--- title: 'Kurepa Conjecture: Trees and Factorials' url: https://www.emergentmind.com/topics/kurepa-conjecture type: topic --- # Kurepa Conjecture: Trees and Factorials In the literature surveyed here, the term **Kurepa Conjecture** is used in two distinct senses. In set theory, the **Kurepa hypothesis** is the assertion that there exists a Kurepa tree, usually an $\omega_1$-tree with countable levels and at least $\omega_2$ many cofinal branches. In number theory, **Kurepa’s conjecture** concerns the left factorial $$ !n=0!+1!+\cdots+(n-1)!, $$ and states that $$ \gcd(!n,n!)=2 \qquad \text{for all } n>1, $$ equivalently that there is no odd prime $p$ such that $p\mid !p$ [2404.02715] [1409.0800]. ## 1. Terminological scope and canonical formulations Set-theoretic papers in this corpus treat the classical Kurepa hypothesis as a branch-existence statement for $\omega_1$-trees, and also in the modern combinatorial form $\mathrm{KH}(\omega_1,\omega)$: a family $\mathcal F\subseteq \mathcal P(\omega_1)$ of size at least $\omega_2$ such that for every $x\in[\omega_1]^\omega$, $$ |\mathcal F\!\upharpoonright x|<\omega, \qquad \mathcal F\!\upharpoonright x=\{\,t\cap x:t\in\mathcal F\,\}. $$ More generally, for infinite cardinals $\lambda<\kappa$, a $\mathrm{KH}(\kappa,\lambda)$-family is a family $\mathcal F\subseteq\mathcal P(\kappa)$ with $|\mathcal F|\ge\kappa^+$ and $|\mathcal F\!\upharpoonright x|<\lambda$ for every $x\in[\kappa]^\lambda$ [1510.02932]. Number-theoretic papers use the same name for the left-factorial divisibility problem. There the central function is $$ !n=\sum_{k=0}^{n-1}k!, $$ and the conjecture is $$ \gcd(!n,n!)=2 \qquad (n>1), $$ equivalently $!n$ is not divisible by $n$ for any $n>2$, and it is enough to check primes [1409.0800]. | Area | Classical statement | Basic object | |---|---|---| | Set theory | There exists a Kurepa tree | An $\omega_1$-tree with countable levels and at least $\omega_2$ many cofinal branches | | Number theory | $\gcd(!n,n!)=2$ for $n>1$ | The left factorial $!n=\sum_{k=0}^{n-1}k!$ | This split usage suggests that “Kurepa conjecture” functions less as a single theorem-name than as a label for a family of Kurepa-type problems, united more by historical attribution than by subject matter. ## 2. The classical set-theoretic object and branch-guessing principles A binary Kurepa tree is defined as a downward-closed $T\subseteq{}^{<\omega_1}2$ with countable levels $$ T_\alpha:=\{t\in T\mid \operatorname{dom}(t)=\alpha\}, $$ such that the set of cofinal branches $$ B(T):=\{f\in{}^{\omega_1}2\mid \forall \alpha<\omega_1\ (f\upharpoonright\alpha\in T_\alpha)\} $$ has size at least $\aleph_2$. More generally, an $\omega_1$-tree is a tree of height $\omega_1$ with countable levels; a Kurepa tree is such a tree with at least $\aleph_2$ many branches; and an Aronszajn tree is an $\omega_1$-tree with no uncountable branches [2404.02715]. A recent refinement is the branch-relative guessing principle $\diamond(T)$. For a binary Kurepa tree $T$, $\diamond(T)$ asserts the existence of a transversal $$ (t_\alpha)_{\alpha<\omega_1}\in \prod_{\alpha<\omega_1}T_\alpha $$ such that for every $f\in B(T)$, the set $$ G(f):=\{\alpha<\omega_1\mid f\upharpoonright\alpha=t_\alpha\} $$ is stationary. The principle is implied by full $\diamond$, and even by $\diamond(\omega,=)$; it is compatible with Martin’s axiom; and it may consistently fail, even for rigid or homogeneous binary Kurepa trees. The same paper proves that there is a model with $$ 2^{\aleph_0}=\aleph_2,\quad \mathrm{MA},\quad \text{and }\diamond(T)\text{ for some binary Kurepa tree }T $$ [2404.02715]. These results separate the bare existence of a Kurepa tree from stronger branch-guessing behavior. The existence of $T$ does not by itself force Jensen-style prediction principles on its branch space. ## 3. Forcing, topology, and strengthened Kurepa trees A topological strengthening introduced by Bergfalk, Chodounský, Guzmán, and Hrušák is the notion of a **strong Kurepa tree**: a Kurepa tree $T$ such that ${}^{\omega_1}2$ is a continuous image of the branch space $[T]$ equipped with the cone topology. In that topology, a basic open set has the form $$ \{g\in[T]:x\in g\} $$ for some node $x\in T$. Under $\mathsf{CH}$, for every cardinal $1\le\kappa\le 2^{\omega_1}$ there is a forcing poset that is $\sigma$-closed, $\omega_2$-c.c., and adds a Kurepa tree $T$ together with a surjective $\kappa$-to-one continuous function $$ F:[T]\to{}^{\omega_1}2. $$ The forcing yields a canonical map via $$ F(b)=\bigcup\{H(b\upharpoonright\gamma):\gamma<\omega_1\}, $$ and any Kurepa tree with this strong property contains an Aronszajn subtree [2507.01133]. Another forcing-theoretic refinement is the notion of a **sealed** Kurepa tree. A Kurepa tree $T$ in a model $M$ is sealed if every forcing notion $P\in M$ that adds a new branch to $T$ collapses $\omega_1$ or $\omega_2$. Relative to a strongly inaccessible cardinal, it is consistent that there exists an $\omega$-distributive forcing notion $R$ of cardinality $\aleph_1$ such that for every Kurepa tree $T$ in the generic extension, $R$ adds a branch to $T$; in particular, no such $T$ is sealed. The same work also gives a proper forcing notion that makes an arbitrary Kurepa tree non-distributive while preserving $\omega_1$ and $\omega_2$ [2401.00572]. Taken together, these results show that the classical Kurepa hypothesis admits substantial strengthening and substantial destabilization. A Kurepa tree can be forced to have strong topological universality, but it can also be placed in models where every Kurepa tree admits small branch-adding forcing. ## 4. Weak Kurepa hypotheses and guessing-model principles A **weak Kurepa tree** is a tree of height and size $\omega_1$ with at least $\omega_2$ many cofinal branches. The **weak Kurepa Hypothesis** $(\mathsf{wKH})$ is the assertion that such a tree exists. This weak form drops the countable-level requirement from the ordinary Kurepa hypothesis [2603.10539]. Recent work connects weak Kurepa phenomena to guessing-model principles. One theorem states that if $wAGP(\mu,\mu^+,\mu^+)$ holds for a regular uncountable $\mu$, then there are no weak $\mu$-Kurepa trees. The same paper proves, however, that starting from a supercompact cardinal and assuming $\mathsf{GCH}$ above it, there is a generic extension in which $$ 2^\omega=\kappa=\omega_2, $$ there is an $\omega_1$-Kurepa tree, and $$ (\omega_2,\omega_2,\geq\omega_2) $$ holds. This separates different tree and guessing principles sharply: some eliminate weak Kurepa trees, while others are compatible with classical $\omega_1$-Kurepa trees [2209.01925]. Further consistency separations strengthen this picture. Relative to a Mahlo cardinal, it is consistent that Martin’s axiom holds at $\omega_2$ while the weak Kurepa Hypothesis fails, and the supplied description states that the model is constructed so that $\mathsf{wKH}$ fails in any c.c.c. forcing extension [2411.04835]. Relative to a supercompact cardinal, it is consistent that $GMP$ holds together with an almost Kurepa Suslin tree; in the same line of work, there is also a model with $$ 2^\omega=\omega_2,\quad GMP^{\omega_2}(\omega_2),\quad \neg\mathsf{KH},\quad \mathsf{wKH}, $$ so there are no Kurepa trees but there is a weak Kurepa tree [2603.10539]. From an inaccessible cardinal, it is consistent that there exists a strongly non-saturated Aronszajn tree and there does not exist a weak Kurepa tree; from a Mahlo cardinal, one can moreover obtain c.c.c.-indestructible failure of the generic Kurepa hypothesis [2506.06878]. The cumulative effect is a highly stratified landscape. Ordinary Kurepa trees, weak Kurepa trees, almost Kurepa Suslin trees, and branch-guessing principles are not interchangeable; forcing can separate them in multiple directions. ## 5. Generalized and higher-cardinal Kurepa hypotheses For an infinite cardinal $\lambda$, the generalized Kurepa hypothesis $KH_\lambda$ asserts the existence of a family $\mathcal F\subseteq\mathcal P(\lambda)$ such that $|\mathcal F|>\lambda$ and for every infinite $X\subseteq\lambda$ with $|X|<\lambda$, $$ |\mathcal F\restriction X|\le |X|, \qquad \mathcal F\restriction X=\{t\cap X:t\in\mathcal F\}. $$ At singular cardinals this principle becomes delicate. It is shown that $\mathsf{GCH}$ does not imply $KH_{\aleph_\omega}$, nor the existence of a family $\mathcal F\subseteq[\aleph_\omega]^{\aleph_0}$ of size $\aleph_{\omega+1}$ such that $\mathcal F\restriction X$ has size $\aleph_0$ for every countable $X$. More specifically, under $$ \mathrm{GCH}+\mathrm{Chang's\ conjecture}\ (\aleph_{\omega+1},\aleph_\omega)\to(\aleph_1,\aleph_0), $$ $KH_{\aleph_\omega}$ fails [1712.02487]. Chang’s generalized Kurepa hierarchy admits further separation. For fixed finite $n\ge 1$, the **Gap-$n$-Kurepa hypothesis** is $$ \forall \lambda \text{ infinite},\quad \mathrm{KH}(\lambda^{+n},\lambda). $$ Relative to an inaccessible cardinal, one can force a model in which $\mathsf{GCH}$ holds, the Gap-$m$-Kurepa hypothesis holds for all $m\neq n$, but the Gap-$n$-Kurepa hypothesis fails. Conversely, if $\mathrm{KH}(\kappa,\lambda)$ fails for regular $\kappa$ and $\lambda<\kappa$, then $\kappa^+$ is inaccessible in $L$ [1510.02932]. The two-cardinal form $\mathsf{KH}(\kappa,\lambda)$ is formulated via slim families or, equivalently, slim $(\kappa,\lambda)$-trees. A slim $(\kappa,\lambda)$-family is a collection $\mathcal F\subseteq\mathcal P(\lambda)$ such that for every $x\in\mathcal P_\kappa(\lambda)$, $$ |\mathcal F\upharpoonright x|<\kappa. $$ The 2025 analysis isolates a key ZFC implication: $$ \lnot\mathsf{KH}(\kappa,\lambda)\land\mathsf{KH}(\kappa,\mu)\rightarrow\mathsf{KH}(\lambda^+,\mu) \qquad (\kappa\le\lambda<\mu). $$ For singular $\lambda$, failure of $\mathsf{KH}$ is tied to Chang’s Conjecture and to scales; for example, if $n<\omega$ and there is a scale on $\aleph_\omega$ that is good at all points of cofinality $\aleph_{n+1}$, then $$ \mathsf{KH}(\aleph_{n+1},\aleph_\omega) $$ holds [2510.08860]. ## 6. Model-theoretic and order-theoretic counterparts Kurepa trees have precise analogues in linear orders. A linear order $L$ of cardinality $\aleph_2$ is **Kurepa** if its density is $\aleph_1$ and the closure of every countable subset of $L$ is countable. The supplied summary states that Kurepa lines exist iff there are Kurepa trees. It is consistent with $\mathsf{ZFC}$ that there is an everywhere Kurepa line which is order-isomorphic to all of its dense $\aleph_2$-dense suborders and contains no Aronszajn suborder; it is also consistent that there is a minimal Kurepa line with no Aronszajn suborder [2310.12834]. In infinitary logic, a single $\mathcal L_{\omega_1,\omega}$-sentence $\psi$ can be constructed so as to code Kurepa trees. Its spectrum is controlled by the existence and exact branch numbers of Kurepa trees. The cited spectrum theorem gives $$ [\aleph_0,2^{\aleph_0}] \subseteq \operatorname{Spec}(\psi), $$ and if there exists a Kurepa tree with $\kappa$ branches, then $$ [\aleph_0,\kappa]\subseteq \operatorname{Spec}(\psi). $$ Conversely, if $\psi$ has a model of size $\kappa$, then either $\kappa\le 2^{\aleph_0}$ or there exists a Kurepa tree with $\kappa$ branches. The same sentence can consistently have right-open or right-closed spectrum and amalgamation spectrum, depending on the surrounding set theory [1705.05821]. This coding program extends to higher cardinals. For countable $\alpha$, an $\mathcal L_{\omega_1,\omega}$-sentence $\psi_\alpha$ is built to code $\aleph_{\alpha+1}$-Kurepa trees. One of the stated consistency results is that $$ 2^{\aleph_\alpha}<2^{\aleph_{\alpha+1}}, $$ $2^{\aleph_{\alpha+1}}$ is weakly inaccessible, and $$ \operatorname{Spec}(\psi_\alpha)=[\aleph_0,2^{\aleph_{\alpha+1}}). $$ The same work uses Kurepa-tree coding to show that $\kappa$-amalgamation for $\mathcal L_{\omega_1,\omega}$-sentences is not absolute [2410.19515]. These model-theoretic and order-theoretic developments indicate that Kurepa phenomena are not confined to tree combinatorics. They govern spectra, amalgamation, rigidity, and minimality in adjacent structures. ## 7. The left-factorial conjecture in number theory In number theory, Kurepa’s conjecture concerns the left factorial $$ !n=0!+1!+\cdots+(n-1)!. $$ The classical statement is $$ \gcd(!n,n!)=2 \qquad \text{for all } n>1, $$ which is equivalent to the claim that there is no odd prime $p$ such that $p\mid !p$. A large-scale computational search checked all primes $$ p<2^{34} $$ and found no counterexample. That work computed the residues $$ r_p=!p\bmod p $$ for all such primes, verified all residues previously reported in the literature, and found a new prime with residue $r_p=2$, $$ p=6855730873. $$ It also studied generalized left factorials $$ !^k n=(0!)^k+(1!)^k+\cdots+((n-1)!)^k $$ and proved that for every integer $15$, $$ K_p=\sum_{k=0}^{p-1}\frac{(-1)^k}{k!}. $$ That paper also proposed a stronger statement, namely $K_n\not\equiv 0\pmod n$ for all $n>7$. It proved the “even part” of this strong hypothesis, since $K_n$ is odd for every integer $n\ge 7$, but it also showed that the odd composite part fails: for odd composite $n\ge 9$, $$ 8K_n\equiv -S_{n-1}+2\pmod n, $$ and $$ 11563=31\cdot 373 $$ is a counterexample, with $$ K_{11563}\equiv 0\pmod{11563} $$ [1312.7037]. Later work develops additional reformulations. One paper introduces $$ \mathbb F_n(x)=\sum_{k=0}^n k!x^k, $$ identifies $\mathbb F_n(1)$ with the Kurepa factorial, and states an equivalent formulation $$ \gcd(\mathbb F_n,(n+1)!)=2. $$ The same paper gives decompositions in terms of Dobinski numbers, Bell numbers, derangement numbers, and complementary Bell numbers, and extends the discussion to partition functions and occupation-number formalisms [2509.06077]. By contrast, an arXiv abstract from 2022 states simply, “We answer Kurepa’s conjecture on the left factorials in affirmative,” but the supplied material for that item contains no theorem statements, proof, or further mathematical details beyond the abstract itself [2209.12677]. Within the supplied record, the number-theoretic Kurepa conjecture therefore appears in three simultaneously relevant forms: as a classical divisibility conjecture with extensive computational evidence, as a source of determinant and Bell-number reformulations, and as the subject of an abstract claiming a full affirmative solution. Source: https://www.emergentmind.com/topics/kurepa-conjecture