
Let n ≥ 0 be an integer. The nth Mersenne number Mn and the nth Mersenne–Lucas number Hn are defined recursively by
{M_n} = 3{M_{n - 1}} - 2{M_{n - 2}},{\rm{ for }}n \ge 2{\rm{ with}}\,{M_0} = 0,{M_1} = 1
and
{H_n} = 3{H_{n - 1}} - 2{H_{n - 2}},\,\,{\rm{ for }}\,\,n \ge 2{\rm{with}}\,\,{H_0}{\rm{ }} = {\rm{ 2}},{\rm{ }}{H_1} = 3
respectively. Note that Mersenne–Lucas numbers are also called as Fermat numbers. The Binet type formulas of these sequences have the form Mn =2n− 1 and Hn = 2n + 1, so Hn = Mn + 2.
Mersenne sequence has been studied in many papers, see for example [2, 3, 6, 7, 9]. In the literature, we can find some generalizations of Mersenne numbers, see [4, 10]. In [8], Ochalik and Włoch introduced the generalized Mersenne numbers as follows. Let k ≥ 3 be a fixed integer. For any integer n ≥ 0 let M (k, n) be the nth generalized Mersenne number defined by the second order linear recurrence relation of the form
M(k,n) = kM(k,n - 1) - (k - 1)M(k,n - 2)
for n ≥ 2 with M (k, 0) = 0 and M (k, 1) = 1.
For n = 0, 1, 2, 3, 4, . . . the generalized Mersenne numbers are 0, 1, k, k2− k + 1, k3− 2k2 + 2k, . . .. Moreover, M(3, n) = Mn.
By analogy, we define the generalized Mersenne–Lucas numbers in the following way. Let k ≥ 3 be a fixed integer. For any integer n ≥ 0 let H(k, n) be the nth generalized Mersenne–Lucas number defined by
H(k,n) = kH(k,n - 1) - (k - 1)H(k,n - 2)
for n ≥ 2 with H(k, 0) = 2 and H (k, 1) = 3.
Then the first few terms of the generalized Mersenne–Lucas sequence are 2, 3, k + 2, k2− k + 3, k3− 2k2 + 2k + 2, . . .. It is easily seen that H(3, n) = Hn.
Proposition 1.1
Let k ≥ 3 be a fixed integer. For any integer n ≥ 0 we have H(k, n) = M(k, n) + 2.
Proof
(By induction on n.) If n = 0 then M0 = 0, H0 = 2. If n = 1 then M1 = 1, H1 = 3. Now assume that for any n ≥ 0, we have H(k, n) = M(k, n)+2 and H(k, n+1) = M(k, n+1)+2. We shall show that H(k, n+2) = M(k, n + 2) + 2. Applying the induction’s hypothesis we obtain
\matrix{
{H(k,n + 2)} \hfill & = \hfill & {kH(k,n + 1) - (k - 1)H(k,n)} \hfill \cr
{} \hfill & = \hfill & {k(M(k,n + 1) + 2) - (k - 1)(M(k,n) + 2)} \hfill \cr
{} \hfill & = \hfill & {kM(k,n + 1) - (k - 1)M(k,n) + 2} \hfill \cr
{} \hfill & = \hfill & {M(k,n + 2) + 2,} \hfill \cr
}
and by the induction’s rule the formula follows.
Some identities, properties, combinatorial interpretations and matrix generators of M(k, n) were given in [8] and [11]. In the next part of the paper we use the following results.
Let n ≥ 0, k ≥ 3 be integers. ThenM(k,n + 1) = \left( {k - 1} \right)M\left( {k,n} \right) + 1.
Using the fact that H(k, n) = M(k, n) + 2, we can give some properties of generalized Mersenne–Lucas numbers.
Corollary 1.5
Let n ≥ 0, k ≥ 3 be integers. ThenH(k,n) = {{{{(k - 1)}^n} + 2k - 5} \over {k - 2}},H(k,n + 1) - H(k,n) = {(k - 1)^n}andH(k,n + 1) = (k - 1)H(k,n) - 2k + 5.
The Mersenne numbers and their generalizations have applications also in the theory of hypercomplex numbers. In [5], Daşdemir and Bilgici introduced and studied Mersenne quaternions, Gaussian Mersenne numbers and generalized Mersenne quaternions. In [11], the authors considered the Mersenne hybrid numbers and generalized Mersenne hybrid numbers. In this paper, we use the Mersenne, Mersenne–Lucas numbers and their generalizations in the theory of bihyperbolic numbers.
Hyperbolic numbers are two dimensional number system. Hyperbolic imaginary unit, so-called unipotent, is an element h ≠ ±1 such that h2 = 1. Bihyperbolic numbers are a generalization of hyperbolic numbers. Let ℍ2 be the set of bihyperbolic numbers ζ of the form
\varsigma = {x_0} + {x_1}{j_1} + {x_2}{j_2} + {x_3}{j_3},
where x0, x1, x2, x3 ∈ and j1, j2, j3 ∉ R are operators such that
j_1^2 = j_2^2 = j_3^2 = 1,{j_1}{j_2} = {j_2}{j_1} = {j_3},{j_1}{j_3} = {j_3}{j_1} = {j_2},{j_2}{j_3} = {j_3}{j_2} = {j_1}.
From the above rules the multiplication of bihyperbolic numbers can be made analogously to the multiplication of algebraic expressions. The addition and the subtraction of bihyperbolic numbers is done by adding and subtracting corresponding terms and hence their coefficients. The addition and multiplication on ℍ2 are commutative and associative, (ℍ2, +, ·) is a commutative ring. For the algebraic properties of bihyperbolic numbers, see [1].
Let n ≥ 0 be an integer. The nth bihyperbolic Mersenne number BhMn and the nth bihyperbolic Mersenne–Lucas number BhHn are defined by
Bh{M_n} = {M_n} + {M_{n + 1}}{j_1} + {M_{n + 2}}{j_2} + {M_{n + 3}}{j_3},Bh{H_n} = {H_n} + {H_{n + 1}}{j_1} + {H_{n + 2}}{j_2} + {H_{n + 3}}{j_3},
respectively, where Mn is the nth Mersenne number, Hn is the nth Mersenne–Lucas number and j1, j2, j3 are units which satisfy (1.9).
The nth generalized bihyperbolic Mersenne number
BhM_n^k
we define in the following way
BhM_n^k = M(k,n) + M(k,n + 1){j_1} + M(k,n + 2){j_2} + M(k,n + 3){j_3},
where M(k, n) denotes the nth generalized Mersenne number, defined by (1.1). By analogy, the nth bihyperbolic Mersenne–Lucas number
BhH_n^k
is defined by
BhH_n^k = H(k,n) + H(k,n + 1){j_1} + H(k,n + 2){j_2} + H(k,n + 3){j_3},
where H(k, n) denotes the nth generalized Mersenne–Lucas number, defined by (1.2). For k = 3 we have
BhM_n^3 = Bh{M_n}
and
BhH_n^3 = Bh{H_n}
.
For simplicity of notation let A = 1 + j1 + j2 + j3. Using (1.3), (1.6) and (1.12), we can write (2.1) and (2.2) as
BhM_n^k = A \cdot M(k,n) + {(k - 1)^n}BhM_0^k
and
BhH_n^k = A \cdot H(k,n) + {(k - 1)^n}BhM_0^k,
respectively.
Using the Binet formula (2.3) and identity (1.3), we can derive the Catalan identity for the generalized bihyperbolic Mersenne numbers.
Theorem 2.8
Let n ≥ 0, r ≥ 0, k ≥ 3 be integers such that n ≥ r. Then\eqalign{
& BhM_{n + r}^k \cdot BhM_{n - r}^k - {\left( {BhM_n^k} \right)^2} \cr
& \,\,\,\,\,\,\,\,\,\, = {{{{(k - 1)}^n} - 1} \over {k - 2}}({k^2} + 2){(k - 1)^{n - r}}{(1 - {(k - 1)^r})^2}(1 + {j_1} + {j_2} + {j_3}). \cr}
Let n ≥ 1 be an integer. ThenBh{M_{n + 1}} \cdot Bh{M_{n - 1}} - {\left( {Bh{M_n}} \right)^2}\, = 11({2^n} - 1){2^{n - 1}}(1 + {j_1} + {j_2} + {j_3}).
Corollary 2.15
Let n ≥ 1 be an integer. ThenBh{H_{n + 1}} \cdot Bh{H_{n - 1}} - {\left( {Bh{H_n}} \right)^2}\, = 11({2^n} + 1){2^{n - 1}}(1 + {j_1} + {j_2} + {j_3}).
Now we give ordinary generating functions for the generalized bihyperbolic Mersenne and Mersenne–Lucas numbers.
Theorem 2.16
The generating function for the generalized bihyperbolic Mersenne number sequence\left\{ {BhM_n^k} \right\}isG(t) = {{BhM_0^k + (BhM_1^k - kBhM_0^k)t} \over {1 - kt + (k - 1){t^2}}}
Proof
Assume that the generating function of the generalized bihyperbolic Mersenne number sequence
\left\{ {BhM_n^k} \right\}
has the form
G(t) = \sum\nolimits_{n = 0}^\infty {BhM_n^k{t^n}}
Then
\eqalign{
& (1 - kt + (k - 1){t^2})G(t) \cr
& \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, = (1 - kt + (k - 1){t^2})(BhM_0^k + BhM_1^kt + BhM_2^k{t^2} + \ldots ) \cr
& \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, = BhM_0^k + BhM_1^kt + BhM_2^k{t^2} + \ldots \cr
& \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, - kBhM_0^kt - kBhM_1^k{t^2} - kBhM_2^k{t^3} - \ldots \cr
& \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, + (k - 1)BhM_0^k{t^2} + (k - 1)BhM_1^k{t^3} + (k - 1)BhM_2^k{t^4} + \ldots \cr
& \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, = BhM_0^k + (BhM_1^k - kBhM_0^k)t, \cr}
since
BhM_n^k = kBhM_{n - 1}^k - \left( {k - 1} \right)\,BhM_{n - 2}^k
and the coefficients of tn for n ≥ 2 are equal to zero. Moreover,
BhM_0^k = {j_1} + k{j_2} + ({k^2} - k + 1){j_3}
,
BhM_1^k - kBhM{_0^k} = 1 + \left({ - k + 1} \right){j_2} + \left({ - k^2} + k \right){j_ 3}
.
Theorem 2.17
The generating function for the generalized bihyperbolic Mersenne-Lucas number sequence\left\{ {BhM_n^k} \right\}isg(t) = {{BhH_0^k + (BhH_1^k - kBhH_0^k)t} \over {1 - kt + (k - 1){t^2}}}.
Proof
The proof of this theorem is similar to the proof of the previous theorem. Note only that
BhH_0^k = 2 + 3{j_1} + \left( {k + 2} \right){j_2} + \left( {{k^2} - k + 3} \right){j_3}
and
BhH_1^k - kBhH_0^k = \left( {3 - 2k} \right) + \left( {2 - 2k} \right){j_1} + \left( {3 - 3k} \right)j{_2} + \left({2 - k - k^2} \right){j_3}
.
Remark 2.18
The generating function γ(t) for the bihyperbolic Mersenne number sequence {BhMn} is
\gamma (t) = {{Bh{M_0} + (Bh{M_1} - 3Bh{M_0})t} \over {1 - 3t + 2{t^2}}},
where BhM0 = j1 + 3j2 + 7j3 and BhM1− 3BhM0 = 1 − 2j2− 6j3.
Remark 2.19
The generating function η(t) for the bihyperbolic Mersenne–Lucas number sequence {BhHn} is
\eta (t) = {{Bh{H_0} + (Bh{H_1} - 3Bh{H_0})t} \over {1 - 3t + 2{t^2}}},
where BhH0 = 2+3j1 +5j2 +9j3 and BhH1−3BhH0 = −3−4j1−6j2−10j3.
At the end, we give the matrix representations of the defined bihyperbolic numbers.
(By induction on n.) If n = 0 then assuming that the matrix to the power 0 is the identity matrix the result is obvious. Now suppose that for any n ≥ 0 holds
\left[ {\matrix{
{BhM_{n + 2}^k} & {BhM_{n + 1}^k} \cr
{BhM_{n + 1}^k} & {BhM_n^k} \cr
} } \right] = \left[ {\matrix{
{BhM_2^k} & {BhM_1^k} \cr
{BhM_1^k} & {BhM_0^k} \cr
} } \right] \cdot {\left[ {\matrix{
k \hfill & 1 \hfill \cr
{ - (k - 1)} \hfill & 0 \hfill \cr
} } \right]^n}.
Note that multiplication of bihyperbolic numbers is commutative and determinant properties can be used. For example, calculating determinants in Theorems 2.20–2.21 and Corollaries 2.22–2.23, we can also obtain Cassini identities. Using algebraic operations and matrix algebra could give many other interesting properties of these numbers.
Declarations
Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Conflict of Interest: The authors declare that they have no conflict of interest.