\documentclass[12pt,reqno]{article} \usepackage[usenames]{color} \usepackage{amssymb} \usepackage{amsmath} \usepackage{amsthm} \usepackage{amsfonts} \usepackage{amscd} \usepackage{graphicx} \usepackage[colorlinks=true, linkcolor=webgreen, filecolor=webbrown, citecolor=webgreen]{hyperref} \definecolor{webgreen}{rgb}{0,.5,0} \definecolor{webbrown}{rgb}{.6,0,0} \usepackage{color} \usepackage{fullpage} \usepackage{float} \usepackage{psfig} \usepackage{graphics} \usepackage{latexsym} \usepackage{epsf} \usepackage{breakurl} \setlength{\textwidth}{6.5in} \setlength{\oddsidemargin}{.1in} \setlength{\evensidemargin}{.1in} \setlength{\topmargin}{-.1in} \setlength{\textheight}{8.4in} \newcommand{\seqnum}[1]{\href{https://oeis.org/#1}{\rm \underline{#1}}} \DeclareMathOperator{\GL}{GL} \DeclareMathOperator{\Mat}{Mat} \DeclareMathOperator{\lcm}{lcm} \DeclareMathOperator{\ord}{ord} \DeclareMathOperator{\diag}{diag} \DeclareMathOperator{\mmod}{mod\!\ } \def\modd#1 #2{#1\ \mbox{\rm (mod}\ #2\mbox{\rm )}} \begin{document} \begin{center} \epsfxsize=4in \leavevmode\epsffile{logo129.eps} \end{center} \theoremstyle{plain} \newtheorem{theorem}{Theorem} \newtheorem{corollary}[theorem]{Corollary} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{proposition}[theorem]{Proposition} \theoremstyle{definition} \newtheorem{definition}[theorem]{Definition} \newtheorem{example}[theorem]{Example} \newtheorem{conjecture}[theorem]{Conjecture} \theoremstyle{remark} \newtheorem{remark}[theorem]{Remark} \begin{center} \vskip 1cm{\LARGE\bf Carmichael Numbers for $\GL(m)$} \vskip 1cm \large Eugene Karolinsky and Dmytro Seliutin \\ Department of Pure Mathematics\\ Kharkiv National University\\ Ukraine \\ \href{mailto:ekarolinsky@univer.kharkov.ua}{\tt ekarolinsky@univer.kharkov.ua} \\ \href{mailto:selyutind1996@gmail.com}{\tt selyutind1996@gmail.com} \end{center} \vskip .2 in \begin{abstract} We propose a generalization of Carmichael numbers, where the multiplicative group $\mathbb G_{m} = \GL(1)$ is replaced by $\GL(m)$ for $m\geq 2$. We prove basic properties of these families of numbers and give some examples. \end{abstract} \section{Introduction}\label{sec:intro} Recall that a composite number $n\in\mathbb N$ is called \emph{Carmichael} if $a^{n-1} \equiv \modd{1} {n}$ for every $a\in(\mathbb Z/n\mathbb Z)^\times$. In other words, Carmichael numbers are Fermat pseudoprimes to all values of $a$ that are coprime to $n$. Alford, Granville, and Pomerance \cite{AGP} proved that there are infinitely many Carmichael numbers. Recently, various generalizations and analogues of Carmichael numbers were proposed, see, e.g., \cite{Howe, McIntosh, Steele} and references therein. In this paper, we introduce another analogue of Carmichael numbers, where the multiplicative group $\mathbb G_{m} = \GL(1)$ is replaced by $\GL(m)$ for $m\geq 2$. Namely, we start with the exponent $K_m(p)$ of the group $\GL(m,\mathbb F_p)$, extrapolate it naturally to all naturals as $K_m(n)$, and then define a composite number $n\in\mathbb N$ to be $m$-Carmichael if $A^{K_m(n)}=I$ for all $A\in\GL(m, \mathbb Z/n\mathbb Z)$. Thus, the ``classical'' Carmichael numbers can be recovered as $1$-Carmichael numbers. We study basic properties of $m$-Carmichael numbers, including an analogue of Korselt's criterion for a number to be Carmichael in terms of its prime divisors. This criterion appears to be practical for numbers of reasonable size. Using this, we compute all $m$-Carmichael numbers less than or equal to $10^5$ for $2\leq m\leq10$. Also, we describe the structure of $m$-Carmichael numbers with given prime factors. Some properties of $m$-Carmichael numbers for $m\geq 2$ appear to be rather different from those of the ``classical'' Carmichael numbers. Namely, $m$-Carmichael numbers for $m\geq 2$ can be non-squarefree or even. Moreover, all prime powers are $m$-Carmichael for $m\geq 2$. Possible explanation of these phenomena is the fact that the groups $\GL(m)$ for $m\geq2$ contain (many copies of) the additive group $\mathbb G_{a}$. The paper is organized as follows. In Section \ref{sec:sec_2}, we define $m$-Carmichael numbers and discuss their basic properties. The main result of the paper is Theorem \ref{thm_Korselt}, an analogue of Korselt's criterion for $m$-Carmichael numbers. In Section \ref{sec:sec_2e}, we consider the distribution of $m$-Carmichael numbers with prescribed prime factors, give several particular examples, and summarize the general pattern in Theorem \ref{thm_P}. In Section \ref{sec:sec_4}, we list some open questions and discuss possible generalizations. In Appendix, we describe our computations of relatively small $m$-Carmichael numbers. Throughout this paper, $p$ denotes a prime number. In particular, $\prod_{p\mid n}$ stands for a product taken over all prime divisors of $n$. \section{$m$-Carmichael numbers}\label{sec:sec_2} Let $\Phi_k(X)$ be the $k$th cyclotomic polynomial. The following proposition is well known, but for the reader's convenience we provide a proof. \begin{proposition} If $a\in \mathbb Z$, then \[ \lcm(a-1,a^2-1,\ldots,a^m-1)=\prod_{k=1}^m\Phi_k(a). \] \end{proposition} \begin{proof} We proceed by induction with the obvious base. We have \begin{gather*} \lcm(a-1,a^2-1,\ldots,a^m-1)=\lcm\left(\prod_{k=1}^{m-1}\Phi_k(a),a^m-1\right)=\\ \left(\prod_{d\mid m, d1$, then $p^k$ is an $m$-Carmichael number. \end{proposition} \begin{proof} Let $A\in\Mat(m,\mathbb Z)$, $\gcd(\det A, p)=1$. By Theorem \ref{fermat_matr}, we have $B:=A^{K_m(p)}\equiv I\ (\mmod p)$. Therefore, by Lemma \ref{lemma_matr_mod_p_pow} we have $B^{p^{k-1}}\equiv I\ (\mmod p^k)$. Since $\nabla_m(p^k)=\nabla_m(p)$ and, by Lemma \ref{lemma_cycl_div}, $D_m(p)\mid D_m(p^k)$, the equation $B^{p^{k-1}}=A^{p^k\nabla_m(p)D_m(p)}\equiv I\ (\mmod p^k)$ implies $A^{K_m(p^k)}=A^{p^k\nabla_m(p^k)D_m(p^k)}\equiv I\ (\mmod p^k)$. \end{proof} Now we present the main theorem of the paper, an analogue of Korselt's criterion for a number to be $m$-Carmichael. \begin{theorem}\label{thm_Korselt} Let $n\in\mathbb N$ be composite. The following are equivalent: (1) $n$ is an $m$-Carmichael number, (2) if $p\mid n$, then $D_m(p)\mid K_m(n)$. \end{theorem} \begin{proof} \noindent (2) $\implies$ (1): Let $D_m(p)\mid K_m(n)$ for all $p\mid n$. Since $p^{\ord_p(n)}\nabla_m(p)\mid K_m(n)$ and \[ \gcd(p^{\ord_p(n)}\nabla_m(p),D_m(p))=1, \] we have $p^{\ord_p(n)}\nabla_m(p)D_m(p)\mid K_m(n)$ for all $p\mid n$. Now consider $A\in\Mat(m,\mathbb Z)$, $\gcd(\det A, n)=1$. By Theorem \ref{fermat_matr}, we have $A^{p\nabla_m(p)D_m(p)}\equiv I\ (\mmod p)$ for all $p\mid n$. Lemma \ref{lemma_matr_mod_p_pow} implies $A^{p^{\ord_p(n)}\nabla_m(p)D_m(p)}\equiv I\ (\mmod p^{\ord_p(n)})$. Thus, $A^{K_m(n)}\equiv I\ (\mmod p^{\ord_p(n)})$ for all $p\mid n$. By the Chinese remainder theorem, $A^{K_m(n)}\equiv I\ (\mmod n)$. Therefore, $n$ is an $m$-Carmichael number. \bigskip \noindent(1) $\implies$ (2): Conversely, assume that for some $p\mid n$ we have $D_m(p)\nmid K_m(n)$. Since $D_m(p)=\lcm(p-1,p^2-1,\ldots,p^m-1)$, there exists $k\in\{1, 2,\ldots, m\}$ such that $p^k-1\nmid K_m(n)$. We construct an element $A\in\GL(m,\mathbb Z/n\mathbb Z)$ of order $p^k-1$. Therefore, $A^{K_m(n)}\neq I$, and $n$ is not $m$-Carmichael. To this end, let $\alpha$ be a generator of the cyclic group $\mathbb F_{p^k}^\times$. Consider the polynomial $(X-\alpha)(X-\alpha^p)\ldots(X-\alpha^{p^{k-1}})\in \mathbb F_p[X]$, and let $B\in\GL(k,\mathbb F_p)$ be its Frobenius companion matrix. Then $B$ is of order $p^k-1$, and the same is true for $C=\diag(B,I)\in\GL(m,\mathbb F_p)$. Lift $C$ to an element of $\Mat(m,\mathbb Z)$. Then $C^{p^k-1}\equiv I\ (\mmod p)$. By Lemma \ref{lemma_matr_mod_p_pow}, we have $(C^{p^{\ord_p(n)-1}})^{p^k-1}\equiv I\ (\mmod p^{\ord_p(n)})$. Moreover, since $\gcd(p^{\ord_p(n)}, p^k-1)=1$, we see that $C^{p^{\ord_p(n)-1}}\bmod p^{\ord_p(n)}$ is also of order $p^k-1$. Finally, by the Chinese remainder theorem, consider $A\in\GL(m,\mathbb Z/n\mathbb Z)$ such that $A\equiv C^{p^{\ord_p(n)-1}}\ (\mmod p^{\ord_p(n)})$ and, for example, $A\equiv I\ (\mmod \frac{n}{p^{\ord_p(n)}})$. By construction, $A$ is of order $p^k-1$, which finishes the proof. \end{proof} \begin{remark} Proposition \ref{prop_prime_pow} also follows easily from Theorem \ref{thm_Korselt}. \end{remark} Moreover, applying Theorem \ref{thm_Korselt} and Lemma \ref{lemma_cycl_div}, we get \begin{corollary}\label{cor_powers} If $n$ is an $m$-Carmichael number, and $k\in\mathbb N$, then $n^k$ is also an $m$-Carmichael number. \end{corollary} Finally, we present one necessary condition for a number to be $m$-Car\-mi\-chael. \begin{proposition} Assume that $n$ is an $m$-Carmichael number. Then $n\not\equiv 2\ (\mmod 4)$. \end{proposition} \begin{proof} Let $n\in\mathbb N$ be composite, $n\equiv 2\ (\mmod 4)$. Then $D_m(n)$ is odd, and thus $\ord_2(K_m(n))=\lceil\log_2m\rceil$. On the other hand, consider an odd $p\mid n$. Since $\Phi_{2^k}(p)=p^{2^{k-1}}+1$ is even for $k\geq 1$, and $8\mid\Phi_1(p)\Phi_2(p)=p^2-1$, we have $\ord_2(D_m(p))\geq\lfloor\log_2m\rfloor+2>\lceil\log_2m\rceil$. Thus, $D_m(p)\nmid K_m(n)$, and $n$ is not $m$-Carmichael. \end{proof} \section{$m$-Carmichael numbers with prescribed prime factors}\label{sec:sec_2e} In Theorem \ref{thm_Korselt}, the divisibility condition for small values of $m$ is transparent enough to find some infinite families of $m$-Carmichael numbers (apart of prime powers). Let us start with $m=2$. Let $d_2(p, n)$ denote the condition $D_2(p)\mid K_2(n)$, i.e., $p^2-1\mid n(n^2-1)$. We have \begin{itemize} \item $d_2(2, n)$ is equivalent to $3\mid n(n^2-1)$, which is satisfied for all $n$. \item $d_2(3, n)$ is equivalent to $8\mid n(n^2-1)$, which is satisfied if and only if $n$ is odd or $8\mid n$. \item $d_2(5, n)$ is equivalent to $3\cdot8\mid n(n^2-1)$, which is also satisfied if and only if $n$ is odd or $8\mid n$. \item $d_2(7, n)$ is equivalent to $3\cdot16\mid n(n^2-1)$, which is satisfied if and only if $n\equiv \pm1\ (\mmod 8)$ or $16\mid n$. \item $d_2(11, n)$ is equivalent to $3\cdot5\cdot8\mid n(n^2-1)$, which is satisfied if and only if $d_2(5, n)$ and $5\mid n(n^2-1)$ are satisfied; the latter is satisfied if and only if $n\equiv 0,\pm1\ (\mmod 5)$. \end{itemize} Using the above, the following propositions are proven by a direct application of Theorem \ref{thm_Korselt}. \begin{proposition}\label{prop_2_Carm_7} Let $n\in\mathbb N$ be a composite $7$-smooth number which is not a prime power. Then $n$ is $2$-Carmichael if and only if $n$ belongs to one of the following families: \begin{enumerate} \item $n=2^k\cdot3^l\cdot5^r$, where $k\geq3$, \item $n=2^k\cdot3^l\cdot5^r\cdot7^s$, where $k\geq4$, $s\geq1$, \item $n=3^l\cdot5^r$, \item $n=3^l\cdot5^r\cdot7^s$, where $l\equiv r\ (\mmod 2)$, $s\geq1$. \end{enumerate} \end{proposition} \begin{proposition}\label{prop_2_Carm_11} Let $n\in\mathbb N$ be a composite $11$-smooth number which is not $7$-smooth and not a prime power. Then $n$ is $2$-Carmichael if and only if $n$ belongs to one of the following families: \begin{enumerate} \item $n=2^k\cdot3^l\cdot5^r\cdot11^t$, where $k\geq3$, $r\geq1$, \item $n=2^k\cdot3^l\cdot5^r\cdot7^s\cdot11^t$, where $k\geq4$, $r\geq1$, $s\geq1$, \item $n=2^k\cdot3^l\cdot11^t$, where $k\geq3$, $k\equiv l\ (\mmod 2)$, \item $n=2^k\cdot3^l\cdot7^s\cdot11^t$, where $k\geq4$, $s\geq1$, $k+l+s$ is even, \item $n=3^l\cdot5^r\cdot11^t$, where $r\geq1$, \item $n=3^l\cdot5^r\cdot7^s\cdot11^t$, where $r\geq1$, $s\geq1$, $l+r+t$ is even, \item $n=3^l\cdot11^t$, where $l$ is even, \item $n=3^l\cdot7^s\cdot11^t$, where $s\geq1$, $l\equiv s\equiv t\ (\mmod 2)$. \end{enumerate} \end{proposition} We now consider $m = 3$ or $m = 4$. We restrict ourselves to composite numbers of the form $n = 2^k3^l$. \begin{proposition}\label{prop_3_Carm} Let $n = 2^k3^l$, where $k, l\geq1$. Then $n$ is $3$-Carmichael if and only if $k\geq2$, and $(k, l)$ belongs to one of the following families: \begin{enumerate} \item $k\equiv 0\ (\mmod 12)$, $l\equiv 0, 2, 3, 4\ (\mmod 6)$, \item $k\equiv \pm2\ (\mmod 12)$, $l\equiv \pm4\ (\mmod 6)$, \item $k\equiv \pm4\ (\mmod 12)$, $l\equiv 0, \pm1, 2, 4\ (\mmod 6)$, \item $k\equiv 6\ (\mmod 12)$, $l\equiv 0\ (\mmod 3)$. \end{enumerate} \end{proposition} \begin{proposition}\label{prop_4_Carm} Let $n = 2^k3^l$, where $k, l\geq1$. Then $n$ is $4$-Carmichael if and only if $k\geq3$, and $(k, l)$ belongs to one of the families 1) -- 4) in Proposition \ref{prop_3_Carm} or to one of the following families: \begin{enumerate} \setcounter{enumi}{4} \item $k\equiv \pm1\ (\mmod 12)$, $l\equiv \pm2\ (\mmod 6)$, \item $k\equiv \pm3\ (\mmod 12)$, $l\equiv 0\ (\mmod 3)$, \item $k\equiv \pm5\ (\mmod 12)$, $l\equiv \pm4\ (\mmod 6)$. \end{enumerate} \end{proposition} \noindent\emph{Proof of Propositions \ref{prop_3_Carm} and \ref{prop_4_Carm}.} We have $D_3(2)=3\cdot7$, $D_4(2)=3\cdot5\cdot7$, $D_3(3)=2^3\cdot13$, $D_4(3)=2^4\cdot5\cdot13$, $\nabla_3(n)=2$, $\nabla_4(n)=2\cdot3$. Therefore, $n$ is $3$-Carmichael if and only if $K_3(n)=2n(n^2-1)(n^2+n+1)$ is divisible by $2^3$, $3$, $7$, and $13$, which is equivalent to the conditions $4\mid n$ (i.e., $k\geq2$), $n\equiv\pm1, 2, 4\ (\mmod 7)$, $n\equiv\pm1, 3, 9\ (\mmod 13)$. Similarly, $n$ is $4$-Carmichael if and only if $K_4(n)=6n(n^2-1)(n^2+n+1)(n^2+1)$ is divisible by $2^4$, $3$, $5$, $7$, and $13$, which is equivalent to $8\mid n$ (i.e., $k\geq3$), $n\equiv\pm1, 2, 4\ (\mmod 7)$, $n\equiv\pm1, 3, 9, \pm5\ (\mmod 13)$. Since $|\mathbb F_7^\times|=6$, $|\mathbb F_{13}^\times|=12$, and $3\bmod 13$ is of order $3$, we see that the conditions on $n$ modulo $7$ and $13$ depend only on $k\bmod 12$, $l\bmod 6$. The corresponding values of $k\bmod 12$, $l\bmod 6$ are obtained by a direct calculation. \medskip Now we describe the general pattern of the distribution of $m$-Carmichael numbers with prescribed prime factors. Let $P$ be a finite nonempty subset of primes. Let $D_m(P)$ denote the least common multiple of $D_m(p)$ for all $p\in P$. Let us say that $n\in\mathbb N$ is a \emph{$P$-number}, if $n$ is divisible precisely by the primes in $P$. By Theorem \ref{thm_Korselt}, a $P$-number $n$ is $m$-Carmichael if and only if $D_m(P)\mid K_m(n)$. Further, write $D_m(P)=D_m'(P)D_m''(P)$, where $D_m'(P)$ is a product of primes in $P$, and $D_m''(P)$ is coprime to all $p\in P$. Then a $P$-number $n$ is $m$-Carmichael if and only if $D_m'(P)\mid K_m(n)$ and $D_m''(P)\mid K_m(n)$. First, notice that, since $n\mid K_m(n)$, $\nabla_m(P):=\nabla_m(n)$ depends only on $P$, and $D_m'(P)$ is coprime to $D_m(n)$, it follows that the condition $D_m'(P)\mid K_m(n)$ is satisfied if and only if $\ord_pn\geq \ord_pD_m'(P)-\ord_p\nabla_m(P)$ for all primes $p\in P$. Secondly, since a $P$-number $n$ is, by construction, invertible modulo $D_m''(P)$, we see that the condition $D_m''(P)\mid K_m(n)$ depends only on the values of $\ord_pn\bmod\lambda(D_m''(P))$ for $p\in P$. Here $\lambda$ is the Carmichael function, i.e., $\lambda(D_m''(P))$ is the exponent of the group $(\mathbb Z/D_m''(P)\mathbb Z)^\times$. Moreover, if $p\in P$, let $v_{m, P}(p)$ denote the order of $p\bmod D_m''(P)$ in the group $(\mathbb Z/D_m''(P)\mathbb Z)^\times$. Then the condition $D_m''(P)\mid K_m(n)$ depends only on the values of $\ord_pn\bmod v_{m, P}(p)$ for $p\in P$. Thus, we get the following \begin{theorem}\label{thm_P} For any $p\in P$, the set of $P$-numbers which are $m$-Car\-mi\-chael is invariant under multiplication by $p^{v_{m, P}(p)}$. \end{theorem} \begin{corollary}\label{cor_inf} Assume that there exists a $P$-number which is $m$-Car\-mi\-chael. Then there are infinitely many of them. \end{corollary} \begin{remark} Corollary \ref{cor_inf} follows also from Corollary \ref{cor_powers}. \end{remark} All propositions of this section can be viewed as examples of Theorem \ref{thm_P}. For example, for $m=3$ and $P=\{2,3\}$ we have $D_m'(P)=2^3\cdot3$, $D_m''(P)=7\cdot13$, and $\lambda(7\cdot13)=\lcm(6,12)=12$, $v_{m, P}(2)=12$, $v_{m, P}(3)=6$, which is in accordance with Proposition \ref{prop_3_Carm}. \section{Concluding remarks}\label{sec:sec_4} We list some natural questions that remain open. \begin{itemize} \item For what $m$ and $P$ are there $P$-numbers which are $m$-Carmichael? \item Are there squarefree $m$-Carmichael numbers for $m\geq3$? \end{itemize} \begin{remark} One can consider an analogous notion for other affine group schemes of finite type defined over $\mathbb Z$. Namely, if $G$ is such a group scheme, $K_G(p)$ the exponent of the group $G(\mathbb F_p)$, and $K_G(n)$ its reasonable extra\-po\-lation to all $n\in\mathbb N$, then one can consider $G$-Carmichael numbers, i.e., composite $n\in\mathbb N$ such that $g^{K_G(n)}=1$ for all $g\in G(\mathbb Z/n\mathbb Z)$. \end{remark} \section{Acknowledgments}\label{sec:ackn} The research of the second author was partially supported by the Akhiezer Foundation. \appendix \section{Numerical experiments}\label{sec:sec_3} We calculate, via brute force, all $m$-Carmichael numbers up to $10^5$ for $2\leq m\leq10$. Let us call an $m$-Carmichael number \emph{nontrivial} if it is not a prime power. There are $1330$ nontrivial $2$-Carmichael numbers, $44$ nontrivial $3$-Carmichael numbers, and $28$ nontrivial $4$-Carmichael numbers on the researched interval. There are no nontrivial $m$-Carmichael numbers for $5\leq m\leq10$ on the researched interval. Among $16$ Carmichael numbers less than $10^5$, the following four, namely, \begin{center} \begin{tabular}{llll} $1729=7\cdot13\cdot19$, & $2465=5\cdot17\cdot29$, & $6601=7\cdot23\cdot41$, & $41041=7\cdot11\cdot13\cdot41$, \end{tabular} \end{center} \noindent are $2$-Carmichael. None of these Carmichael numbers are $m$-Carmichael for $3\leq m\leq10$. Moreover, none of $m$-Carmichael numbers for $3\leq m\leq10$ on the researched interval are squarefree. There are $18$ numbers on the researched interval, namely, \begin{center} \begin{tabular}{llll} $48=2^4\cdot3$, & $144=2^4\cdot3^2$, & $1296=2^4\cdot3^4$, & $1728=2^6\cdot3^3$,\\ $2304=2^8\cdot3^2$, & $5760=2^7\cdot3^2\cdot5$, & $9216=2^{10}\cdot3^2$, & $11664=2^4\cdot3^6$,\\ $20736=2^8\cdot3^4$, & $25600=2^{10}\cdot5^2$, & $27000=2^3\cdot3^3\cdot5^3$, & $30720=2^{11}\cdot3\cdot5$,\\ $34992=2^4\cdot3^7$, & $36864=2^{12}\cdot3^2$, & $46656=2^6\cdot3^6$, & $62208=2^8\cdot3^5$,\\ $96768=2^9\cdot3^3\cdot7$, & $99225=3^4\cdot5^2\cdot7^2$, \end{tabular} \end{center} \noindent that are nontrivial $m$-Carmichael numbers for all $m\in\{2, 3, 4\}$, and one number, $22815=3^3\cdot5\cdot13^2$, that is nontrivial $3$-Carmichael and $4$-Carmichael, but not $2$-Carmichael. Also, on the researched interval there are $19$ numbers, namely, \begin{center} \begin{tabular}{llll} $160=2^5\cdot5$, & $448=2^6\cdot7$, & $704=2^6\cdot11$, & $800=2^5\cdot5^2$,\\ $1056=2^5\cdot3\cdot11$, & $2640=2^4\cdot3\cdot5\cdot11$, & $3136=2^6\cdot7^2$, & $5929=7^2\cdot11^2$,\\ $7744=2^6\cdot11^2$, & $18144=2^5\cdot3^4\cdot7$, & $20000=2^5\cdot5^4$, & $21952=2^6\cdot7^3$,\\ $28672=2^{12}\cdot7$, & $29952=2^8\cdot3^2\cdot13$, & $31744=2^{10}\cdot31$, & $34496=2^6\cdot7^2\cdot11$,\\ $39424=2^9\cdot7\cdot11$, & $45056=2^{12}\cdot11$, & $85184=2^6\cdot11^3$, \end{tabular} \end{center} \noindent that are nontrivial $2$- and $3$-Carmichael, but not $4$-Carmichael; $8$ numbers, namely, \begin{center} \begin{tabular}{llll} $216=2^3\cdot3^3$, & $1152=2^7\cdot3^2$, & $2592=2^5\cdot3^4$, & $4000=2^5\cdot5^3$,\\ $5832=2^3\cdot3^6$, & $13824=2^9\cdot3^3$, & $28800=2^7\cdot3^2\cdot5^2$, & $73728=2^{13}\cdot3^2$, \end{tabular} \end{center} \noindent that are nontrivial $2$- and $4$-Carmichael, but not $3$-Carmichael; $6$ numbers, namely, \begin{center} \begin{tabular}{lll} $324=2^2\cdot3^4$, & $900=2^2\cdot3^2\cdot5^2$, & $1404=2^2\cdot3^3\cdot13$,\\ $39204=2^2\cdot3^4\cdot11^2$, & $74088=2^3\cdot3^3\cdot7^3$, & $74536=2^3\cdot7\cdot11^3$, \end{tabular} \end{center} \noindent that are nontrivial $3$-Carmichael, but not $2$- or $4$-Carmichael. Finally, one number, $26112=2^9\cdot3\cdot17$, is nontrivial $4$-Carmichael, but not $2$- or $3$-Carmichael. Below we list all nontrivial $2$-Carmichael numbers up to $3000$ that are not treated by Propositions \ref{prop_2_Carm_7} and \ref{prop_2_Carm_11} (i.e., not $11$-smooth). \begin{center} \begin{tabular}{llll} $104=2^3\cdot13$ & $171=3^2\cdot19$ & $195=3\cdot5\cdot13$ & $273=3\cdot7\cdot13$\\ $351=3^3\cdot13$ & $435=3\cdot5\cdot29$ & $455=5\cdot7\cdot13$ & $609=3\cdot7\cdot29$\\ $615=3\cdot5\cdot41$ & $624=2^4\cdot3\cdot13$ & $665=5\cdot7\cdot19$ & $715=5\cdot11\cdot13$\\ $736=2^5\cdot23$ & $759=3\cdot11\cdot23$ & $832=2^6\cdot13$ & $855=3^2\cdot5\cdot19$\\ $903=3\cdot7\cdot43$ & $1001=7\cdot11\cdot13$ & $1015=5\cdot7\cdot29$ & $1045=5\cdot11\cdot19$\\ $1071=3^2\cdot7\cdot17$ & $1088=2^6\cdot17$ & $1183=7\cdot13^2$ & $1216=2^6\cdot19$\\ $1265=5\cdot11\cdot23$ & $1352=2^3\cdot13^2$ & $1377=3^4\cdot17$ & $1431=3^3\cdot53$\\ $1456=2^4\cdot7\cdot13$ & $1520=2^4\cdot5\cdot19$ & $1539=3^4\cdot19$ & $1560=2^3\cdot3\cdot5\cdot13$\\ $1595=5\cdot11\cdot29$ & $1625=5^3\cdot13$ & $1729=7\cdot13\cdot19$ & $1856=2^6\cdot29$\\ $1881=3^2\cdot11\cdot19$ & $1911=3\cdot7^2\cdot13$ & $1984=2^6\cdot31$ & $2001=3\cdot23\cdot29$\\ $2009=7^2\cdot41$ & $2015=5\cdot13\cdot31$ & $2080=2^5\cdot5\cdot13$ & $2211=3\cdot11\cdot67$\\ $2255=5\cdot11\cdot41$ & $2365=5\cdot11\cdot43$ & $2375=5^3\cdot19$ & $2457=3^3\cdot7\cdot13$\\ $2465=5\cdot17\cdot29$ & $2535=3\cdot5\cdot13^2$ & $2565=3^3\cdot5\cdot19$ & $2624=2^6\cdot41$\\ $2639=7\cdot13\cdot29$ & $2736=2^4\cdot3^2\cdot19$ & $2808=2^3\cdot3^3\cdot13$ & $2871=3^2\cdot11\cdot29$\\ $2912=2^5\cdot7\cdot13$ & $2925=3^2\cdot5^2\cdot13$ \end{tabular} \end{center} Also, we manage to compute a few larger $m$-Carmichael numbers for $m\geq5$. We consider numbers of the form $2^k3^l$ for relatively small values of $k$ and $l$ and apply Theorem \ref{thm_Korselt}. For instance, $2^{22}\cdot3^2$ is $2$- (by Proposition \ref{prop_2_Carm_7}), $3$- (by Proposition \ref{prop_3_Carm}), $4$- (by Proposition \ref{prop_4_Carm}), $5$- and $6$-Carmichael, but not $7$- or $8$-Carmichael. Similarly, $2^{286}\cdot3^{36}$ is $2$-, $6$-, $7$-, and $8$-Carmichael, but not $3$-, $4$-, or $5$-Car\-mi\-chael. \begin{thebibliography}{10} \bibitem{AGP} W.~R.~Alford, A.~Granville, and C.~Pomerance, There are infinitely many Carmichael numbers, {\em Ann.\ of Math.\ (2)} {\bf 139} (1994), 703--722. \bibitem{Ge} Y.~Ge, Elementary properties of cyclotomic polynomials, {\em Math.\ Reflec.} {\bf 2} (2008). \bibitem{Howe} E.~W.~Howe, Higher-order Carmichael numbers, {\em Math.\ Comp.} {\bf 69} (2000), 1711--1719. \bibitem{Marshall} J.~B.~Marshall, On the extension of Fermat's theorem to matrices of order $n$, {\em Proc.\ Edinb.\ Math.\ Soc.} {\bf 6} (1939), 85--91. \bibitem{McIntosh} R.~J.~McIntosh, M.~Dipra, Carmichael numbers with $p+1\mid n+1$, {\em J.\ Number Theory} {\bf 147} (2015), 81--91. \bibitem{Niven} I.~Niven, Fermat's theorem for matrices, {\em Duke Math.\ J.} {\bf 15} (1948), 823--826. \bibitem{Steele} G.~A.~Steele, Carmichael numbers in number rings, {\em J.\ Number Theory} {\bf 128} (2008), 910--917. \end{thebibliography} \bigskip \hrule \bigskip \noindent 2010 {\it Mathematics Subject Classification}: Primary 11A51, Secondary 11Y11, 20G30. \noindent \emph{Keywords: } Carmichael number, pseudoprime, general linear group. \bigskip \hrule \bigskip \noindent (Concerned with sequences \seqnum{A002997} and \seqnum{A336663}.) \bigskip \hrule \bigskip \vspace*{+.1in} \noindent Received March 3 2020; revised version received July 30 2020. Published in {\it Journal of Integer Sequences}, October 18 2020. \bigskip \hrule \bigskip \noindent Return to \htmladdnormallink{Journal of Integer Sequences home page}{http://www.cs.uwaterloo.ca/journals/JIS/}. \vskip .1in \end{document}