Abstract
Cyclic lattices and ideal lattices were introduced by Micciancio (2002), Lyubashevsky and Micciancio (2006), respectively, which play an efficient role in Ajtai’s construction of a collision resistant Hash function (see Ajtai (1996), Ajtai and Dwork (1997)) and in Gentry’s construction of fully homomorphic encryption (see Gentry (2009)). Let \(R=Z[x]/\langle \phi (x)\rangle \) be a quotient ring of the integer coefficients polynomials ring, Lyubashevsky and Micciancio regarded an ideal lattice as the correspondence of an ideal of R, but they neither explain how to extend this definition to whole Euclidean space \(\mathbb {R}^n\), nor exhibit the relationship of cyclic lattices and ideal lattices. In this chapter, we regard the cyclic lattices and ideal lattices as the correspondences of finitely generated R-modules, so that we may show that ideal lattices are actually a special subclass of cyclic lattices, namely, cyclic integer lattices. In fact, there is a one to one correspondence between cyclic lattices in \(\mathbb {R}^n\) and finitely generated R-modules (see Theorem 4). On the other hand, since R is a Noether ring, each ideal of R is a finitely generated R-module, so it is natural and reasonable to regard ideal lattices as a special subclass of cyclic lattices (see Corollary 7). It is worth noting that we use a more general rotation matrix here, so our definition and results on cyclic lattices and ideal lattices are more general forms. As an application, we provide a cyclic lattice with an explicit and countable upper bound for the smoothing parameter (see Theorem 5). It is an open problem that is the shortest vector problem on cyclic lattice NP-hard (see Micciancio (2002)). Our results may be viewed as a substantial progress in this direction.
You have full access to this open access chapter, Download conference paper PDF
Similar content being viewed by others
Keywords
1 Discrete Subgroup in \(\mathbb {R}^n\)
Let \(\mathbb {R}\) be the real numbers field, \(\mathbb {Z}\) be the integers ring, and \(\mathbb {R}^n\) be Euclidean space of which is an n-dimensional linear space over \(\mathbb {R}\) with the Euclidean norm |x| given by
We use column vector notation for \(\mathbb {R}^n\) through out this chapter, and \(x'=(x_1,x_2, \dots ,x_n)\) is transpose of x, which is called row vector of \(\mathbb {R}^n\).
Definition 1
Let \(L\subset \mathbb {R}^n\) be a non-trivial additive subgroup, it is called a discrete subgroup if there is a positive real number \(\lambda >0\) such that
As usual, a ball of center \(x_0\) with radius \(\delta \) is defined by
If L is a discrete subgroup of \(\mathbb {R}^n\), then there are only finitely many vectors of L lie in every ball \(b(0,\delta )\), thus we always find a vector \(\alpha \in L\) such that
\(\alpha \) is called one of shortest vector of L and \(\lambda \) is called the minimum distance of L.
Let \(B=[\beta _1,\beta _2,\dots ,\beta _m]\in \mathbb {R}^{n\times m}\) be a \(n\times m\) dimensional matrix with rank\((B) = m\leqslant n\), it means that \(\beta _1,\beta _2,\dots ,\beta _m\) are m linearly independent vectors in \(\mathbb {R}^n\). The lattice L(B) generated by B is defined by
which is all linear combinations of \(\beta _1,\beta _2,\dots ,\beta _m\) over \(\mathbb {Z}\). If \(m=n\), L(B) is called a full-rank lattice.
It is a well-known conclusion that a discrete subgroup L in \(\mathbb {R}^n\) is just a lattice L(B). Firstly, we give a detailed proof here by making use of the simultaneous Diophantine approximation theory in real number field \(\mathbb {R}\) (see Cassels (1971) and Cassels (1963)).
Lemma 1
Let \(L\subset \mathbb {R}^n\) be a discrete subgroup, \(\alpha _1,\alpha _2,\dots ,\alpha _m \in L\) be m vectors of L. Then \(\alpha _1,\alpha _2,\dots ,\alpha _m\) are linearly independent over \(\mathbb {R}\), if and only if which are linearly independent over \(\mathbb {Z}\).
Proof
If \(\alpha _1,\alpha _2,\dots ,\alpha _m\) are linearly independent over \(\mathbb {R}\), trivially which are linearly independent over \(\mathbb {Z}\). Suppose that \(\alpha _1,\alpha _2,\dots ,\alpha _m\) are linearly independent over \(\mathbb {Z}\), we consider arbitrary linear combination over \(\mathbb {R}\). Let
We should prove (1.4) is equivalent to \(a_1=a_2=\cdots =a_m=0\), which implies that \(\alpha _1,\alpha _2,\dots ,\alpha _m\) are linearly independent over \(\mathbb {R}\).
By Minkowski’s Third Theorem (see Theorem VII of Cassels (1963)), for any sufficiently large \(N>1\), there are a positive integer \(q\geqslant 1\) and integers \(p_1,p_2,\dots ,p_m \in \mathbb {Z}\) such that
By (1.4), we have
Let \(\lambda \) be the minimum distance of L, \(\varepsilon >0\) be any positive real number. We select N such that
It follows that \(mN^{-\frac{1}{m}}<\varepsilon \) and
By (1.6) we have
Since \(p_1 \alpha _1+p_2 \alpha _2+\cdots +p_m \alpha _m \in L\), thus we have \(p_1 \alpha _1+p_2 \alpha _2+\cdots +p_m \alpha _m=0\), and \(p_1=p_2=\cdots =p_m=0\). By (1.5) we have \(q|a_i|<\frac{1}{m} \varepsilon \) for all i, \(1\leqslant i\leqslant m\). Since \(\varepsilon \) is a sufficiently small positive number, we must have \(a_1=a_2=\cdots =a_m=0\). We complete the proof of lemma.
Suppose that \(B\in \mathbb {R}^{n\times m}\) is an \(n\times m\)-dimensional matrix and rank\((B) = m\), \(B'\) is the transpose of B. It is easy to verify
which implies that \(B'B\) is an invertible square matrix of \(m\times m\) dimension. Since \(B'B\) is a positive defined symmetric matrix, then there is an orthogonal matrix \(P\in \mathbb {R}^{m\times m}\) such that
where \(\delta _i>0\) are the characteristic value of \(B'B\), and diag\(\{\delta _1,\delta _2,\dots ,\delta _m\}\) is the diagonal matrix of \(m\times m\) dimension.
Lemma 2
Suppose that \(B\in \mathbb {R}^{n\times m}\) with rank\((B) = m\), \(\delta _1,\delta _2,\dots ,\delta _m\) are m characteristic values of \(B'B\), and \(\lambda (L(B))\) is the minimum distance of lattice L(B), then we have
where \(\delta =\min \{\delta _1,\delta _2,\dots ,\delta _m\}\).
Proof
Let \(A=B'B\), by (1.7), there exists an orthogonal matrix \(P\in \mathbb {R}^{m\times m}\) such that
If \(x\in \mathbb {Z}^m\), \(x\ne 0\), we have
Since \(x\in \mathbb {Z}^m\) and \(x\ne 0\), we have \(|x|^2\geqslant 1\), it follows that
We have Lemma 2 immediately.
Another application of Lemma 2 is to give a countable upper bound for smoothing parameter (see Theorem 5). Combining Lemmas 1 and 2, we show the following assertion.
Theorem 1
Let \(L\subset \mathbb {R}^n\) be a subset, then L is a discrete subgroup if and only if there is an \(n\times m\) dimensional matrix \(B\in \mathbb {R}^{n\times m}\) with rank\((B) = m\) such that
Proof
If \(L \subset \mathbb {R}^n\) is a discrete subgroup, then L is a free \(\mathbb {Z}\)-module. By Lemma 1, we have \(\text {rank}_{\mathbb {Z}}(L) = m\leqslant n\). Let \(\beta _1,\beta _2,\dots ,\beta _m\) be a \(\mathbb {Z}\)-basis of L, then
Writing \(B=[\beta _1,\beta _2,\dots ,\beta _m]_{n\times m}\), then the rank of matrix B is m, and
Conversely, let L(B) be arbitrary lattice generated by B, obviously, L(B) is an additive subgroup of \(\mathbb {R}^n\), by Lemma 2, L(B) is also a discrete subgroup, we have Theorem 1 at once.
Corollary 1
Let \(L\subset \mathbb {R}^n\) be a lattice and \(G\subset L\) be an additive subgroup of L, then G is a lattice of \(\mathbb {R}^n\).
Corollary 2
Let \(L\subset \mathbb {Z}^n\) be an additive subgroup, then L is a lattice of \(\mathbb {R}^n\). These lattices are called integer lattices.
According to above Theorem 1, a lattice L(B) is equivalent to a discrete subgroup of \(\mathbb {R}^n\). Suppose \(L=L(B)\) is a lattice with generated matrix \(B\in \mathbb {R}^{n\times m}\), and rank\((B) = m\), we write rank\((L) = \)rank(B), and
In particular, if rank\((L) = n\) is a full-rank lattice, then \(d(L)=|\text {det}(B)|\) as usual. A sublattice N of L means a discrete additive subgroup of L, the quotient group is written by L/N, and the cardinality of L/N is denoted by |L/N|.
Lemma 3
Let \(L\subset \mathbb {R}^n\) be a lattice and \(N\subset L\) be a sublattice. If rank\((N) = \)rank(L), then the quotient group L/N is a finite group.
Proof
Let rank\((L) = m\), and \(L=L(B)\), where \(B\in \mathbb {R}^{n\times m}\) with rank\((B) = m\). We define a mapping \(\sigma \) from L to \(\mathbb {Z}^m\) by \(\sigma (Bx)=x\). Clearly, \(\sigma \) is an additive group isomorphism, \(\sigma (N)\subset \mathbb {Z}^m\) is a full-rank lattice of \(\mathbb {Z}^m\), and \(L/N \cong \mathbb {Z}^m/\sigma (N)\). It is a well-known result that
It follows that
Lemma 3 follows.
Suppose that \(L_1\subset \mathbb {R}^n\), \(L_2\subset \mathbb {R}^n\) are two lattices of \(\mathbb {R}^n\), we define \(L_1+L_2=\{a+b|a\in L_1,b\in L_2\}\). Obviously, \(L_1+L_2\) is an additive subgroup of \(\mathbb {R}^n\), but generally speaking, \(L_1+L_2\) is not a lattice of \(\mathbb {R}^n\) again.
Lemma 4
Let \(L_1\subset \mathbb {R}^n\), \(L_2\subset \mathbb {R}^n\) be two lattices of \(\mathbb {R}^n\). If rank\((L_1 \cap L_2) = \)rank\((L_1)\) or rank\((L_1 \cap L_2) = \)rank\((L_2)\), then \(L_1+L_2\) is again a lattice of \(\mathbb {R}^n\).
Proof
To prove \(L_1+L_2\) is a lattice of \(\mathbb {R}^n\), by Theorem 1, it is sufficient to prove \(L_1+L_2\) is a discrete subgroup of \(\mathbb {R}^n\). Suppose that rank\((L_1 \cap L_2) = \)rank\((L_1)\), for any \(x\in L_1\), we define a distance function \(\rho (x)\) by
Since there are only finitely many vectors in \(L_2\cap b(x,\delta )\), where \(b(x,\delta )\) is any a ball of center x with radius \(\delta \). Therefore, we have
On the other hand, if \(x_1\in L_1\), \(x_2\in L_1\), and \(x_1-x_2\in L_2\), then there is \(y_0\in L_2\) such that \(x_1=x_2+y_0\), and we have \(\rho (x_1)=\rho (x_2)\). It means that \(\rho (x)\) is defined over the quotient group \(L_1+L_2/L_2\). Because we have the following group isomorphic theorem
By Lemma 3, it follows that
In other words, \(L_1+L_2/L_2\) is also a finite group. Let \(x_1,x_2,\dots ,x_k\) be the representative elements of \(L_1+L_2/L_2\), we have
Therefore, \(L_1+L_2\) is a discrete subgroup of \(\mathbb {R}^n\), thus it is a lattice of \(\mathbb {R}^n\) by Theorem 1.
Remark 1
The condition rank\((L_1 \cap L_2) = \)rank\((L_1)\) or rank\((L_1 \cap L_2) = \)rank\((L_2)\) in Lemma 4 seems to be necessary. As a counterexample, we see the real line \(\mathbb {R}\), let \(L_1=\mathbb {Z}\) and \(L_2=\sqrt{2}\mathbb {Z}\), then \(L_1+L_2\) is not a discrete subgroup of \(\mathbb {R}\), thus \(L_1+L_2\) is not a lattice in \(\mathbb {R}\). Because \(L_1+L_2=\{n+\sqrt{2}m\big | n\in \mathbb {Z},m\in \mathbb {Z}\}\) is dense in \(\mathbb {R}\) by Dirichlet’s Theorem (see Theorem I of Cassels (1963)).
As a direct consequence, we have the following generalized form of Lemma 4.
Corollary 3
Let \(L_1,L_2,\dots ,L_m\) be m lattices of \(\mathbb {R}^n\) and
Then \(L_1+L_2+\cdots +L_m\) is a lattice of \(\mathbb {R}^n\).
Proof
Without loss of generality, we assume that
Let \(L_1+L_2+\cdots +L_{m-1}=L'\), then
Since rank\((L'\cap L_m) = \)rank\((L_m)\), by Lemma 4, we have \(L'+L_m=L_1+L_2+\cdots +L_m\) is a lattice of \(\mathbb {R}^n\) and the corollary follows.
2 Ideal Matrices
Let \(\mathbb {R}[x]\) and \(\mathbb {Z}[x]\) be the polynomials rings over \(\mathbb {R}\) and \(\mathbb {Z}\) with variable x, respectively. Suppose that
is a polynomial with integer coefficients of which has no multiple roots in complex numbers field \(\mathbb {C}\). Let \(w_1,w_2,\dots ,w_n\) be the n different roots of \(\phi (x)\) in \(\mathbb {C}\), the Vandermonde matrix \(V_{\phi }\) is defined by
According to the given polynomial \(\phi (x)\), we define a rotation matrix \(H=H_{\phi }\) by
where \(I_{n-1}\) is the \((n-1)\times (n-1)\) unit matrix. Obviously, the characteristic polynomial of H is just \(\phi (x)\).
We use column notation for vectors in \(\mathbb {R}^n\), for any \(f=\begin{pmatrix} f_0 \\ f_1 \\ \vdots \\ f_{n-1} \end{pmatrix}\in \mathbb {R}^n\), the ideal matrix generated by vector f is defined by
which is a block matrix in terms of each column \(H^k f\ (0\leqslant k\leqslant n-1)\). Sometimes, f is called an input vector. It is easily seen that \(H^*(f)\) is a more general form of the classical circulant matrix (see Davis (1994)) and r-circulant matrix (see Shi (2018), Yasin and Taskara (2013)). In fact, if \(\phi (x)=x^n-1\), then \(H^*(f)\) is the ordinary circulant matrix generated by f. If \(\phi (x)=x^n-r\), then \(H^*(f)\) is the r-circulant matrix.
By (2.4), it follows immediately that
Moreover, \(H^*(f)=0\) is a zero matrix if and only if \(f=0\) is a zero vector, thus one has \(H^*(f)=H^*(g)\) if and only if \(f=g\). Let \(M^*\) be the set of all ideal matrices, namely
We may regard \(H^*\) as a mapping from \(\mathbb {R}^n\) to \(M^*\) of which is a one to one correspondence.
In Zheng et al. (2023), we have shown some basic properties of ideal matrix, most of them may be summarized as the following theorem.
Theorem 2
Suppose that \(\phi (x)\in \mathbb {Z}[x]\) is a fixed polynomial with no multiple roots in \(\mathbb {C}\), then for any two column vectors f and g in \(\mathbb {R}^n\), we have
-
(i)
\(H^*(f)=f_0 I_n+f_1 H+\cdots +f_{n-1}H^{n-1}\);
-
(ii)
\(H^*(f)H^*(g)=H^*(H^*(f)g)\) and \(H^*(f)H^*(g)=H^*(g)H^*(f)\);
-
(iii)
\(H^*(f)=V_{\phi }^{-1}\ \text {diag}\{f(w_1),f(w_2),\dots ,f(w_n)\}V_{\phi }\);
-
(iv)
det \((H^*(f))=\Pi _{i=1}^n f(w_i)\);
-
(v)
\(H^*(f)\) is an invertible matrix if and only if \((f(x),\phi (x))=1\) in \(\mathbb {R}[x]\),
where \(V_{\phi }\) is the Vandermonde matrix given by (2.2), \(w_i\ (1\leqslant i\leqslant n)\) are all roots of \(\phi (x)\) in \(\mathbb {C}\), and diag\(\{f(w_1),f(w_2),\dots ,f(w_n)\}\) is the diagonal matrix.
Proof
See Theorem 2 of Zheng et al. (2023).
Let \(e_1,e_2,\dots ,e_n\) be unit vectors of \(\mathbb {R}^n\), that is
It is easy to verify that
This means that the unit matrix \(I_n\) and rotation matrices \(H^k\ (1\leqslant k\leqslant n-1)\) are all the ideal matrices.
Let \(\phi (x)\mathbb {R}[x]\) and \(\phi (x)\mathbb {Z}[x]\) be the principal ideals generated by \(\phi (x)\) in \(\mathbb {R}[x]\) and \(\mathbb {Z}[x]\), respectively, we denote the quotient rings R and \(\overline{R}\) by
There is a one to one correspondence between \(\overline{R}\) and \(\mathbb {R}^n\) given by
We denote this correspondence by t, that is
If we restrict t in the quotient ring R, then which gives a one to one correspondence between R and \(\mathbb {Z}^n\). First, we show that t is also a ring isomorphism.
Definition 2
For any two column vectors f and g in \(\mathbb {R}^n\), we define the \(\phi \)-convolutional product \(f*g\) by \(f*g=H^*(f)g\).
By Theorem 2, it is easy to see that
Lemma 5
For any two polynomials f(x) and g(x) in \(\overline{R}\), we have
Proof
Let \(g(x)=g_0+g_1 x+\cdots +g_{n-1}x^{n-1}\in \overline{R}\), then
It follows that
Hence, for any \(0\leqslant k\leqslant n-1\), we have
Let \(f(x)=f_0+f_1 x+\cdots +f_{n-1}x^{n-1}\in \overline{R}\), by (i) of Theorem 2, we have
The lemma follows.
Theorem 3
Under \(\phi \)-convolutional product, \(\mathbb {R}^n\) is a commutative ring with identity element \(e_1\) and \(\mathbb {Z}^n\subset \mathbb {R}^n\) is its subring. Moreover, we have the following ring isomorphisms:
where \(M^*\) is the set of all ideal matrices given by (2.6), and \(M_{\mathbb {Z}}^{*}\) is the set of all integer ideal matrices.
Proof
Let \(f(x)\in \overline{R}\) and \(g(x)\in \overline{R}\), then
and
This means that t is a ring isomorphism. Since \(f*g=g*f\) and \(e_1*g=H^*(e_1)g=I_n g=g\), then \(\mathbb {R}^n\) is a commutative ring with \(e_1\) as the identity elements. Noting \(H^*(f)\) is an integer matrix if and only if \(f\in \mathbb {Z}^n\) is an integer vector, the isomorphism of subrings follows immediately.
According to property (v) of Theorem 2, \(H^*(f)\) is an invertible matrix whenever \((f(x),\phi (x))=1\) in \(\mathbb {R}[x]\), we show that the inverse of an ideal matrix is again an ideal matrix.
Lemma 6
Let \(f(x)\in \overline{R}\) and \((f(x),\phi (x))=1\) in \(\mathbb {R}[x]\), then
where \(u(x)\in \overline{R}\) is the unique polynomial such that \(u(x)f(x)\equiv 1\) (mod \(\phi (x)\)).
Proof
By Lemma 5, we have \(u*f=e_1\), it follows that
Thus we have \((H^*(f))^{-1}=H^*(u)\). It is worth to note that if \(H^*(f)\) is an invertible integer matrix, then \((H^*(f))^{-1}\) is not an integer matrix in general.
Sometimes, the following lemma may be useful, especially, when we consider an integer matrix.
Lemma 7
Let \(f(x)\in \mathbb {Z}[x]\) and \((f(x),\phi (x))=1\) in \(\mathbb {Z}[x]\), then we have \((f(x),\phi (x))=1\) in \(\mathbb {R}[x]\).
Proof
Let Q be the rational number field. Since \((f(x),\phi (x))=1\) in \(\mathbb {Z}[x]\), then \((f(x),\phi (x))=1\) in \(\mathbb {Q}[x]\). We know that \(\mathbb {Q}[x]\) is a principal ideal domain, thus there are two polynomials a(x) and b(x) in \(\mathbb {Q}[x]\) such that
This means that \((f(x),\phi (x))=1\) in \(\mathbb {R}[x]\).
3 Cyclic Lattices and Ideal Lattices
As we know that cyclic code plays a central role in the algebraic coding theorem (see Chap. 6 of Lint (1999)). In Zheng et al. (2023), we extended ordinary cyclic code to more general forms, namely \(\phi \)-cyclic codes. To obtain an analogous concept of \(\phi \)-cyclic code in \(\mathbb {R}^n\), we note that every rotation matrix H defines a linear transformation of \(\mathbb {R}^n\) by \(x\rightarrow Hx\).
Definition 3
A linear subspace \(C\subset \mathbb {R}^n\) is called a \(\phi \)-cyclic subspace if \(\forall \alpha \in C\Rightarrow H\alpha \in C\). A lattice \(L\subset \mathbb {R}^n\) is called a \(\phi \)-cyclic lattice if \(\forall \alpha \in L\Rightarrow H\alpha \in L\).
In other words, a \(\phi \)-cyclic subspace C is a linear subspace of \(\mathbb {R}^n\), of which is closed under linear transformation H. A \(\phi \)-cyclic lattice L is a lattice of \(\mathbb {R}^n\) of which is closed under H. If \(\phi (x)=x^n-1\), then H is the classical circulant matrix and the corresponding cyclic lattice first appeared in Micciancio (2002), but he does not discuss the further property for these lattices. To obtain the explicit algebraic construction of \(\phi \)-cyclic lattice, we first show that there is a one to one correspondence between \(\phi \)-cyclic subspaces of \(\mathbb {R}^n\) and the ideals of \(\overline{R}\).
Lemma 8
Let t be the correspondence between \(\overline{R}\) and \(\mathbb {R}^n\) given by (2.9), then a subset \(C\subset \mathbb {R}^n\) is a \(\phi \)-cyclic subspace of \(\mathbb {R}^n\), if and only if \(t^{-1}(C)\subset \overline{R}\) is an ideal.
Proof
We extend the correspondence t to subsets of \(\overline{R}\) and \(\mathbb {R}^n\) by
Let \(C(x)\subset \overline{R}\) be an ideal, it is clear that \(C\subset t(C(x))\) is a linear subspace of \(\mathbb {R}^n\). To prove C is a \(\phi \)-cyclic subspace, we note that if \(c(x)\in C(x)\), then by (2.11)
Therefore, if C(x) is an ideal of \(\overline{R}\), then \(t(C(x))=C\) is a \(\phi \)-cyclic subspace of \(\mathbb {R}^n\). Conversely, if \(C\subset \mathbb {R}^n\) is a \(\phi \)-cyclic subspace, then for any \(k\geqslant 1\), we have \(H^k c\in C\) whenever \(c\in C\), it implies
which means that C(x) is an ideal of \(\overline{R}\). We complete the proof.
By the above lemma, to find a \(\phi \)-cyclic subspace in \(\mathbb {R}^n\), it is enough to find an ideal of \(\overline{R}\). There are two trivial ideals \(C(x)=0\) and \(C(x)=\overline{R}\), the corresponding \(\phi \)-cyclic subspace are \(C=0\) and \(C=\mathbb {R}^n\). To find non-trivial \(\phi \)-cyclic subspaces, we make use of the homomorphism theorems, which is a standard technique in algebra. Let \(\pi \) be the natural homomorphism from \(\mathbb {R}[x]\) to \(\overline{R}\), ker\(\pi =\phi (x)\mathbb {R}[x]\). We write \(\phi (x)\mathbb {R}[x]\) by \(<\phi (x)>\). Let N be an ideal of \(\mathbb {R}[x]\) satisfying
Since \(\mathbb {R}[x]\) is a principal ideal domain, then \(N=<g(x)>\) is a principal ideal generated by a monic polynomial \(g(x)\in \mathbb {R}[x]\). It is easy to see that
It follows that all ideals N satisfying (2) are given by
We write by \(<g(x)>\) mod \(\phi (x)\), the image of \(<g(x)>\) under \(\pi \), i.e.
It is easy to check
more precisely, which is a representative elements set of \(<g(x)>\) mod \(\phi (x)\). By homomorphism theorem in ring theory, all ideals of \(\overline{R}\) are given by
Let d be the number of monic divisors of \(\phi (x)\) in \(\mathbb {R}[x]\), we have the following.
Corollary 4
The number of \(\phi \)-cyclic subspace of \(\mathbb {R}^n\) is d.
Next, we discuss \(\phi \)-cyclic lattice, which is the geometric analogy of cyclic code. The \(\phi \)-cyclic subspace of \( \mathbb {R}^{n}\) may be regarded as the algebraic analogy of cyclic code. Let the quotient rings R and \(\overline{R}\) be given by (2.8). A R-module is an Abel group \(\wedge \) such that there is an operator \(\lambda \alpha \in \wedge \) for all \(\lambda \in R\) and \(\alpha \in \wedge \), satisfying \(1\cdot \alpha =\alpha \) and \((\lambda _1 \lambda _2)\alpha =\lambda _1 (\lambda _2 \alpha )\). It is easy to see that \(\overline{R}\) is a R-module, if \(\wedge \subset \overline{R}\) and \(\wedge \) is a R-module, then \(\wedge \) is called a R-submodule of \(\overline{R}\). All R-modules we discuss here are R-submodule of \(\overline{R}\). On the other hand, if \(I\subset R\), then I is an ideal of R, if and only if I is a R-module. Let \(\alpha \in \overline{R}\), the cyclic R-module generated by \(\alpha \) be defined by
If there are finitely many polynomials \(\alpha _1,\alpha _2,\dots ,\alpha _k\) in \(\overline{R}\) such that \(\wedge =R\alpha _1+R\alpha _2+\cdots +R\alpha _k\), then \(\wedge \) is called a finitely generated R-module, which is a R-submodule of \(\overline{R}\).
Now, if \(L\subset \mathbb {R}^n\) is a \(\phi \)-cyclic lattice, \(g\in \mathbb {R}^n\), \(H^*(g)\) is the ideal matrix generated by vector g, and \(L(H^*(g))\) is the lattice generated by \(H^*(g)\). It is easy to show that any \(L(H^*(g))\) is a \(\phi \)-cyclic lattice and
which implies that \(L(H^*(g))\) is the smallest \(\phi \)-cyclic lattice of which contains vector g. Therefore, we call \(L(H^*(g))\) is a minimal \(\phi \)-cyclic lattice in \(\mathbb {R}^n\).
Lemma 9
There is a one to one correspondence between the minimal \(\phi \)-cyclic lattice in \(\mathbb {R}^n\) and the cyclic R-submodule in \(\overline{R}\), namely,
and
Proof
Let \(b(x)\in R\), by Lemma 5, we have
and \(t(Rg(x))\subset L(H^*(g))\). Conversely, if \(\alpha \in L(H^*(g))\), and \(\alpha =H^*(g)b\) for some integer vector b, by Lemma 5 again, we have \(b(x)g(x)\in Rg(x)\), and \(t(b(x)g(x))=\alpha \). This implies that \(L(H^*(g))\subset t(Rg(x))\), and
The lemma follows immediately.
Suppose \(L=L(\beta _1,\beta _2,\dots ,\beta _m)\) is arbitrary \(\phi \)-cyclic lattice, where \(B=[\beta _1,\beta _2, \dots ,\beta _m]_{n\times m}\) is the generated matrix of L. L may be expressed as the sum of finitely many minimal \(\phi \)-cyclic lattices, in fact, we have
To state and prove our main results, first, we give a definition of prime spot in \(\mathbb {R}^n\).
Definition 4
Let \(g\in \mathbb {R}^n\), and \(g(x)=t^{-1}(g)\in \overline{R}\). If \((g(x),\phi (x))=1\) in \(\mathbb {R}[x]\), we call g is a prime spot of \(\mathbb {R}^n\).
By (v) of Theorem 2, \(g\in \mathbb {R}^n\) is a prime spot if and only if \(H^*(g)\) is an invertible matrix, thus the minimal \(\phi \)-cyclic lattice \(L(H^*(g))\) generated by a prime spot is a full-rank lattice.
Lemma 10
Let g and f be two prime spots of \(\mathbb {R}^n\), then \(L(H^*(g))+L(H^*(f))\) is a full-rank \(\phi \)-cyclic lattice.
Proof
According to Lemma 4, it is sufficient to show that
In fact, we should prove in general
Since \(H^*(g)\cdot H^*(f)\) is an invertible matrix, then rank\(\big (L(H^*(g)\cdot H^*(f))\big ) = n\), and (8) follows immediately.
To prove (9), we note that
It follows that
It is easy to see that
Therefore, we have
This is the proof of Lemma 10.
It is worth to note that (9) is true for the more general case and does not need the condition of prime spot.
Corollary 5
Let \(\beta _1,\beta _2,\dots ,\beta _m\) be arbitrary m vectors in \(\mathbb {R}^n\), then we have
Proof
If \(\beta _1,\beta _2,\dots ,\beta _m\) are integer vectors, then (10) is trivial. For the general case, we write
where \(\beta _1 *\beta _2 *\cdots *\beta _m\) is the \(\phi \)-convolutional product, then
Since
It follows that
We have this corollary.
By Lemma 10, we also have the following assertion.
Corollary 6
Let \(\beta _1,\beta _2,\dots ,\beta _m\) be m prime spots of \(\mathbb {R}^n\), then \(L(H^*(\beta _1))+L(H^*(\beta _2))+\cdots +L(H^*(\beta _m))\) is a full-rank \(\phi \)-cyclic lattice.
Proof
It follows immediately from Corollary 3.
Our main result in this chapter is to establish the following one to one correspondence between \(\phi \)-cyclic lattices in \(\mathbb {R}^n\) and finitely generated R-modules in \(\overline{R}\).
Theorem 4
Let \(\wedge =R\alpha _1(x)+R\alpha _2(x)+\cdots +R\alpha _m(x)\) be a finitely generated R-module in \(\overline{R}\), then \(t(\wedge )\) is a \(\phi \)-cyclic lattice in \(\mathbb {R}^n\). Conversely, if \(L\subset \mathbb {R}^n\) is a \(\phi \)-cyclic lattice in \(\mathbb {R}^n\), then \(t^{-1}(L)\) is a finitely generated R-module in \(\overline{R}\), that is a one to one correspondence.
Proof
If \(\wedge \) is a finitely generated R-module, by Lemma 9, we have
The main difficulty is to show that \(t(\wedge )\) is a lattice of \(\mathbb {R}^n\), we require a surgery to embed \(t(\wedge )\) into a full-rank lattice. To do this, let \((\alpha _i(x),\phi (x))=d_i(x)\), \(d_i(x)\in \mathbb {Z}[x]\), and \(\beta _i(x)=\alpha _i(x)/d_i(x)\), \(1\leqslant i\leqslant m\). Since \(\phi (x)\) has no multiple roots by assumption, then \((\beta _i(x),\phi (x))=1\) in \(\mathbb {R}[x]\). In other words, each \(t(\beta _i(x))=\beta _i\) is a prime spot. It is easy to verify \(R\alpha _i(x)\subset R\beta _i(x)\ (1\leqslant i\leqslant m)\), thus we have
By Corollaries 6 and 1, we have \(t(\wedge )\) is \(\phi \)-cyclic lattice. Conversely, if \(L\subset \mathbb {R}^n\) is a \(\phi \)-cyclic lattice of \(\mathbb {R}^n\), and \(L=L(\beta _1,\beta _2,\dots ,\beta _m)\), by (7), we have
which is a finitely generated R-module in \(\overline{R}\). We complete the proof of Theorem 4.
As we introduced in abstract, since R is a Noether ring, then \(I\subset R\) is an ideal if and only if I is a finitely generated R-module. On the other hand, if \(I\subset R\) is an ideal, then \(t(I)\subset \mathbb {Z}^n\) is a discrete subgroup of \(\mathbb {Z}^n\), thus t(I) is a lattice, we define the following.
Definition 5
Let \(I\subset R\) be an ideal, t(I) is called the \(\phi \)-ideal lattice.
Ideal lattice first appeared in Lyubashevsky and Micciancio (2006) (see Definition 3.1 of Lyubashevsky and Micciancio (2006)). As a direct consequence of Theorem 4, we have the following.
Corollary 7
Let \(L\subset \mathbb {R}^n\) be a subset, then L is a \(\phi \)-cyclic lattice if and only if
where \(\beta _i\in \mathbb {R}^n\) and \(m\leqslant n\). Furthermore, L is a \(\phi \)-ideal lattice if and only if every \(\beta _i\in \mathbb {Z}^n\), \(1\leqslant i\leqslant m\).
Corollary 8
Suppose that \(\phi (x)\) is an irreducible polynomial in \(\mathbb {Z}[x]\), then any non-zero ideal I of R defines a full-rank \(\phi \)-ideal lattice \(t(I)\subset \mathbb {Z}^n\).
Proof
Let \(I\subset R\) be a non-zero ideal, then we have \(I=R\alpha _1(x)+R\alpha _2(x)+\cdots +R\alpha _m(x)\), where \(\alpha _i(x)\in R\) and \((\alpha _i(x),\phi (x))=1\). It follows that
Since each \(\alpha _i\) is a prime spot, we have rank\((t(I))=n\) by Corollary 6, and the corollary follows at once.
According to Definition 3.1 of Lyubashevsky and Micciancio (2006), we have proved that any an ideal of R corresponding to a \(\phi \)-ideal lattice, which just is a \(\phi \)-cyclic integer lattice under the more general rotation matrix \(H=H_{\phi }\). Cyclic lattice and ideal lattice were introduced in Lyubashevsky and Micciancio (2006), Micciancio (2002), respectively, to improve the space complexity of lattice-based cryptosystems. Ideal lattices allow to represent a lattice using only two polynomials. Using such lattices, class lattice-based cryptosystems can diminish their space complexity from \(O(n^2)\) to O(n). Ideal lattices also allow to accelerate computations using the polynomial structure. The original structure of Micciancio’s matrices uses the ordinary circulant matrices and allows for an interpretation in terms of arithmetic in polynomial ring \(\mathbb {Z}[x]/<x^n-1>\). Lyubashevsky and Micciancio (2006) later suggested to change the ring to \(\mathbb {Z}[x]/<\phi (x)>\) with an irreducible \(\phi (x)\) over \(\mathbb {Z}[x]\). Our results here suggest to change the ring to \(\mathbb {Z}[x]/<\phi (x)>\) with any polynomial \(\phi (x)\). There are many works subsequent to Micciancio (2002, Lyubashevsky and Micciancio (2006), such as (Feige & Micciancio, 2004; Micciancio & Regev, 2009; Peikert, 2016; Plantard & Schneider, 2013; Pradhan et al., 2019; Stehle & Steinfeld, 2011).
Example 1
It is interesting to find some examples of \(\phi \)-cyclic lattices in an algebraic number field K. Let Q be a rational number field, without loss of generality, an algebraic number field K of degree n is just \(K=Q(w)\), where \(w=w_i\) is a root of \(\phi (x)\). If all \(Q(w_i)\subset \mathbb {R}\ (1\leqslant i\leqslant n)\), then K is called a totally real algebraic number field. Let \(O_K\) be the ring of algebraic integers of K, and \(I\subset O_K\) be an ideal, \(I\ne 0\). Since there is an integral basis \(\{\alpha _1,\alpha _2,\dots ,\alpha _n\}\subset I\) such that
We may regard every ideal of \(O_K\) as a lattice in \(Q^n\), and our assertion is that every non-zero ideal of \(O_K\) is corresponding to a full-rank \(\phi \)-cyclic lattice of \(Q^n\). To see this example, let
It is known that \(K=Q[w]\), thus every \(\alpha \in K\) corresponds to a vector \(\overline{\alpha }\in Q^n\) by
If \(I\subset O_K\) is an ideal of \(O_K\) and \(I=\mathbb {Z}\alpha _1+\mathbb {Z}\alpha _2+\cdots +\mathbb {Z}\alpha _n\), let \(B=[\overline{\alpha _1},\overline{\alpha _2},\dots ,\overline{\alpha _n}]\in Q^{n\times n}\), which is full-rank matrix. We have \(\tau (I)=L(B)\) as a full-rank lattice. It remains to show that \(\tau (I)\) is a \(\phi \)-cyclic lattice, we only prove that if \(\alpha \in I\Rightarrow H\overline{\alpha }\in \tau (I)\). Suppose that \(\alpha \in I\), then \(w\alpha \in I\). It is easy to verify that \(\tau (w)=e_2\) (see (2.7)) and
This means that \(\tau (I)\) is a \(\phi \)-cyclic lattice of \(Q^n\), which is a full-rank lattice.
4 Smoothing Parameter
As an application of the algebraic structure of \(\phi \)-cyclic lattice, we show an explicit upper bound of the smoothing parameter for the \(\phi \)-cyclic lattices. Firstly, we introduce some basic notations.
A Gauss function \(\rho _{s,c}(x)\) in \(\mathbb {R}^n\) is given by
where \(x\in \mathbb {R}^n\), \(c\in \mathbb {R}^n\), and \(s>0\) is a positive real number. \(\rho _{s,c}(x)\) is called the Gauss function around original point c with parameter s. It is easy to see that
Thus, we may define a probability density function \(D_{s,c}(x)\) by
Suppose \(L\subset \mathbb {R}^n\) is a lattice, let
The discrete Gauss distribution over L is a probability distribution \(D_{L,s,c}\) over L given by
If \(c=0\) is the zero vector of \(\mathbb {R}^n\), we write \(\rho _{s,0}(x)=\rho _{s}(x)\), \(\rho _{s,0}(L)=\rho _{s}(L)\), \(D_{s,0}(x)=D_{s}(x)\), and \(D_{s,0}(L)=D_{s}(L)\). Suppose that L is a full-rank lattice and \(L^*\) is its dual lattice, we define the smoothing parameter \(\eta _{\varepsilon }(L)\) of L to be the smallest s such that \(\rho _{1/s}(L^*)\leqslant 1+\varepsilon \), more precisely,
where \(\varepsilon >0\) is a positive number. Notice that \(\rho _{1/s}(L^*)\) is a continuous and strictly decreasing function of s, thus the smoothing parameter \(\eta _{\varepsilon }(L)\) is a continuous and strictly decreasing function of \(\varepsilon \).
Let \(L=L(\beta _1,\beta _2,\dots ,\beta _n)\subset \mathbb {R}^n\) be a full-rank lattice with a basis \(\beta _1,\beta _2,\dots ,\beta _n\), the fundamental region P(L) is given by
Suppose that X and Y are two discrete random variables on \(\mathbb {R}^n\), the statistical distance between X and Y over L is defined by
If X and Y are continuous random variables with probability density function \(T_1\) and \(T_2\), respectively, then \(\triangle (X,Y)\) is defined by
The smoothing parameter was introduced by Micciancio and Regev (2007), which plays an important role in the statistical information of lattices. An important property of smoothing parameter is for any lattice \(L=L(B)\) and any \(\varepsilon >0\), the statistical distance between \(D_s\) mod L and the uniform distribution over the fundamental region P(L) is at most \(\frac{1}{2}(\rho _{1/s}(L(B)^*))\). More precisely, for any \(\varepsilon >0\) and any \(s\geqslant \eta _{\varepsilon } (L(B))\), the statistical distance is at most \(\frac{1}{2}\varepsilon \), namely
Lemma 11
Let \(L\subset \mathbb {R}^n\) be a full-rank lattice, we have
where \(L^*\) is the dual lattice of L, and \(\lambda _1 (L^*)\) is the minimum distance of \(L^*\).
Proof
See Lemma 3.2 of Micciancio and Regev (2007), or Banaszczyk (1993).
Lemma 12
Suppose that \(L_1\) and \(L_2\) are two full-rank lattices in \(\mathbb {R}^n\), and \(L_1 \subset L_2\), then for any \(\varepsilon >0\), we have
Proof
Let \(\eta _{\varepsilon }(L_1)=s\), we are to show that \(\eta _{\varepsilon }(L_2)\leqslant s\). Since
It is easy to check that \(L_2^*\subset L_1^*\), it follows that
which implies
and \(\eta _{\varepsilon }(L_2)\leqslant s=\eta _{\varepsilon }(L_1)\), thus we have Lemma 12.
According to (2.4), the ideal matrix \(H^*(f)\) with input vector \(f\in \mathbb {R}^n\) is just the ordinary circulant matrix when \(\phi (x)=x^n-1\). Next lemma shows that the transpose of a circulant matrix is still a circulant matrix. For any \(g=\begin{pmatrix} g_0 \\ g_1 \\ \vdots \\ g_{n-1} \end{pmatrix}\in \mathbb {R}^n\), we denote \(\overline{g}=\begin{pmatrix} g_{n-1} \\ g_{n-2} \\ \vdots \\ g_{0} \end{pmatrix}\), which is called the conjugation of g.
Lemma 13
Let \(\phi (x)=x^n-1\), then for any \(g=\begin{pmatrix} g_0 \\ g_1 \\ \vdots \\ g_{n-1} \end{pmatrix}\in \mathbb {R}^n\), we have
Proof
Since \(\phi (x)=x^n-1\), then \(H=H_{\phi }\) (see (2.3)) is an orthogonal matrix, and we have \(H^{-1}=H^{n-1}=H'\). We write \(H_1=H'=H^{-1}\). The following identity is easy to verify
It follows that
and we have the lemma.
Lemma 14
Suppose that \(g\in \mathbb {R}^n\) and the circulant matrix \(H^*(g)\) is invertible. Let \(A=(H^*(g))'H^*(g)\), then all characteristic values of A are given by
where \(\theta _i^n=1\ (1\leqslant i\leqslant n)\) are the n-th roots of unity.
Proof
By Lemma 13 and (ii) of Theorem 2, we have
where \(g''=H^*(H\overline{g})g\). Let \(g''(x)=t^{-1}(g'')\) be the corresponding polynomial of \(g''\). By (iii) of Theorem 2, all characteristic values of A are given by
Let \(g=\begin{pmatrix} g_0 \\ g_1 \\ \vdots \\ g_{n-1} \end{pmatrix}\in \mathbb {R}^n\). It is easy to see that
where \(g_{-i}=g_{n-i}\) for all \(1\leqslant i\leqslant n-1\), then the lemma follows at once.
By definition 4, if \(g\in \mathbb {R}^n\) is a prime spot, then there is a unique polynomial \(u(x)\in \overline{R}\) such that \(u(x)g(x)\equiv 1\) (mod \(\phi (x)\)). We define a new vector \(T_g\) and its corresponding polynomial \(T_g(x)\) by
If \(g\in \mathbb {Z}^n\) is an integer vector, then \(T_g\in \mathbb {Z}^n\) is also an integer vector, and \(T_g(x)\in \mathbb {Z}[x]\) is a polynomial with integer coefficients. Our main result on smoothing parameter is the following theorem.
Theorem 5
Let \(\phi (x)=x^n-1\), \(L\subset \mathbb {R}^n\) be a full-rank \(\phi \)-cyclic lattice, then for any prime spots \(g\in L\), we have
where \(\theta _i^n=1\), \(1\leqslant i\leqslant n\), and \(T_g(x)\) is given by (4.14).
Proof
Let \(g\in L\) be a prime spot, by Lemma 12, we have
To estimate the smoothing parameter of \(L(H^*(g))\), the dual lattice of \(L(H^*(g))\) is given by
where \(u(x)\in \overline{R}\) and \(u(x)g(x)\equiv 1\) (mod \(x^n-1\)), and \(T_g\) is given by (4.14). Let \(A=(H^*(T_g))'H^*(T_g)\), by Lemma 14, all characteristic values of A are
By Lemma 2, the minimum distance \(\lambda _1 (L(H^*(g))^*)\) is bounded by
Now, Theorem 5 follows from Lemma 11 immediately.
Let \(L=L(B)\) be a full-rank lattice and \(B=[\beta _1,\beta _2,\dots ,\beta _n]\). We denote by \(B^*=[\beta _1^*,\beta _2^*,\dots ,\beta _n^*]\) the Gram-Schmidt orthogonal vectors \(\{\beta _i^*\}\) of the ordered basis \(B=\{\beta _i\}\). It is a well-known conclusion that
which yields by Lemma 11 the following upper bound
where \(B_0^*\) is the orthogonal basis of dual lattice \(L^*\) of L.
For a \(\phi \)-cyclic lattice L, we observe that the upper bound (4.17) is always better than (4.18) by numerical testing, we give two examples here.
Example 2
Let \(n=3\) and \(\phi (x)=x^3-1\), the rotation matrix H is
We select a \(\phi \)-cyclic lattice \(L=L(B)\), where
Since \(L=\mathbb {Z}^3\), thus L is a \(\phi \)-cyclic lattice. It is easy to check
On the other hand, we randomly find a prime spot \(g=\begin{pmatrix} 0\\ 0\\ 1 \end{pmatrix}\in L\) and \(g(x)=x^2\). Since \(xg(x)\equiv 1\) (mod \(x^3-1\)), we have \(T_g(x)=x^2\), it follows that \(|T_g(\theta _1)|=|T_g(\theta _2)|=|T_g(\theta _3)|=1\), and
Example 3
Let \(n=4\) and \(\phi (x)=x^4-1\), the rotation matrix H is
We select a \(\phi \)-cyclic lattice \(L=L(B)\), where
Since \(L=\mathbb {Z}^4\), thus L is a \(\phi \)-cyclic lattice. It is easy to check
On the other hand, we randomly find a prime spot \(g=\begin{pmatrix} -2 \\ 1 \\ 0 \\ 0 \end{pmatrix}\in L\) and \(g(x)=x-2\). Since \((\frac{1}{7}x^3-\frac{1}{7}x^2-\frac{2}{7}x-\frac{5}{7})g(x)\equiv 1\) (mod \(x^4-1\)), we have \(T_g(x)=-\frac{2}{7}x^3-\frac{1}{7}x^2+\frac{1}{7}x-\frac{5}{7}\), it follows that \(|T_g(\theta _1)|=1\), \(|T_g(\theta _2)|=|T_g(\theta _3)|=|T_g(\theta _4)|=\frac{5}{7}\), and
References
Ajtai, M. (1996). Generating hard instances of the short basis problem. In Proceedings of 28th STOC (pp. 99–108).
Ajtai, M., & Dwork, C. (1997). A public-key cryptosystem with worst-case/average-case equivalence. In Proceedings of 29th STOC (pp. 284–293).
Banaszczyk, W. (1993). New bounds in some transference theorems in the geometry of numbers. Mathematische Annalen, 296(4), 625–635.
Cassels, J. W. S. (1963). Introduction to diophantine approximation. Cambridge University Press.
Cassels, J. W. S. (1971). An introduction to the geometry of numbers. Springer.
Davis, P. J. (1994). Circulant matrices (2nd ed.). Chelsea Publishing.
Feige, U., & Micciancio, D. (2004). The inapproximability of lattice and coding problems with preprocessing. Journal of Computer and System Sciences, 69(1), 45–67.
Gentry, C. (2009). Fully homomorphic encryption using ideal lattices. Stoc.
Lint, J. H. V. (1999) Introduction to coding theory. Springer.
Lyubashevsky, V., & Micciancio, D. (2006) Generalized compact knapsacks are collision resistant. In Proceedings of the 33rd international conference on Automata, Languages and Programming—Proceedings of ICALP 2006 (Vol. 4052, pp. 144–155). Springer LNCS.
Micciancio, D. (2001). The hardness of the closest vector problem with preprocessing. IEEE Transactions on Information Theory, 47(3), 1212–1215.
Micciancio, D. (2022). Generalized compact knapsacks, cyclic lattices, and efficient one-way functions from worst-case complexity assumptions: (extended abstract). In Annual Symposium on Foundations of Computer Science.
Micciancio, D., & Regev, O. (2007). Worst-case to average-case reductions based on gaussian measures. SIAM Journal on Computing, 37(1), 267–302.
Micciancio, D., & Regev, O. (2009). Lattice-based cryptography. In D. J. Bernstein, J. Buchmann & E. Dahmen (Eds.), Post-quantum cryptography (pp. 147–191). Springer.
Peikert, C. (2016). A decade of lattice cryptography. Foundations and trends in theoretical computer science.
Plantard, T., & Schneider, M. (2013). Creating a challenge for ideal lattices (pp. 1–17).
Pradhan, P. K., Rakshit, S., & Datta, S. (2019). Lattice based cryptography: Its applications, areas of interest and future scope. In Proceedings of the Third International Conference on Computing Methodologies and Communication (pp. 988–993).
Regev, O. (2004). Improved inapproximability of lattice and coding problems with preprocessing. IEEE Transactions on Information Theory, 50(9), 2031–2037.
Shi, B. J. (2018). The spectral norms of geometric circulant matrices with the generalized k-Horadam numbers. Journal of Inequalities and Applications, 14.
Stehle, D., & Steinfeld, R.: Making NTRU as secure as worst-case problems over ideal lattices. In K. G. Paterson(Eds.), Advances in cryptology, lecture notes in computer sciences (Vol. 6632, pp. 27–47). Springer.
Yasin, Y., & Taskara, N. (2013). On the inverse of circulant matrix via generalized k-Horadam numbers. Applied Mathematics and Computation, 223, 191–196.
Zheng, Z.Y., Huang, W. L., Xu, J., & Tian, K. A generalization of cyclic code and applications to public key cryptosystems.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Copyright information
© 2023 The Author(s)
About this paper
Cite this paper
Zhiyong, Z., Fengxia, L., Yunfan, L., Kun, T. (2023). Cyclic Lattices, Ideal Lattices, and Bounds for the Smoothing Parameter. In: Zheng, Z. (eds) Proceedings of the Second International Forum on Financial Mathematics and Financial Technology. IFFMFT 2021. Financial Mathematics and Fintech. Springer, Singapore. https://doi.org/10.1007/978-981-99-2366-3_7
Download citation
DOI: https://doi.org/10.1007/978-981-99-2366-3_7
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-99-2365-6
Online ISBN: 978-981-99-2366-3
eBook Packages: Economics and FinanceEconomics and Finance (R0)