%%%%% journal info -- DO NOT CHANGE %%%%%%%%% %%%%%%%% end journal info %%%%%%%%%%%%% %%%%% author macros %%%%%%%%% % place your own macros HERE %%%%% end %%%%%%%%% %%%%%%%%% already loaded packages -- DO NOT RELOAD %%%%% % \usepackage{amssymb} % \usepackage{amsmath,amsthm} % \usepackage{latexsym} % \usepackage{amsfonts} % \usepackage{amsbsyt} % \usepackage{graphicx} %%%%%% end %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \documentclass{mc} \usepackage{amsfonts} \usepackage{float} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \usepackage{url} \usepackage[displaymath,mathlines,pagewise]{lineno} \setcounter{page}{91} \renewcommand\thisnumber{x} \renewcommand\thisyear {2019} \renewcommand\thismonth{xxx} \renewcommand\thisvolume{24} \renewcommand\datereceived{April 10, 2018} \renewcommand\dateaccepted{September 9, 2018} %Rad zaprimljen dana: 10.04.2018. %Rad prihvacen dana: 09.09.2018. \begin{document} %\begin{linenumbers} \title[Formally self-dual codes from block $\lambda$-circulant matrices]{% Constructing formally self-dual codes from block $\lambda$-circulant matrices% } \author[A.\,Kaya, B.\,Yildiz]{Abidin Kaya\affil{1}\comma\corrauth and Bahattin Yildiz\affil{2}} \markboth{A.~Kaya and B.~Yildiz}{Formally self-dual codes from block $\lambda$-circulant matrices} % multiple authors: % Please mark \corrauth after the name of the corresponding author. % different addresses: \address{\affilnum{1}\ Sampoerna Academy, L'Avenue Campus, 12\,780, Jakarta, Indonesia\\ \affilnum{2}\ Department of Mathematics and Statistics, Northern Arizona University, Flagstaff, AZ 86\,011, USA} \emails{{\texttt{abidin.kaya@sampoernaacademy.sch.id} (A.\,Kaya)}, {\texttt{% bahattin.yildiz@nau.edu} (B.\,Yildiz)}} \begin{abstract} In this work, construction methods for formally self-dual codes are generalized in the form of block $\lambda$-circulant matrices. The constructions are applied over the rings $\mathbb{F}_{2}$, $R_{1}=\mathbb{F}% _{2}+u\mathbb{F}_{2}$ and $S=\mathbb{F}_{2}[u]/(u^{3}-1)$. Using $n$-block $% \lambda $-circulant matrices for suitable integers $n$ and units $\lambda $, many binary FSD codes (as Gray images) with a higher minimum distance than best known self-dual codes of lengths 34, 40, 44, 54, 58, 70, 72 and 74 were obtained. In particular, ten new even FSD $\left[ 72,36,14\right]$ codes were constructed together with eight new near-extremal FSD even codes of length 44 and twenty-five new near-extremal FSD even codes of length 36. \end{abstract} \ams{94B05, 94B99.} \keywords{formally self-dual codes, near-extremal codes, circulant codes } \maketitle %%%%% title and author(s): % \markboth{Author(s)}{Short Title} % \title{Title} \ams{Primary 94B05, 94B99; Secondary 11T71, 13M99 } \section{Introduction} Formally self-dual codes are those that have the same weight enumerators as their duals. Iso-dual codes are codes that are equivalent to their duals. Of course self-dual codes are both iso-dual and formally self-dual but there are formally self-dual codes that are not self-dual. Formally self-dual codes can have better minimum distances than self-dual codes of the same lengths, as will be showcased throughout the paper. Formally self-dual codes work in the exact same way for the Assmus-Mattson theorem as self-dual codes and they have been studied quite extensively in the literature. For shortness of the notation, throughout the paper, we will use the abbreviation FSD to describe formally self-dual codes. Betsumiya and Harada studied binary optimal odd FSD codes in \cite{betsumiya}% , whereas Dougherty et al. considered optimal FSD codes over $\mathbb{F}_{5}$ and $\mathbb{F}_{7}$ in \cite{dougherty}. In \cite{betsumiya2}, a classification of FSD even codes of lengths up to 16 was done. In \cite% {fields}, a classification of extremal even FSD codes of lengths 20 and 22 was done. Extremal double circulant FSD even codes were classified in \cite% {gulliver}. Betsumiya and Harada considered FSD codes related to Type II codes in \cite{betsumiya3}. Near extremal FSD even codes of lengths 24 and 32 and later FSD even codes of lengths divisible by 8 were studied in \cite% {gulliver2} and \cite{kim}, respectively. In \cite{bouyuklieva}, a classification of extremal even FSD codes of length 30 was made. In \cite% {freibert}, optimal subcodes of FSD codes of lengths 16-22 were analysed by using their distance profiles. Parallel to our main methods, some recent studies have shown that good FSD codes can be obtained from FSD codes over certain rings. A family of rings of characteristic 2 was used quite effectively in \cite{karadeniz} to obtain FSD codes with better minimum distances than the self-dual ones as well as some near-extremal FSD codes. In \cite{batoul}, constructions were applied to the rings $\mathbb{F}_{3}+v\mathbb{F}_{3}$ and $\mathbb{F}_{5}+v\mathbb{F}% _{5}$ to find some good FSD codes over the fields $\mathbb{F}_{3}$ and $% \mathbb{F}_{5}$ with better minimum distances than the self-dual codes. Two main construction methods for FSD codes use double circulant and bordered double circulant matrices over the binary field as well as over appropriate rings. In this work, we will generalize these methods by introducing a general block-matrix construction as well as $\lambda $% -circulant matrices. Our construction methods generalize all the existing methods of constructions and their effectiveness is demonstrated by the highly substantial number of good FSD codes (near-extremal or better than self-dual) that we have obtained. We will apply our methods to three main alphabets, all of characteristic 2, namely the binary field $\mathbb{F}_{2}$% , the ring $R_{1}=\mathbb{F}_{2}+u\mathbb{F}_{2}$ and $S=\mathbb{F}% _{2}[u]/(u^{3}-1)$. In the case of $S$, since the Gray map is not duality-preserving, we will prove the existence of the MacWilliams identities that are preserved under the Gray map, which help us find binary FSD codes as the images of FSD codes over $S$. Using $n$-block $\lambda $% -circulant matrices for suitable integers $n$ and units $\lambda $, we were able to obtain many binary FSD codes with a higher minimum distance than best known self-dual codes of lengths 34, 40, 44, 54, 58, 70, 72 and 74. Ten new even $\left[ 72,36,14\right]$ codes were also constructed. In addition to these, eight new near-extremal FSD even codes of length 44 and twenty-five new near-extremal FSD even codes of length 36 were obtained. The rest of the paper is organized as follows: In Section 2, we give the necessary preliminaries about FSD codes as well as the rings over which the constructions are described. In particular, the MacWilliams identities are given for the Lee weight enumerators of codes over $S$, to demonstrate that the image of FSD codes are binary FSD codes. In Section 3, the main construction methods are described with the proofs. Section 4 includes numerical results, where many good FSD codes of different lengths constructed through our constructions are tabulated. \section{Preliminaries} In this paper, we will work mainly on the binary field $\mathbb{F}_{2}$ and two of its ring extensions, namely $\mathbb{F}_{2}+u\mathbb{F}_{2}$, where $% u^{2}=0$ and $S=\mathbb{F}_{2}[u]/(u^{3}-1)$. In order to understand the constructions better, we will recall some of the properties of these rings, leaving aside the binary field as an obvious case. We refer the reader to \cite{huffman} for the definitions and notations on codes. Let $R$ be one of the rings; $\mathbb{F}_{2}$, $R_{1}$ or $S$ and $\mathcal{C% }$ a code of length $m$ over $R$. The dual of $\mathcal{C}$ is defined as $\mathcal{C}^{\perp }=\{\overline{y}\in R^{m}\ |\ \overline{y}\cdot \overline{x}=0$ for all $\overline{x}\in \mathcal{C}\}$, where $\cdot$ denotes the Euclidean inner product $\overline{a}\cdot \overline{b}=\sum a_{i}b_{i}$. A code $\mathcal{C}$ is said to be self-orthogonal if $\mathcal{% C}\subseteq \mathcal{C}^{\perp}$, self-dual if $\mathcal{C}=\mathcal{C}% ^{\perp }$ and iso-dual if $\mathcal{C}$ is permutation equivalent to $% \mathcal{C}^{\perp }$. The code is said to be even if all the codewords have even Lee weight. \begin{definition} A code $\mathcal{C}$ is called FSD if the code $\mathcal{C}$ and its dual $\mathcal{C}^{\perp }$ have the same weight enumerator. A binary FSD even code of length $m$ is called {\it extremal} if $d=2\left\lfloor m/4\right\rfloor +2$ and {\it near-extremal} if $d=2\left\lfloor m/4\right\rfloor $% . \end{definition} The ring $\mathbb{F}_{2}+u\mathbb{F}_{2}\simeq \mathbb{F}_{2}[u]/(u^{2})$ is a commutative ring of characteristic 2 and it contains the binary field as a subring. It has just one non-trivial ideal, given by $(u)$. Type II, Type IV, self-dual codes and cyclic codes over $\mathbb{F}_{2}+u\mathbb{F}_{2}$ have been studied extensively in \cite{Dougherty1}. Through its different extensions that appeared later in the literature, we now refer to this ring as $R_{1}.$ We recall that a linear code $\mathcal{C}$ of length $m$ over the ring $% R_{1} $ is an $R_{1}$-submodule of $R_{1}^{m}$. The elements of $R_1$ are $0,1,u,1+u$ and their Lee weights are defined as $% 0,1,2,1$, respectively. A Gray map $\phi $ is defined as $\phi :R_{1}^{m}\longrightarrow \mathbb{F}% _{2}^{2m}$% \begin{equation} \phi \left( \overline{a}+\overline{b}u\right) =\left( \overline{b},\overline{% a}+\overline{b}\right), \label{Gray} \end{equation}% where $\overline{a},\overline{b}$ in $\mathbb{F}_{2}^{m}$. $\phi $ is a distance preserving, duality preserving isometry from $\left( R_{1}^{m},d_{L}\right) $ to $\left( \mathbb{F}_{2}^{2m},d_{H}\right) $, where $d_{L}$ and $d_{H}$ denote the Lee and Hamming distance in $R_{1}^{m}$ and $\mathbb{F}_{2}^{2m}$, respectively. This means if $\mathcal{C}$ is a linear code over $R_{1}$ with parameters $\left[ m,2^{k},d\right] $, (here $% 2^{k}$ means the number of the codewords), then $\phi \left( \mathcal{C}% \right) $ is a binary linear code of parameters $\left[ 2m,k,d\right] $. Moreover, $\phi (\mathcal{C}^{\perp })=\phi (\mathcal{C})^{\perp }$. Since the Gray map is a duality-preserving map, self-dual codes over $R_1$ have self-dual binary images, but more importantly for our work we have the following theorem, as a consequence of the duality-preserving property of $% \phi$: \begin{theorem} \label{R1image} If $\mathcal{C}$ is an iso-dual code over $R_{1}$ of length $m$, then $\phi (\mathcal{C})$ is an iso-dual binary code of length $% 2m$. If $\mathcal{C}$ is an FSD code over $R_{1}$ of length $m$, then $% \phi (\mathcal{C})$ is an FSD binary code of length $2m$. \end{theorem} \subsection{The ring $S = \mathbb{F}_2[u]/(u^3-1)$ and the MacWilliams Identity} The ring $\mathbb{F}_2[u]/(u^3-1)$, which will be denoted by $S$ here, was used quite effectively in \cite{u3=1} to construct many new binary self-dual codes of length $68$. Thus we recall some properties of the ring $S$ from \cite{u3=1}, leaving some of the details for the work in question. We start by observing that the ring is structurally quite familiar, namely $% S\cong \mathbb{F}_{2}\times \mathbb{F}_{4}$. $S$ is not local and its ideal structure is given by $I_{0}\subset I_{u+u^{2}},I_{1+u+u^{2}}\subset S$, where;% \begin{eqnarray*} I_{1+u+u^{2}} &=&\left( 1+u+u^{2}\right) =\left\{ 0,1+u+u^{2}\right\} , \\ I_{u+u^{2}} &=&\left( 1+u\right) =\left\{ 0,u+u^{2},1+u,1+u^{2}\right\} \text{.} \end{eqnarray*}% The units in $S$ are given by $\left\{ 1,u,u^{2}\right\} $ and the non-units are \[ \left\{ u+u^{2},1+u,1+u^{2},1+u+u^{2}\right\} \text{.} \] A linear code $\mathcal{C}$ of length $m$ over $S$ is an $S$-submodule of $% S^{m}$ and has a generating matrix that is permutation equivalent to \[ G=\left( \begin{array}{cccc} I_{k_{1}} & A & B & C \\ 0 & \left( u+u^{2}\right) I_{k_{2}} & 0 & \left( u+u^{2}\right) D \\ 0 & 0 & \left( 1+u+u^{2}\right) I_{k_{3}} & \left( 1+u+u^{2}\right) E% \end{array}% \right) \] We define a Gray map as follows:% \begin{eqnarray*} \varphi &:&S^{m}\rightarrow \mathbb{F}_{2}^{3m} \\ \overline{a}+\overline{b}u+\overline{c}u^{2} &\mapsto &\left( \overline{a},% \overline{b},\overline{c}\right) . \end{eqnarray*} \begin{definition} The Lee weight of an element $x=a+bu+cu^{2}\in S$ is the Hamming weight of its Gray image, i.e. $w_{L}\left( a+bu+cu^{2}\right) =w_{H}\left( a\right) +w_{H}\left( b\right) +w_{H}\left( c\right) $. \end{definition} Since the Gray map $\varphi$ introduced above does not preserve orthogonality, we cannot immediately have a theorem such as Theorem \ref% {R1image}. In order to have an analogous result, we need to show that the MacWilliams identities for the Lee weight enumerators of codes over $S$ exist. We will quickly establish this by using Wood's results in \cite{Wood}. Let us label the elements in $S= \{ g_1,g_2, \dots, g_{8} \}$ as $% S=\{0,u^2,u,u+u^2,1,1+u^2,1+u,1+u+u^2\}$ \begin{definition} The complete weight enumerator of a linear code $\mathcal{C}$ over $S$ is defined as \begin{equation*} cwe_{\mathcal{C}}(X_{1},X_{2},\dots ,X_{8})=\sum_{\overline{c}\in \mathcal{C}% }{X_{1}^{n_{g_{1}}(\overline{c})}X_{2}^{n_{g_{2}}(\overline{c})}\cdots X_{8}^{n_{g_{8}}(\overline{c})}}, \end{equation*} where $n_{g_i}(\overline{c})$ is the number of appearances of $g_i$ in the vector $\overline{c}$. \end{definition} The complete weight enumerator gives us a lot of information about the code, such as the size of the code, the minimum weight of the code and the weight enumerator of the code for any weight function. Now, since $S$ is a Frobenius ring, the MacWilliams identities for the complete weight enumerator hold. To find the exact identities we define the following character on $S:$ \begin{definition} Define $\chi: S \rightarrow \mathbb{C}^{\times}$ by \begin{equation*} \chi(a+bu+cu^2) = (-1)^{c}. \end{equation*} \end{definition} It is easy to verify that $\chi$ is a non-trivial character when restricted to each non-zero ideal, hence it is a generating character for $S$. Then we make up an $8 \times 8$ matrix $T$, by letting $T(i,j) = \chi(g_ig_j).$ The matrix $T$ is given as follows: \[ T=\left [ \begin{array}{rrrrrrrr} 1 & 1 & 1 & 1 & 1 & 1 & 1 & 1 \\ 1 & 1 & 1 & 1 & -1 & -1 & -1 & -1 \\ 1 & 1 & -1 & -1 & 1 & 1 & -1 & -1 \\ 1 & 1 & -1 & -1 & -1 & -1 & 1 & 1 \\ 1 & -1 & 1 & -1 & 1 & -1 & 1 & -1 \\ 1 & -1 & 1 & -1 & -1 & 1 & -1 & 1 \\ 1 & -1 & -1 & 1 & 1 & -1 & -1 & 1 \\ 1 & -1 & -1 & 1 & -1 & 1 & 1 & -1% \end{array} \right ]. \] The following theorem then follows from \cite{Wood} (Theorem 8.1) quite easily: \begin{theorem} \label{cweMacWilliams} Let $\mathcal{C}$ be a linear code over $S$ of length $m$ and suppose $\mathcal{C}^{\perp }$ is its dual. Then we have \begin{equation*} cwe_{\mathcal{C}^{\perp }}(X_{1},X_{2},\dots ,X_{8})=\frac{1}{|\mathcal{C}|}% cwe_{\mathcal{C}}(T\cdot (X_{1},X_{2},\dots ,X_{8})^{t}), \end{equation*}% where $()^{t}$ denotes the transpose. \end{theorem} Now, in $S$, we have $w_{L}(0)=0$; $w_{L}(1)=w_{L}(u)=w_{L}(u^{2})=1$; $% w_{L}(1+u)=w_{L}(1+u^{2})=w_{L}(u+u^{2})=2$ and $w_{L}(1+u+u^{2})=3$. Thus, identifying the variables corresponding to the elements with the same weight we get the symmetrized weight enumerator, namely, \begin{equation} swe_{\mathcal{C}}(X,Y,Z,W)=cwe_{\mathcal{C}}(X,Y,Y,Z,Y,Z,Z,W). \end{equation} Now, combining Theorem \ref{cweMacWilliams} and the definition of the symmetrized weight enumerator, we obtain the following theorem: \begin{theorem} \label{sweMacWilliams} Let $\mathcal{C}$ be a linear code over $S$ of length $m$ and let $\mathcal{C}^{\perp }$ be its dual. Then we have \begin{equation*} swe_{\mathcal{C}^{\perp }}(X,Y,Z,W)= \end{equation*}% \begin{equation*} \frac{1}{|\mathcal{C}|}swe_{\mathcal{C}}(X+3Y+3Z+W,X+Y-Z-W,X-Y-Z+W,X-3Y+3Z-W). \end{equation*} \end{theorem} We are now ready to obtain the MacWilliams identity for the Lee weight enumerator. Define the Lee weight enumerator for codes over $S$ as the homogeneous polynomial \begin{equation} Lee_{\mathcal{C}}(W,X)=\sum_{\overline{c}\in \mathcal{C}}W^{3n-w_{L}(% \overline{c})}X^{w_{L}(\overline{c})}. \label{Leeenum} \end{equation} Considering the weights that the variables $X,Y,Z,W$ of the symmetrized weight enumerator represent, we easily get the following theorem: \begin{theorem} \label{swelee} Let $\mathcal{C}$ be a linear code over $S$ of length $m$. Then \begin{equation*} Lee_{\mathcal{C}}(W,X)=swe_{\mathcal{C}}(W^{3},W^{2}X,WX^{2},X^{3}). \end{equation*} \end{theorem} Now combining Theorem \ref{sweMacWilliams} and Theorem \ref{swelee} we obtain the following theorem: \begin{theorem} \label{leeMacWilliams} Let $\mathcal{C}$ be a linear code over $S$ of length $m$ and let $\mathcal{C}^{\perp }$ be its dual. With $Lee_{\mathcal{C}}(W,X)$ denoting its Lee weight enumerator as given in (\ref{Leeenum}), we have \begin{equation*} Lee_{\mathcal{C}^{\perp }}(W,X)=\frac{1}{|\mathcal{C}|}Lee_{\mathcal{C}% }(W+X,W-X). \end{equation*} \end{theorem} Note that this is exactly the same identity that is satisfied by the Hamming weight enumerators of binary codes. The existence of the identity gives us the following corollary that is of central importance for our work: \begin{corollary} If $\mathcal{C}$ is an FSD code over $S$ of length $m$, then $\varphi (% \mathcal{C})$ is a binary FSD code of length $3m$. Moreover, the Lee weight enumerator of $\mathcal{C}$ is the same as the Hamming weight enumerator of $% \varphi (\mathcal{C})$. \end{corollary} \section{Constructions\label{constructions}} In this section, we introduce the construction methods for FSD codes over commutative Frobenius rings by block $\lambda $-circulant matrices. Let $R$ be a commutative Frobenius ring and $v=\left( v_{1},v_{2},\ldots ,v_{m}\right) $ an element of the $R$-module $R^{m}$. The $\lambda $% -cyclic shift of $v$ is defined as $\sigma _{\lambda }\left( v\right) =\left( \lambda v_{m},v_{1},v_{2},\ldots ,v_{m-1}\right) $, where $\lambda \in R$. An $m\times m$ square matrix $A$ is called $\lambda $-circulant if every row is a $\lambda $-cyclic shift of the previous one, in other words, $% A $ is in the following form; \[ \left( \begin{array}{ccccc} a_{1} & a_{2} & a_{3} & \cdots & a_{m} \\ \lambda a_{m} & a_{1} & a_{2} & \cdots & a_{n-1} \\ \lambda a_{m-1} & \lambda a_{m} & a_{1} & \cdots & a_{m-2} \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ \lambda a_{2} & \lambda a_{3} & \lambda a_{4} & \cdots & a_{1}% \end{array}% \right) . \]% If $\lambda =1$ the matrix is simply called circulant. Recently, in \cite% {karadeniz}, Karadeniz et al. applied the double and bordered double circulant constructions for a family of commutative Frobenius rings of characteristic $2$. The constructions were generalized to double $\lambda$% -circulant codes in \cite{batoul}. $\lambda$-circulant matrices share properties of circulant matrices such as they commute with each other and they are permutation equivalent to their transposes. \begin{theorem} \label{pure}Let $A_{i}$ be a $\lambda $-circulant matrix of size $m\times m$ for $1\leq i\leq n$ over $R$. Suppose that there is a weight function on $R$ such that $wt(a) = wt(-a)$ for all $a \in R$. Then the code $\mathcal{C}$ generated by \begin{equation} G=\left[ I_{mn}\ \begin{array}{|c} M% \end{array}% \right] \text{ where }M=\left[ \begin{array}{cccc} A_{1} & A_{2} & \cdots & A_{n} \\ A_{n} & A_{1} & \cdots & A_{n-1} \\ \vdots & \vdots & \ddots & \vdots \\ A_{2} & \cdots & A_{n} & A_{1}% \end{array}% \right] \label{matrix} \end{equation}% is an FSD code over $R$. Moreover, $\mathcal{C}$ is iso-dual if $char\left( R\right) =2$. The code is called an $n$-block $\lambda $-circulant~code. \end{theorem} \begin{proof} Let $\mathcal{C}$ be the code generated by $G$. Then the dual $\mathcal{C}% ^{\perp }$ of $\mathcal{C}$ is generated by \begin{equation*} G^{\prime }=\left[ \begin{array}{c|c} M^{t} & -I_{mn}% \end{array}% \right] \text{ .} \end{equation*}% Let $\Pi _{n}$ be the permutation \begin{equation*} \Pi _{n}=\left( 1,n\right) \left( 2,n-1\right) \cdots \left( k-1,n-k+2\right) \left( k,n-k+1\right), \end{equation*}% where $k=\left\lfloor n/2\right\rfloor $. It is easily observed that $A_{i}$ and $A_{i}^{t}$ are permutation equivalent since when $\Pi _{m}$ is applied to rows and then to columns of $A_{i}$, we obtain $A_{i}^{t}$. So the matrix \begin{equation*} M^{\prime }=\left[ \begin{array}{cccc} A_{1}^{t} & A_{2}^{t} & \cdots & A_{n}^{t} \\ A_{n}^{t} & A_{1}^{t} & \cdots & A_{n-1}^{t} \\ \vdots & \vdots & \ddots & \vdots \\ A_{2}^{t} & \cdots & A_{n}^{t} & A_{1}^{t}% \end{array}% \right] \end{equation*}% is obtained by applying $\Pi _{m}$ to each block of $M$. Then $M^{t}$ is obtained by applying $\Pi _{n}$ to $M^{\prime }$ block-wise. Let $\mathcal{C}% ^{\prime \prime }$ be the code generated by $G^{\prime \prime }=\left[ \begin{array}{c|c} M^{T} & I_{mn}% \end{array}% \right] $. The matrices $G$ and $G^{\prime \prime }$ are permutation equivalent and so are the codes they generate. On the other hand, $\mathcal{C}% ^{\perp }$ and $\mathcal{C}^{\prime \prime }$ have the same weight enumerator since $wt(a) = wt(-a)$ for each $a \in R$. Hence $\mathcal{C}$ is formally self-dual. Moreover, $\mathcal{C} $ is isodual if $R$ is a ring of characteristic $2$ since $\mathcal{C}% ^{\perp }=\mathcal{C}^{\prime \prime }$. \end{proof} \begin{theorem} \label{bordered}Let $M$ and $R$ be as in (\ref{matrix}); $a,b\in R$ and let $\boldsymbol{b% }$ denote the all-$b$ row of length $mn$. Then the code generated by \begin{equation*} \left[ I_{mn+1}\ \begin{array}{|cc} a & \boldsymbol{b} \\ \boldsymbol{b}^{t} & M% \end{array}% \right] \end{equation*}% is an FSD code over $R$. The code is iso-dual when $char\left( R\right) =2$ and it is called a bordered $n$-block $\lambda $-circulant code. \end{theorem} \begin{proof} Let $\mathcal{C}$ be the code generated by $G$. Then the dual $\mathcal{C}% ^{\perp }$ of $\mathcal{C}$ is generated by \begin{equation*} G' = \left[ \begin{array}{cc|} a & \boldsymbol{b} \\ \boldsymbol{b}^{t} & M^t% \end{array} -I_{mn+1}\ % \right] \end{equation*}% Applying the same permutations as in the proof of Theorem \ref{pure}, we see that $C$ is equivalent to $C''$, which is generated by \[G'' = \left[ \begin{array}{cc|} a & \boldsymbol{b} \\ \boldsymbol{b}^{t} & M^t% \end{array} I_{mn+1}\ % \right]. \] On the other hand, $\mathcal{C}% ^{\perp }$ and $\mathcal{C}^{\prime \prime }$ have the same weight enumerator since $wt(a) = wt(-a)$ for each $a \in R$. Hence $\mathcal{C}$ is formally self-dual. Moreover, $\mathcal{C} $ is isodual if $R$ is a ring of characteristic $2$ since $\mathcal{C}% ^{\perp }=\mathcal{C}^{\prime \prime }$. \end{proof} The constructions given in theorems \ref{pure} and \ref{bordered} can be generalized further by considering $\lambda _{0}$-circulating blocks of matrices. Such a code is called $\lambda _{0}$-$n$-block $\lambda $% -circulant code. We illustrate this construction in the following example; \begin{example} \label{lambda}Let $R=R_1$ and $r_{1}=\left( 1+u,u\right) $, $r_{2}=\left( 1,1+u\right) $% , $r_{3}=\left( 1+u,1+u\right) $ and $r_{4}=\left( 0,u\right) $ $\ $be the first rows of circulant matrices $A_{1}$, $A_{2}$, $A_{3}$ and $A_{4}$, respectively. Let $\mathcal{C}$ be the $\lambda _{0}$ block circulant code over $R_{1}$ generated by \begin{equation*} G=\left[ I_{8}\ \begin{array}{|c} \begin{array}{cccc} A_{1} & A_{2} & A_{3} & A_{4} \\ \lambda _{0}A_{4} & A_{1} & A_{2} & A_{3} \\ \lambda _{0}A_{3} & \lambda _{0}A_{4} & A_{1} & A_{2} \\ \lambda _{0}A_{2} & \lambda _{0}A_{3} & \lambda _{0}A_{4} & A_{1}% \end{array}% \end{array}% \right], \end{equation*}% where $\lambda _{0}=1+u$. Then $\mathcal{C}$ is an iso-dual code. The binary code $\varphi \left( \mathcal{C}\right) $ is a binary even FSD code of parameters $\left[ 32,16,8\right]$, which is optimal. A partial weight distribution of the code is given by $1+364z^{8}+2048z^{10}+6720z^{12}+\cdots $. Even FSD codes of these parameters have been studied extensively in \cite{gulliver2}. \end{example} More examples for the generalization of Theorem \ref{bordered} are given in Tables \ref{tab:44},\ref{tab:40} and \ref{tab:54}. \section{Computational results} In this section, the methods from Section \ref{constructions} are applied to the rings $\mathbb{F}_{2}$, $R_{1}$ and $S$. FSD codes with parameters better than self-dual codes are tabulated. New near-extremal FSD even codes are constructed for lengths $36$, $38$ and $44$. For the rest of the paper $(.)^{\ast }$ is used to denote that the code is optimal as a linear code and $(.)^{b}$ is used to denote that the minimum distance of the code is the distance of the best known linear code with respect to the online database \cite{grassl}. \begin{remark} In the tables that follow, $n$ denotes the number of blocks, $m~$denotes the length of each block, $a$ and $b$ denote the borders of the block circulant matrix (In the case of bordered double circulant constructions given in Theorem \ref{bordered}), $A_{i} $ is a $\lambda $-circulant matrix for each $1\leq i$ $\leq n$ and $\lambda _{0}$ is as in Example \ref{lambda}. The order of the automorphism group of $\mathcal{C}$ is denoted by $\left\vert Aut(\mathcal{C})\right\vert $. In order to fit the upcoming tables we denote the element $1+u$ of $R_{1}$ as $3$ and $% 1+u+u^{2}$ of $S$ as $7$. All computations were run on an average PC, using the magma algebra system (\cite{Magma}). \end{remark} \begin{example} Let $A_{1}$ and $A_{2}$ be $2\times 2$ $\left( 1+u\right) $-circulant matrices over $R_{1}$ with first rows $r_{1}=\left( 1,1+u\right) $ and $% r_{2}=\left( 1+u,u\right) $. Then the bordered $2$-block $\left( 1+u\right) $% -circulant code generated by \begin{equation*} G=\left[ I_{5}\ \begin{array}{|c} \begin{array}{ccccc} u & 1 & 1 & 1 & 1 \\ 1 & 1 & 1+u & 1+u & u \\ 1 & 1 & 1 & u & 1+u \\ 1 & 1+u & u & 1 & 1+u \\ 1 & u & 1+u & 1 & 1% \end{array}% \end{array}% \right] \end{equation*}% is an iso-dual code over $R_{1}$. The binary image of the code is an odd binary FSD $\left[ 20,10,6\right]^{\ast }$ code that is optimal as a linear code. A partial weight distribution of the code is $% 1+40z^{6}+160z^{7}+130z^{8}+\cdots $ and its automorphism group has order $% 2^{8}\times 3\times 5$. \end{example} \begin{example} Let $\mathcal{C}$ be the pure double circulant code with the first row \begin{equation*} r=\left( 11000100101111110100111\right). \end{equation*}% Then $\mathcal{C}$ is an iso-dual code by Theorem \ref{pure}. The code $\mathcal{C}$ is an optimal linear $\left[ 46,23,11\right]^{\ast }$ code and a partial weight distribution of $\mathcal{C}$ is given by \begin{equation*} 1+3312z^{11}+9660z^{12}+121440z^{15}+235290z^{16}+\cdots . \end{equation*}% With respect to the online database of linear codes, \cite{grassl}, a binary linear $\left[ 46,23,11\right]^{\ast }$ code is constructed as a shortened puncturing of the unique Type II $\left[ 48,24,12\right]^{\ast }$. The code $\mathcal{C}$ is equivalent to the one kept in \cite{grassl}. \end{example} \subsection{Binary even FSD $\left[ 72,36,14\right]$ codes} The existence of a Type II binary self-dual $\left[ 72,36,16\right]$ code is an open problem that has generated a lot of interest in the last four decades. Equivalent to the same problem is the existence of a binary Type I self-dual $\left[ 72,36,14\right]$ code. The best known binary self-dual codes of length 72 have minimum distance 12, see \cite{kovic,gurel}. The first examples of binary $\left[ 72,36,14\right]$ FSD even codes were constructed in \cite{karadeniz}. In this section, we construct $10$ binary FSD even $\left[ 72,36,14\right]$ codes, which are different than the codes given in \cite{karadeniz}. The results are given in Table \ref{tab:72}. \begin{example} Let $\lambda =1+u$, $n=1$ and let $\mathcal{C}$ be the double $\lambda $% -circulant code with the first row $% r_{1}=(u,0,u,u,0,u,0,0,0,u+1,0,1,u+1,u+1,1,0,1,1,)$ over $R_{1}$. Then $% \mathcal{C}$ is an iso-dual code. The Gray image $\varphi \left( \mathcal{C}% \right) $ of $\mathcal{C}$ is a binary FSD even code of parameters $\left[ 72,36,14% \right]$. A partial weight enumerator of the code is given by $1+9036z^{14}+121959z^{16}\cdots $. \end{example} \begin{table}[h!] \begin{center} \begin{tabular}{|l|l|l|} \hline $n$ & $r_{1},\ldots ,r_{n}$ & Partial weight distribution \\ \hline 4 & $000110011,111101100,101111100,100011011$ & $1+9144z^{14}+120897z^{16}+% \cdots $ \\ \hline 4 & $000100101,011101011,101000000,110110000$ & $1+8748z^{14}+123525z^{16}+% \cdots $ \\ \hline 4 & $101110001,000100010,001010011,101011101$ & $1+8748z^{14}+123660z^{16}+% \cdots $ \\ \hline 4 & $000100101,011101011,000101111,000100110$ & $1+9072z^{14}+121383z^{16}+% \cdots $ \\ \hline 4 & $000100110,001010101,101000000,110110000$ & $1+8640z^{14}+124479z^{16}+% \cdots $ \\ \hline 3 & $111100111010,110101101100,111011001011$ & $1+8640z^{14}+125433z^{16}+% \cdots $ \\ \hline 3 & $000111110111,001101011000,100110001101$ & $1+9288z^{14}+120537z^{16}+% \cdots $ \\ \hline 1 & $100100010000010111111111000111010000$ & $1+8820z^{14}+122841z^{16}+% \cdots $ \\ \hline 1 & $100 1 0 0 0 1 0 0 0 0 0 1 0 1 1 1 0 0 0 0 1 1 0 0 0 1 0 0 1 1 1 0 1 1$ & $1+8748z^{14}+123831z^{16}+\cdots $ \\ \hline \end{tabular}% \vspace*{0.2cm}\caption{\rm\em Binary FSD $\left[ 72,36,14\right]$ codes as $n$-block circulant codes over $\mathbb{F}_{2}$} \label{tab:72} \end{center} \end{table} \subsection{New near extremal binary FSD even codes} Extremal binary FSD even codes do not exist for many lengths and the non-existence of near-extremal binary FSD even codes have been studied in \cite{han}. In \cite{gulliver}, possible weight enumerators of near-extremal binary FSD even codes of lengths $16$ to $46$ has been determined. In this section, we construct new binary near-extremal FSD even codes for lengths $36 $, $38$ and $44$, where possible weight enumerators are;% \begin{eqnarray*} W_{36} &=&1+\left( 225+\alpha \right) y^{8}+\left( 2016-6\alpha \right) y^{10}+\cdots , \\ W_{38} &=&1+\left( 171+\alpha \right) y^{8}+\left( 1862-5\alpha \right) y^{10}+\cdots , \\ W_{44} &=&1+\left( 1320+\alpha \right) z^{10}+\left( 10461-8\alpha \right) z^{12}+\cdots . \end{eqnarray*} The existence of near-extremal binary FSD even codes is known for weight enumerators; $\alpha =$0, 9, 13, 18, 27, 36, 45, 47, 63, 64, 72, $-$4, $-$9, $-$21, $-$27, $-$36, $-$38, $-$45, $-$54, $-$55 and $-$72 for $W_{36}$; $% \alpha =$0, 22, 31, 44, 66, 176, $-19$, $-22$, $-44$, $-53$, $-66$, $-176$ for $W_{44}$. In this work, we were able to construct binary near-extremal FSD even codes with weight enumerators; $\alpha =$3, 6, 12, 15, 21, 24, 30, 33, 39, 42, 48, 60, 96, 144, $-3$, $-$6, $-$12, $-$15, $-$18, $-$24, $-$30, $-$33, $-$42, $-$% 144 in $W_{36}$; $\alpha =$10, 30, 50, 80, $-$10, $-$11, $-$30 and $-$70 in $% W_{44}$. \begin{example} For $n=1$ and $\lambda =1$, let $\mathcal{C}$ be the double circulant code over $S$ of length $12$ with the first row of $A_{1}$ as \ $r_{1}=\left( 1,u^{2},1+u,u+u^{2},1,1+u^{2}\right) $. Then $\mathcal{C}$ is iso-dual by Theorem \ref{pure}. The Gray image of $\mathcal{C}$ is a binary near-extremal FSD even code with weight enumerator $\alpha =144$ in $W_{36}$. The weight enumerator of the code is different than the ones in \cite{gulliver}. \end{example} More new near extremal FSD even codes are constructed in Table \ref{tab:36}. \begin{table}[htbp] \begin{center} \begin{tabular}{|c|c|c|c||c|} \hline $\lambda $ & $r_1$ & $a,b$ & $\left\vert Aut(\mathcal{C})\right\vert $ & $\alpha $ \\ \hline $1$ & $\left( u,1+u^{2},0,7,1+u^{2}\right) $ & $1+u^{2},u$ & 30 & $3$ \\ \hline $1$ & $\left( 1+u^{2},1+u^{2},u,0,7\right) $ & $1+u^{2},1$ & 15 & $6$ \\ \hline $u$ & $\left( 0,1+u^{2},1,u+u^{2},u\right) $ & $u+u^{2},u$ & 3 & $12$ \\ \hline $u$ & $\left( 1+u,7,0,1,0\right) $ & $1+u,1$ & 3 & $15$ \\ \hline $u$ & $\left( u,0,u+u^{2},u^{2},1+u\right) $ & $1+u^{2},1$ & 3 & $21$ \\ \hline $u$ & $\left( 1+u^{2},u^{2},u,u^{2},1\right) $ & $u+u^{2},1$ & 3 & $24$ \\ \hline $u$ & $\left( 1,1,0,u^{2},7\right) $ & $1+u^{2},u^{2}$ & 3 & $30$ \\ \hline $1$ & $\left( u^{2},0,1+u^{2},1,1+u\right) $ & $1+u^{2},1$ & 15 & $33$ \\ \hline $u$ & $\left( 1+u,7,7,0,1+u^{2}\right) $ & $1+u,1$ & 3 & $39$ \\ \hline $u$ & $\left( 1+u^{2},7,1,u^{2},u^{2}\right) $ & $1+u,u$ & 3 & $42$ \\ \hline $1$ & $\left( u,u^{2},u,1+u^{2},u\right) $ & $u+u^{2},u$ & 15 & $48$ \\ \hline $u$ & $\left( 7,u^{2},1+u^{2},7,1\right) $ & $1+u,u$ & 3 & $60$ \\ \hline $1$ & $\left( u+u^{2},u+u^{2},1,1,1+u^{2}\right) $ & $1+u,u^{2}$ & 60 & $96$ \\ \hline $u$ & $\left( 7,u,u+u^{2},0,0\right) $ & $1+u^{2},u$ & 3 & $-3$ \\ \hline $u$ & $\left( 0,u^{2},1,u,7\right) $ & $1+u^{2},1$ & 3 & $-6$ \\ \hline $1$ & $\left( 0,7,1+u^{2},0,1\right) $ & $u+u^{2},u$ & 15 & $-12$ \\ \hline $u$ & $\left( 1+u,u,1+u,0,u^{2}\right) $ & $1+u^{2},u$ & 3 & $-15$ \\ \hline $u$ & $\left( 7,u,1+u,1,7\right) $ & $1+u,u^{2}$ & 3 & $-18$ \\ \hline $1$ & $\left( 1+u^{2},u,1,1+u^{2},0\right) $ & $1+u,u$ & 15 & $-24$ \\ \hline $u$ & $\left( 7,0,u,u,1\right) $ & $1+u^{2},u$ & 3 & $-30$ \\ \hline $u$ & $\left( 1+u^{2},u+u^{2},7,u,0\right) $ & $1+u,u$ & 3 & $-33$ \\ \hline $1$ & $\left( 1+u,1+u,u,u^{2},0\right) $ & $1+u^{2},u^{2}$ & 15 & $-42$ \\ \hline $1$ & $\left( 1+u^{2},1+u^{2},u,1+u,u\right) $ & $1+u^{2},u^{2}$ & 1440 & $% -144$ \\ \hline \end{tabular}% \vspace*{0.2cm} \caption{\rm\em 24 new near extremal FSD even codes of length $36$ over $S$ by Theorem \ref{bordered} }\label{tab:36} \end{center} \end{table} For a list of known near-extremal FSD even codes of length $38$ we refer to \cite{gulliver}. In the following example we construct a code with a new weight enumerator by applying bordered 2-block circulant construction on $% \mathbb{F}_{2}$. \begin{example} Let $\mathcal{C}$ be the bordered 2-block circulant code with $r_{1}=\left( 111100111\right) ,$ $r_{2}=\left( 101011100\right) $ and $a=1=b$ over $% \mathbb{F}_{2}$. The code $\mathcal{C}$ is an even code and it is FSD by Theorem \ref{bordered}. $\left\vert Aut\left( \mathcal{C}\right) \right\vert =2\times 3^{2}$ and $\mathcal{C}$ has a weight enumerator with $\alpha =-9$ in $W_{38}$. \end{example} \begin{example} Let $\mathcal{C}$ be the binary bordered 3-block circulant code with $% r_{1}=\left( 1110101\right) ,$ $r_{2}=\left( 1100100\right) ,$ $r_{3}=\left( 0010100\right) $, $a=0$ and $b=1$. Then $\mathcal{C}$ is a near-extremal FSD even code with weight enumerator $\alpha =-11$ in $W_{44}$. The existence of an FSD code with this weight enumerator was unknown. The automorphism group of the code has order $3\times 7$. \end{example} We were able to construct seven near extremal binary FSD even codes of length $44$ with new weight enumerators as Gray images of FSD codes over $% R_{1}$, which are listed in Table \ref{tab:44}. \begin{table}[htbp] \begin{center} \begin{tabular}{|c|c|c|c|c|c||c|} \hline $n$ & $\lambda _{0}$ & $\lambda $ & $r_{1},\ldots ,r_{n}$ & $a,b$ & $% \left\vert Aut(\mathcal{C})\right\vert $ & $\alpha $ \\ \hline $2$ & $3$ & $1$ & $\left( u01u3\right) ,\left( 03u01\right) $ & $3,3$ & $% 2\times 5$ & $30$ \\ \hline $2$ & $3$ & $1$ & $\left( 0330u\right) ,\left( 33101\right) $ & $3,3$ & $% 2\times 5$ & $50$ \\ \hline $5$ & $1$ & $1$ & $\left( 13\right) ,\left( 10\right) ,\left( uu\right) ,\left( 3u\right) \left( 33\right) $ & $1,1$ & $2^{2}5$ & $10$ \\ \hline $5$ & $1$ & $1$ & $\left( 30\right) ,\left( 11\right) ,\left( 1u\right) ,\left( u3\right) \left( u1\right) $ & $1,1$ & $2^{2}5$ & $80$ \\ \hline $5$ & $1$ & $1$ & $\left( u0\right) ,\left( u3\right) ,\left( 03\right) ,\left( 11\right) \left( 13\right) $ & $1,1$ & $2^{2}5$ & $-10$ \\ \hline $5$ & $1$ & $1$ & $\left( 3u\right) ,\left( 0u\right) ,\left( 1u\right) ,\left( uu\right) \left( 13\right) $ & $1,1$ & $2^{2}5$ & $-30$ \\ \hline $5$ & $1$ & $1$ & $\left( 3u\right) ,\left( 03\right) ,\left( 33\right) ,\left( 13\right) \left( u0\right) $ & $1,1$ & $2^{2}5$ & $-70$ \\ \hline \end{tabular}% \vspace*{0.2cm}\caption{\rm\em Optimal $\left[ 44,22,10\right]$ FSD even codes as images of bordered $n$-block circulant codes over $R_{1}$}\label{tab:44} \end{center} \end{table} \begin{remark} Throughout the paper the binary FSD codes that have been listed have better minimum distances than the best known self-dual codes, except for the ones listed in Tables \ref{tab:36} and \ref{tab:76}, which have the same distance as the corresponding extremal self-dual code. \end{remark} Theorem \ref{pure} is applied over $\mathbb{F}_{2}$ in Table \ref{tab:70}. Its generalization is applied over $R_1$ and $S$ in Table \ref{tab:40} and in Table \ref{tab:54}, respectively. Theorem \ref{bordered} is applied in Tables \ref{tab:34}, \ref{tab:58}, \ref{tab:74} and \ref{tab:76}. \begin{table}[H] \begin{center} \begin{tabular}{||c|c|c|c|c||} \hline $\lambda _{0}$ & $\lambda$ & $r_{1}$ & $r_{2}$ & Partial weight distribution \\ \hline $1$ & $1$ & $\left( 01113\right) $ & $\left( 013u0\right) $ & $% 1+280z^{9}+1052z^{10}+\cdots $ \\ \hline $1$ & $1$ & $\left( 1u0uu\right) $ & $\left( 1330u\right) $ & $% 1+320z^{9}+992z^{10}+\cdots $ \\ \hline $1$ & $1$ & $\left( u1u03\right) $ & $\left( u303u\right) $ & $% 1+340z^{9}+972z^{10}+\cdots $ \\ \hline $3$ & $1$ & $\left( 1013u\right) $ & $\left( 3u303\right) $ & $% 1+260z^{9}+1030z^{10}+\cdots $ \\ \hline $3$ & $1$ & $\left( 11u01\right) $ & $\left( 33100\right) $ & $% 1+280z^{9}+1010z^{10}+\cdots $ \\ \hline $3$ & $1$ & $\left( 10u33\right) $ & $\left( u1033\right) $ & $% 1+300z^{9}+1020z^{10}+\cdots $ \\ \hline $3$ & $1$ & $\left( u3u00\right) $ & $\left( 110u1\right) $ & $% 1+320z^{9}+1000z^{10}+\cdots $ \\ \hline $3$ & $1$ & $\left( u3311\right) $ & $\left( 00013\right) $ & $% 1+320z^{9}+982z^{10}+\cdots $ \\ \hline $3$ & $1$ & $\left( 0u31u\right) $ & $\left( 1111u\right) $ & $% 1+360z^{9}+922z^{10}+\cdots $ \\ \hline $3$ & $3$ & $\left( 01133\right) $ & $\left( u31uu\right) $ & $% 1+340z^{9}+982z^{10}+\cdots $ \\ \hline $3$ & $3$ & $\left( 03u30\right) $ & $\left( u1331\right) $ & $% 1+240z^{9}+1166z^{10}+\cdots $ \\ \hline $3$ & $3$ & $\left( 00u33\right) $ & $\left( 3u03u\right) $ & $% 1+300z^{9}+1022z^{10}+\cdots $ \\ \hline $3$ & $3$ & $\left( 1030u\right) $ & $\left( 01311\right) $ & $% 1+320z^{9}+1012z^{10}+\cdots $ \\ \hline \end{tabular}% \vspace*{0.2cm} \caption{\rm\em Binary $\left[ 40,20,9\right]^{b}$ FSD codes as Gray images of $2$-block circulant codes over $R_{1}$} \label{tab:40} \end{center} \end{table} \begin{table}[htbp] \begin{center} \begin{tabular}{|l|l|l|l|l|l|} \hline $n$ & $m$ & $r_{1},\ldots ,r_{n}$ & $a,b$ & $\left\vert Aut(\mathcal{C})\right\vert $ & Partial weight distribution \\ \hline $2$ & $8$ & $11000001,11010100$ & $0,1$ & $2^{6}$ & $1+172z^{8}+352z^{9}+% \cdots $ \\ \hline $2$ & $8$ & $10001010,10110001$ & $0,1$ & $2^{7}$ & $1+204z^{8}+384z^{9}+% \cdots $ \\ \hline $2$ & $8$ & $10001101,00011001$ & $1,1$ & $2^{6}$ & $1+220z^{8}+384z^{9}+% \cdots $ \\ \hline $2$ & $8$ & $10001011,11101111$ & $0,1$ & $2^{6}$ & $1+236z^{8}+416z^{9}+% \cdots $ \\ \hline $2$ & $8$ & $10111110,10000101$ & $0,1$ & $2^{7}$ & $1+268z^{8}+448z^{9}+% \cdots $ \\ \hline $4$ & $4$ & $1110,1000,0110,0111$ & $0,1$ & $2^{9}3$ & $1+140z^{8}+320z^{9}+% \cdots $ \\ \hline $4$ & $4$ & $1001,1000,1000,0111$ & $0,1$ & $2^{9}3$ & $1+204z^{8}+512z^{9}+% \cdots $ \\ \hline $4$ & $4$ & $1010,0110,0001,1100$ & $0,1$ & $2^{9}5^{2}3$ & $% 1+332z^{8}+640z^{9}+\cdots $ \\ \hline $1$ & $16$ & $1000101110011100$ & $0,1$ & $2^{4}$ & $1+192z^{8}+298z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $0111001100101101$ & $0,1$ & $2^{4}$ & $1+200z^{8}+380z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $1100101001101001$ & $0,1$ & $2^{4}$ & $1+208z^{8}+282z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $0001111011101111$ & $0,1$ & $2^{5}$ & $1+208z^{8}+388z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $1110011000111110$ & $0,1$ & $2^{4}$ & $1+210z^{8}+284z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $1000101110110000$ & $0,1$ & $2^{4}$ & $1+212z^{8}+392z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $1100001111101001$ & $1,1$ & $2^{7}$ & $1+220z^{8}+384z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $0110111010000110$ & $0,1$ & $2^{4}$ & $1+224z^{8}+266z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $1010111011110010$ & $0,1$ & $2^{4}$ & $1+224z^{8}+282z^{9}+% \cdots $ \\ \hline $1$ & $16$ & $0101010000011001$ & $0,1$ & $2^{4}$ & $1+240z^{8}+266z^{9}+% \cdots $ \\ \hline $1$ & $17$ & $10101110111110110$ & pure & $17$ & $1+153z^{8}+527z^{9}+\cdots $ \\ \hline $1$ & $17$ & $01111110100011111$ & pure & $17$ & $1+170z^{8}+527z^{9}+\cdots $ \\ \hline $1$ & $17$ & $11101000001001011$ & pure & $17$ & $1+187z^{8}+459z^{9}+\cdots $ \\ \hline $1$ & $17$ & $11001111111100110$ & pure & $2\times 17$ & $% 1+187z^{8}+493z^{9}+\cdots $ \\ \hline $1$ & $17$ & $11011101000100100$ & pure & $17$ & $1+204z^{8}+442z^{9}+\cdots $ \\ \hline $1$ & $17$ & $11111010111011001$ & pure & $17$ & $1+221z^{8}+408z^{9}+\cdots $ \\ \hline $1$ & $17$ & $10011011011000010$ & pure & $17$ & $1+255z^{8}+816z^{9}+\cdots $ \\ \hline \end{tabular}% \vspace*{0.2cm} \caption{\rm\em Optimal binary linear $\left[ 34,17,8\right]$ codes as odd FSD codes over $\mathbb{F}_{2}$} \label{tab:34} \end{center} \end{table} \begin{table}[htbp] \begin{center} \resizebox{1\textwidth}{!}{\begin{tabular}{|l|l|l|l|l|} \hline $n$ & $\lambda _{0}$ & $\lambda$ & $r_{1},\ldots ,r_{n}$ & Partial weight distribution \\ \hline 3 & $u^{2}$ & $1$ & $\left( 1+u,u,u\right) ,\left( u^{2},u+u^{2},0\right) ,\left( 1+u,1,u+u^{2}\right) $ & $1+729z^{11}+3096z^{12}+\cdots $ \\ \hline 3 & $u^{2}$ & $1$ & $\left( 1+u,1+u^{2},0\right) ,\left( 1+u,0,1\right) ,\left( u^{2},1,7\right) $ & $1+864z^{11}+3123z^{12}+\cdots $ \\ \hline 3 & 1 & $1$ & $\left( 1,1,0\right) ,\left( 1,0,1+u^{2}\right) ,\left( u^{2},u+u^{2},1+u^{2}\right) $ & $1+1080z^{11}+5706z^{12}+\cdots $ \\ \hline 3 & $1$ & $1$ & $\left( 0,7,1+u\right) ,\left( 1+u,1+u^{2},1+u^{2}\right) ,\left( u,1,u\right) $ & $1+1404z^{11}+5031z^{12}+\cdots $ \\ \hline 1 & - & $u$ & $\left( 1+u,1+u^{2},1+u^{2},1,u+u^{2},7,u+u^{2},u,1\right) $ & $1+621z^{11}+2970z^{12}+\cdots $ \\ \hline 1 & - & $u$ & $\left( u+u^{2},u^{2},1,u+u^{2},u,u+u^{2},1,u+u^{2},u+u^{2}\right) $ & $% 1+648z^{11}+3051z^{12}+\cdots $ \\ \hline 1 & - & $u$ & $\left( 1,1,u,u,u,1+u^{2},1+u+u^{2},1+u^{2},0\right) $ & $% 1+729z^{11}+3051z^{12}+\cdots $ \\ \hline 1 & - & $u$ & $\left( u^{2},u+u^{2},u+u^{2},7,1+u,u+u^{2},1+u,7,u\right) $ & $1+864z^{11}+2916z^{12}+\cdots $ \\ \hline 1 & - & $u$ & $\left( 1+u,1+u^{2},1+u^{2},1,u+u^{2},1+u,u,1,1\right) $ & $% 1+918z^{11}+3132z^{12}+\cdots $ \\ \hline 1 & - & $u^{2}$ & $\left( 1,1+u^{2},1,7,u+u^{2},u,1,7,1+u\right) $ & $% 1+891z^{11}+2889z^{12}+\cdots $ \\ \hline \end{tabular}} \vspace*{0.2cm} \caption{\rm\em Binary FSD $\left[ 54,27,11\right]^{b}$ codes as Gray images of FSD codes over $S$} \label{tab:54} \end{center} \end{table} \begin{table}[htbp] \begin{center} \begin{tabular}{|l|l|l|l|l|} \hline $r_{1}$ & $r_{2}$ & $r_{3}$ & $r_{4}$ & Partial weight distribution \\ \hline $1111001$ & $1011001$ & $0111111$ & $1000101$ & $1+3290z^{12}+40565z^{14}+% \cdots $ \\ \hline $1011110$ & $0000110$ & $0011011$ & $1111111$ & $1+3388z^{12}+39879z^{14}+% \cdots $ \\ \hline $1001111$ & $0111011$ & $0011000$ & $0101000$ & $1+3395z^{12}+39830z^{14}+% \cdots $ \\ \hline $0000100$ & $0100100$ & $1000101$ & $1001110$ & $1+3402z^{12}+39783z^{14}+% \cdots $ \\ \hline $0001001$ & $1101111$ & $1000101$ & $1110011$ & $1+3437z^{12}+39536z^{14}+% \cdots $ \\ \hline $0111111$ & $0100000$ & $0111001$ & $1001111$ & $1+3437z^{12}+39482z^{14}+% \cdots $ \\ \hline $0011100$ & $1010000$ & $0011111$ & $0011110$ & $1+3500z^{12}+39097z^{14}+% \cdots $ \\ \hline $1001100$ & $0110000$ & $0000101$ & $1001100$ & $1+3500z^{12}+39095z^{14}+% \cdots $ \\ \hline $1000110$ & $0011110$ & $1011111$ & $1001100$ & $1+3507z^{12}+38990z^{14}+% \cdots $ \\ \hline $1111001$ & $1010000$ & $1010100$ & $0011110$ & $1+3542z^{12}+38803z^{14}+% \cdots $ \\ \hline $1001011$ & $1010010$ & $1000111$ & $0111011$ & $1+3563z^{12}+38600z^{14}+% \cdots $ \\ \hline $1110011$ & $1001110$ & $0000110$ & $0000001$ & $1+3591z^{12}+38402z^{14}+% \cdots $ \\ \hline $0101100$ & $0101000$ & $1111010$ & $1001011$ & $1+3598z^{12}+38411z^{14}+% \cdots $ \\ \hline $0011000$ & $0100110$ & $0110110$ & $0111000$ & $1+3619z^{12}+38208z^{14}+% \cdots $ \\ \hline $1100001$ & $0101000$ & $1010001$ & $1000001$ & $1+3640z^{12}+38061z^{14}+% \cdots $ \\ \hline $1101110$ & $1000100$ & $0101100$ & $0101101$ & $1+3668z^{12}+37921z^{14}+% \cdots $ \\ \hline $0101100$ & $0010011$ & $0111001$ & $0011110$ & $1+3703z^{12}+37674z^{14}+% \cdots $ \\ \hline $1110101$ & $1100101$ & $0110111$ & $0111010$ & $1+3745z^{12}+37382z^{14}+% \cdots $ \\ \hline $0001010$ & $1000000$ & $1000011$ & $1101010$ & $1+3766z^{12}+37177z^{14}+% \cdots $ \\ \hline $1010111$ & $1010001$ & $1010011$ & $0110110$ & $1+3850z^{12}+36589z^{14}+% \cdots $ \\ \hline \end{tabular}% \vspace*{0.2cm} \caption{\rm\em Binary $\left[ 58,29,12\right]^{b}$ FSD codes as bordered $4$-block circulant codes over $\mathbb{F}_{2}$ with $a=1=b$} \label{tab:58} \end{center} \end{table} \begin{table}[htbp] \begin{center} \begin{tabular}{|l|l|l|l|l|l|} \hline $r_{1}$ & $r_{2}$ & $r_{3}$ & $r_{4}$ & $r_{5}$ & Partial weight distribution \\ \hline $1000111$ & $0101010$ & $0100110$ & $0100001$ & $0111001$ & $% 1\!+\!910z^{13}\!+\!5880z^{14}\!+\!\cdots $ \\ \hline $1110011$ & $0001000$ & $1000110$ & $0100110$ & $0000011$ & $% 1\!+\!1120z^{13}\!+\!5740z^{14}\!+\!\cdots $ \\ \hline $1101101$ & $0101000$ & $0011010$ & $0100010$ & $1100110$ & $% 1\!+\!1155z^{13}\!+\!6405z^{14}\!+\!\cdots $ \\ \hline $0110111$ & $1001110$ & $0001000$ & $0001000$ & $0110100$ & $% 1\!+\!1295z^{13}\!+\!5810z^{14}\!+\!\cdots $ \\ \hline $0010101$ & $1111000$ & $1001100$ & $1111001$ & $1001001$ & $% 1\!+\!1295z^{13}\!+\!5635z^{14}\!+\!\cdots $ \\ \hline $1100100$ & $1110010$ & $0001110$ & $0110001$ & $1010111$ & $% 1\!+\!1400z^{13}\!+\!6020z^{14}\!+\!\cdots $ \\ \hline $0111100$ & $0111101$ & $1110100$ & $1010101$ & $1001100$ & $% 1\!+\!1435z^{13}\!+\!5565z^{14}\!+\!\cdots $ \\ \hline $1010011$ & $0010010$ & $1001100$ & $1100010$ & $0010111$ & $% 1\!+\!1505z^{13}\!+\!5670z^{14}\!+\!\cdots $ \\ \hline $1001111$ & $0010110$ & $1000000$ & $0100110$ & $0101110$ & $% 1\!+\!1505z^{13}\!+\!6125z^{14}\!+\!\cdots $ \\ \hline $0010111$ & $1001111$ & $0011011$ & $0100110$ & $1111000$ & $% 1\!+\!1540z^{13}\!+\!5530z^{14}\!+\!\cdots $ \\ \hline \end{tabular}% \vspace*{0.2cm} \caption{\rm\em Binary $\left[ 70,35,13\right]$ FSD codes as 5-block circulant codes over $\mathbb{F}_{2}$} \label{tab:70} \end{center} \end{table} \begin{table}[h!] \begin{center} \resizebox{1\textwidth}{!}{\begin{tabular}{|l|l|l|l|} \hline $n$ & $m$ & $r_{1},\ldots ,r_{n}$ & Partial weight distribution \\ \hline $4$ & $9$ & $101000100,101111100,010001001,011100100$ & $% 1+6681z^{14}+98703z^{16}+\cdots $ \\ \hline $4$ & $9$ & $100001110,110100110,000011001,001000010$ & $% 1+6714z^{14}+98538z^{16}+\cdots $ \\ \hline $4$ & $9$ & $100111101,111010111,101000000,101100110$ & $% 1+6723z^{14}+98673z^{16}+\cdots $ \\ \hline $4$ & $9$ & $111011010,010010001,111001000,110100000$ & $% 1+6876z^{14}+97980z^{16}+\cdots $ \\ \hline $4$ & $9$ & $000111101,001011001,011100111,111011000$ & $% 1+7272z^{14}+95568z^{16}+\cdots $ \\ \hline $3$ & $12$ & $001011000110,101011001001,000100000000$ & $% 1+6078z^{14}+102177z^{16}+\cdots $ \\ \hline $3$ & $12$ & $000100001111,100101010110,111000101011$ & $% 1+6222z^{14}+102177z^{16}+\cdots $ \\ \hline $3$ & $12$ & $111101000001,110110100101,010100100011$ & $% 1+6270z^{14}+100497z^{16}+\cdots $ \\ \hline $3$ & $12$ & $001001100101,000000100110,010111110101$ & $% 1+6618z^{14}+99189z^{16}+\cdots $ \\ \hline $3$ & $12$ & $000100011101,010010001100,101101010110$ & $% 1+7374z^{14}+95553z^{16}+\cdots $ \\ \hline $2$ & $18$ & $101001111011011100,010110010001101110$ & $% 1+6390z^{14}+100446z^{16}+\cdots $ \\ \hline $2$ & $18$ & $101000100111110001,010110011100111011$ & $% 1+6552z^{14}+99906z^{16}+\cdots $ \\ \hline $2$ & $18$ & $100011101000011000,011111010001010010$ & $% 1+6732z^{14}+98538z^{16}+\cdots $ \\ \hline $2$ & $18$ & $010111001111100011,011010100101111000$ & $% 1+6930z^{14}+98088z^{16}+\cdots $ \\ \hline $2$ & $18$ & $001101101010011100,110111011100101100$ & $% 1+6957z^{14}+97035z^{16}+\cdots $ \\ \hline \end{tabular}} \vspace*{0.2cm}\caption{\rm\em Bordered $n$-block circulant $\left[ 74,37,14\right]^{b}$ FSD codes over $\mathbb{F}_{2}$ with $a=1=b$} \label{tab:74} \end{center} \end{table} \begin{table}[h!] \begin{center} \begin{tabular}{|l|l|l|l|l|} \hline $n$ & $m$ & $r_{1},\ldots ,r_{n}$ & $a,b$ & Partial weight distribution \\ \hline $2$ & $9$ & $\left( u03033u00\right) ,\left( 01100u131\right) $ & $3,1$ & $% 1+4518z^{14}+80364z^{16}+\cdots $ \\ \hline $2$ & $9$ & $\left( u10013303\right) ,\left( u00111uu0\right) $ & $1,1$ & $% 1+4554z^{14}+81072z^{16}+\cdots $ \\ \hline $2$ & $9$ & $\left( 3u1011111\right) ,\left( u3u331010\right) $ & $3,1$ & $% 1+4740z^{14}+79836z^{16}+\cdots $ \\ \hline $2$ & $9$ & $\left( 0u03013u3\right) ,\left( 1uu33311u\right) $ & $1,3$ & $% 1+5202z^{14}+77610z^{16}+\cdots $ \\ \hline $2$ & $9$ & $\left( 133uu0310\right) ,\left( 3u310u101\right) $ & $3,1$ & $% 1+5208z^{14}+77916z^{16}+\cdots $ \\ \hline $2$ & $9$ & $\left( uuu003131\right) ,\left( 0010uuuu1\right) $ & $1,1$ & $% 1+5262z^{14}+77232z^{16}+\cdots $ \\ \hline $2$ & $9$ & $\left( 03u110100\right) ,\left( 111113033\right) $ & $3,1$ & $% 1+5340z^{14}+77526z^{16}+\cdots $ \\ \hline $3$ & $6$ & $\left( 11303u\right) ,\left( 30u3u3\right) ,\left( 330131\right) $ & $3,1$ & $1+4206z^{14}+82047z^{16}+\cdots $ \\ \hline $3$ & $6$ & $\left( uu0133\right) ,\left( u31u10\right) ,\left( 00u0u0\right) $ & $3,3$ & $1+4716z^{14}+80295z^{16}+\cdots $ \\ \hline $3$ & $6$ & $\left( 103uu3\right) ,\left( 31u1u0\right) ,\left( 1101u3\right) $ & $1,1$ & $1+4842z^{14}+78663z^{16}+\cdots $ \\ \hline $3$ & $6$ & $\left( 331033\right) ,\left( 31u010\right) ,\left( 03u113\right) $ & $3,3$ & $1+4866z^{14}+79239z^{16}+\cdots $ \\ \hline $3$ & $6$ & $\left( 111313\right) ,\left( u033u1\right) ,\left( 0u311u\right) $ & $1,1$ & $1+4866z^{14}+79455z^{16}+\cdots $ \\ \hline $3$ & $6$ & $\left( 111300\right) ,\left( 1001u3\right) ,\left( u1303u\right) $ & $1,3$ & $1+5778z^{14}+75711z^{16}+\cdots $ \\ \hline \end{tabular}% \vspace*{0.2cm} \caption{\rm\em Binary $\left[ 76,38,14\right]^{b}$ FSD codes as Gray images of bordered $n$-block circulant codes over $R_{1}$} \label{tab:76} \end{center} \end{table} \clearpage \begin{thebibliography}{99} \bibitem{batoul} \textsc{A.\,Batoul, K.\,Guenda, A.\,Kaya, B.\,Y\i ld\i z}, \emph{Cyclic, Isodual and Formally Self-dual Codes over $\mathbb{F}_q+v% \mathbb{F}_q$}, Eur. J. Pure Appl. Math. \textbf{8}% (2015), 64--80. \bibitem{betsumiya} \textsc{K.\,Betsumiya, M.\,Harada}, \emph{Binary optimal odd formally self-dual codes}, Des. Codes Cryptogr. \textbf{23}% (2001), 11--21. \bibitem{betsumiya2} \textsc{K.\,Betsumiya, M.\,Harada}, \emph{% Classification of formally self-dual even codes of lengths up to 16}, Des. Codes Cryptogr. \textbf{23}(2001), 325--332. \bibitem{betsumiya3} \textsc{K.\,Betsumiya, M.\,Harada}, \emph{Formally self-dual codes related to Type II codes}, Appl. Algebr. Eng. Comm. \textbf{% 14}(2003), 81--88. \bibitem{Magma} \textsc{W.\,Bosma, J.\,Cannon, C.\,Playoust}, \emph{The Magma algebra system.\,I.\,The user language}, J. Symbolic Comput. \textbf{24}% (1997), 235--265. \bibitem{bouyuklieva} \textsc{S.\,Bouyuklieva, I.\,Bouyukliev}, \emph{% Classification of the extremal formally self-dual even codes of length 30}, Adv. Math. Commun. \textbf{4}(2010), 433--439. \bibitem{kovic} \textsc{D.\,Crnkovic, S.\,Rukavina, L.\,Simcic}, \emph{Binary doubly-even self-dual codes of length 72 with large automorphism groups}, Math. Commun. \textbf{18}(2013), 297--308. \bibitem{dougherty} \textsc{S.\,T.\,Dougherty, T.\,A.\,Gulliver, M.\,Harada}, \emph{Optimal formally self-dual codes over $\mathbb{F}_{5}$ and $\mathbb{F}% _{7}$}, Appl. Algebra Engrg. Comm. Comput. \textbf{10}(2000), 227--236. \bibitem{Dougherty1} \textsc{S.\,T.\,Dougherty, P.\,Gaborit, M.\,Harada, P.\,Sol% \'{e}}, \emph{Type II codes over $\mathbb{F}_2+u\mathbb{F}_2$}, IEEE Trans. Infrom. Theory \textbf{45}(1999), 32--45. \bibitem{fields} \textsc{J.\,E.\,Fields, P.\,Gaborit, W.\,C.\,Huffman, V.\,Pless}% , \emph{On the classification of extremal even formally self-dual codes of lengths 20 and 22}, Discrete Appl. Math. \textbf{111}(2001), 75--86. \bibitem{freibert} \textsc{F.\,Freibert, J.-L.\,Kim}, \emph{Optimal subcodes of formally self-dual codes and their optimum distance profiles}, Appl. Algebra Engrg. Comm. Comput. \textbf{24}(2013), 215--224. \bibitem{grassl} \textsc{M.\,Grassl}, \emph{Bounds on the minimum distance of linear codes and quantum codes}, Online available at http://codetables.de, accessed on 04.10.2018. \bibitem{gulliver} \textsc{T.\,A.\,Gulliver, M.\,Harada}, \emph{% Classification of extremal double circulant formally self-dual even codes}, Des. Codes Cryptogr. \textbf{11}(1997), 25--35. \bibitem{gulliver2} \textsc{T.\,A.\,Gulliver, M.\,Harada, T.\,Nishimura, P.\,R.\,J.\,Ostergard}, \emph{Near extremal formally self-dual even codes of lengths 24 and 32}, Des. Codes Cryptogr. \textbf{37}(2005), 465--471. \bibitem{gurel} \textsc{M.\,G\"{u}rel, N.\,Yankov}, \emph{Self-dual codes with an automorphism group of order $17$}, Math. Commun. \textbf{21}(2016), 1--11. \bibitem{huffman} \textsc{W.\,C.\,Huffman, V.\,Pless}, \emph{Fundamentals of error correcting codes}, Cambridge University Press, New York, 2003. \bibitem{u3=1} \textsc{A.\,Kaya, B.\,Yildiz}, \emph{New extremal binary self-dual codes of length 68}, Jacodesmath \textbf{1}(2014), 29--39. \bibitem{han} \textsc{S.\,Han, J.-L.\,Kim}, \emph{The non-existence of near extremal formally self-dual codes}, Des. Codes Cryptogr. \textbf{51}(2009), 69--77. \bibitem{karadeniz} \textsc{S.\,Karadeniz, S.\,T.\,Dougherty, B.\,Yildiz}, \emph{Constructing Formally Self-dual codes over $R_{k}$}, Discrete Appl. Math. \textbf{167}(2014), 188--196. \bibitem{kim} \textsc{J.-L.\,Kim, V.\,Pless}, \emph{A note on formally self-dual even codes of length divisible by 8}, Finite Fields Appl. \textbf{% 13}(2007), 224--229. \bibitem{Wood} \textsc{J.\,Wood}, \emph{Duality for modules over finite rings and applications to coding theory}, Amer. J. Math. \textbf{121}(1999), 555--575. \end{thebibliography} %\end{linenumbers} \end{document}