8.5.10 Monomorphisms and 2-Categorical Monomorphisms

    In this section, we characterise:

    More specifically:

    Essentially, a monomorphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ in $\mathsf{Rel}$ looks like this:

    In particular:

    1. 1.

      $R$ should contain an injection $f\colon A\hookrightarrow B$ embedding a copy of $A$ into $B$.

    2. 2.

      $R$ can be non-functional, mapping elements of $A$ to multiple elements of $B$ (but not to more than one in $\mathrm{Im}(f)$).

    3. 3.

      $R$ needs to be “weakly injective” in the sense that $R(a)=R(b)$ implies $a=b$, but it can fail to be injective.

    4. 4.

      $R$ doesn’t need to be surjective, so $B$ can have elements that aren’t in the image of $R$.


    1. 1Summary: As it turns out, every 1-morphism in $\boldsymbol {\mathsf{Rel}}$ is representably faithful and most other notions of 2-categorical monomorphism agree with the usual (1-categorical) notion of monomorphism.

    Let $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ be a relation. The following conditions are equivalent:1

    1. 1.

      The relation $R$ is a monomorphism in $\mathsf{Rel}$.

    2. 2.

      The direct image function

      \[ R_{!}\colon \mathcal{P}(A)\to \mathcal{P}(B) \]

      associated to $R$ is injective.

    3. 3.

      The coinverse image function

      \[ R_{-1}\colon \mathcal{P}(B)\to \mathcal{P}(A) \]

      associated to $R$ is surjective.

    4. 4.

      The inverse image function

      \[ R^{-1}\colon \mathcal{P}(B)\to \mathcal{P}(A) \]

      associated to $R$ is surjective.

    5. 5.

      The codirect image function

      \[ R_{*}\colon \mathcal{P}(A)\to \mathcal{P}(B) \]

      associated to $R$ is injective.

    6. 6.

      The direct image functor

      \[ R_{!}\colon (\mathcal{P}(A),\subset )\to (\mathcal{P}(B),\subset ) \]

      associated to $R$ is full.

    7. 7.

      The codirect image functor

      \[ R_{*}\colon (\mathcal{P}(A),\subset )\to (\mathcal{P}(B),\subset ) \]

      associated to $R$ is full.

    8. 8.

      The direct image functor

      \[ R_{!}\colon (\mathcal{P}(A),\subset )\to (\mathcal{P}(B),\subset ) \]

      associated to $R$ is fully faithful.

    9. 9.

      The coinverse image functor

      \[ R_{-1}\colon (\mathcal{P}(B),\subset )\to (\mathcal{P}(A),\subset ) \]

      associated to $R$ is essentially surjective.

    10. 10.

      The inverse image functor

      \[ R_{-1}\colon (\mathcal{P}(B),\subset )\to (\mathcal{P}(A),\subset ) \]

      associated to $R$ is essentially surjective.

    11. 11.

      The codirect image functor

      \[ R_{*}\colon (\mathcal{P}(A),\subset )\to (\mathcal{P}(B),\subset ) \]

      associated to $R$ is fully faithful.

    12. 12.

      For each pair of relations $S,T\colon X\mathrel {\rightrightarrows \kern -9.5pt\mathrlap {|}\kern 6pt}A$, the following condition is satisfied:

      • (★)
      • If $R\mathbin {\diamond }S\subset R\mathbin {\diamond }T$, then $S\subset T$.
    13. 13.

      There exists an injective function $f\colon A\hookrightarrow B$ satisfying the following conditions:2

      1. (a)

        We have $\operatorname {\mathrm{Gr}}(f)\subset R$.3

      2. (b)

        The diagram

        commutes.4

    14. 14.

      We have

      In other words, we have

      \[ U=\underbrace{\left\{ a\in A\ \middle |\ R(a)\subset R(U)\right\} }_{=R_{-1}(R_{!}(U))} \]

      for each $U\in \mathcal{P}(A)$.

    15. 15.

      We have

      In other words, we have

      \[ U=\underbrace{\left\{ a\in A\ \middle |\ \begin{aligned} & \text{there exists some $b\in R(a)$}\\ & \text{such that we have $R^{-1}(b)\subset U$} \end{aligned} \right\} }_{=R^{-1}(R_{*}(U))} \]

      for each $U\in \mathcal{P}(A)$.


    1. 1Item 2, Item 5, Item 6, and Item 7 unwind respectively to the following statements:
      • For each $U,V\in \mathcal{P}(A)$, if $R_{!}(U)=R_{!}(V)$, then $U=V$.

      • For each $U,V\in \mathcal{P}(A)$, if $R_{*}(U)=R_{*}(V)$, then $U=V$.

      • For each $U,V\in \mathcal{P}(A)$, if $R_{!}(U)\subset R_{!}(V)$, then $U\subset V$.

      • For each $U,V\in \mathcal{P}(A)$, if $R_{*}(U)\subset R_{*}(V)$, then $U\subset V$.

    2. 2We are assuming the axiom of choice for this item (Item 13).
    3. 3In other words, for each $a\in A$, we have $f(a)\in R(a)$.
    4. 4In other words, for each $a\in A$, we have $R^{-1}(f(a))=\left\{ a\right\} $.

    We will prove this by showing:

    Step 1: Item 2$\iff $Item 5 and Item 5$\iff $Item 7
    This follows from Item 17 of Proposition 8.7.1.1.4.

    Step 2: First Proof of Item 1$\iff $Item 2
    We claim that Item 1 and Item 2 are equivalent:

    • Item 1$\implies $Item 2: Let $U,V\in \mathcal{P}(A)$ and consider the diagram

      By Remark 8.7.1.1.3, we have

      \begin{align*} R_{!}(U) & = R\mathbin {\diamond }U,\\ R_{!}(V) & = R\mathbin {\diamond }V. \end{align*}

      Now, if $R\mathbin {\diamond }U=R\mathbin {\diamond }V$, i.e. $R_{!}(U)=R_{!}(V)$, then $U=V$ since $R$ is assumed to be a monomorphism, showing $R_{!}$ to be injective.

    • Item 2$\implies $Item 1: Conversely, suppose that $R_{!}$ is injective, consider the diagram

      and suppose that $R\mathbin {\diamond }S=R\mathbin {\diamond }T$. Note that, since $R_{!}$ is injective, given a diagram of the form
      if $R_{!}(U)=R\mathbin {\diamond }U=R\mathbin {\diamond }V=R_{!}(V)$, then $U=V$. In particular, for each $x\in X$, we may consider the diagram
      where we have $R\mathbin {\diamond }S\mathbin {\diamond }[x]=R\mathbin {\diamond }T\mathbin {\diamond }[x]$, implying that we have

      \[ S(x)=S\mathbin {\diamond }[x]=T\mathbin {\diamond }[x]=T(x) \]

      for each $x\in X$. Thus $S=T$ and $R$ is a monomorphism.

    Step 2.5: Second Proof of Item 1$\iff $Item 2
    A more abstract proof can also be given, following [Yuan, Mono's and epi's in the category Rel?]:

    • Proposition 8.5.10.1.2$\implies $Proposition 8.5.10.1.3: Assume that $R$ is a monomorphism.

      • We first notice that the functor $\mathrm{Rel}(\mathrm{pt},-)\colon \mathrm{Rel}\to \mathsf{Sets}$ maps $R$ to $R_{!}$ by Remark 8.7.1.1.3.

      • Since $\mathrm{Rel}(\mathrm{pt},-)$ preserves all limits by Unresolved reference, Unresolved reference of Unresolved reference, it follows by Unresolved reference, Unresolved reference of Unresolved reference that $\mathrm{Rel}(\mathrm{pt},-)$ also preserves monomorphisms.

      • Since $R$ is a monomorphism and $\mathrm{Rel}(\mathrm{pt},-)$ maps $R$ to $R_{!}$, it follows that $R_{!}$ is also a monomorphism.

      • Since the monomorphisms in $\mathsf{Sets}$ are precisely the injections (Unresolved reference, Unresolved reference of Unresolved reference), it follows that $R_{!}$ is injective.

    • Proposition 8.5.10.1.3$\implies $Proposition 8.5.10.1.2: Assume that $R_{!}$ is injective.

      • We first notice that the functor $\mathrm{Rel}(\mathrm{pt},-)\colon \mathrm{Rel}\to \mathsf{Sets}$ maps $R$ to $R_{!}$ by Remark 8.7.1.1.3.

      • Since the monomorphisms in $\mathsf{Sets}$ are precisely the injections (Unresolved reference, Unresolved reference of Unresolved reference), it follows that $R_{!}$ is a monomorphism.

      • Since $\mathrm{Rel}(\mathrm{pt},-)$ is faithful, it follows by Unresolved reference, Unresolved reference of Unresolved reference that $\mathrm{Rel}(\mathrm{pt},-)$ reflects monomorphisms.

      • Since $R_{!}$ is a monomorphism and $\mathrm{Rel}(\mathrm{pt},-)$ maps $R$ to $R_{!}$, it follows that $R$ is also a monomorphism.

    Step 3: Item 2$\iff $Item 6
    We claim that Item 2 and Item 6 are equivalent:

    • Item 2$\implies $Item 6: We proceed in a few steps:

      • Let $U,V\in \mathcal{P}(A)$ such that $R_{!}(U)\subset R_{!}(V)$, assume $R_{!}$ to be injective, and consider the set $U\cup V$.

      • Since $R_{!}(U)\subset R_{!}(V)$, we have

        \begin{align*} R_{!}(U\cup V) & = R_{!}(U)\cup R_{!}(V)\\ & = R_{!}(V), \end{align*}

        where we have used Item 13 of Proposition 8.7.1.1.4 for the first equality.

      • Since $R_{!}$ is injective, this implies $U\cup V=V$.

      • Thus $U\subset V$, as we wished to show.

    • Item 2$\implies $Item 6: We proceed in a few steps:

      • Suppose Item 6 holds, and let $U,V\in \mathcal{P}(A)$ such that $R_{!}(U)=R_{!}(V)$.

      • Since $R_{!}(U)=R_{!}(V)$, we have $R_{!}(U)\subset R_{!}(V)$ and $R_{!}(V)\subset R_{!}(U)$.

      • By assumption, this implies $U\subset V$ and $V\subset U$.

      • Thus $U=V$, showing $R_{!}$ to be injective.

    Step 4: Item 6$\iff $Item 3
    By Unresolved reference, a right adjoint functor between posets, such as $R_{-1}$, is surjective iff its left adjoint, in this case $R_{!}$, is full.

    Step 5: Item 1$\iff $Item 4
    By Item 5 of Proposition 8.7.1.1.5, $R^{-1}$ is surjective iff $R^{\dagger }_{!}$ is surjective. By Item 7 of Proposition 8.5.11.1.2, this happens iff $R^{\dagger }$ is an epimorphism. By the self-duality of $\mathsf{Rel}$ (Proposition 8.5.1.1.1) coupled with Unresolved reference, this happens iff $R$ is a monomorphism.

    Step 6: Item 2$\iff $Item 8
    This follows from the fact that $\mathcal{P}(B)$ is locally posetal.

    Step 7: Item 3$\iff $Item 9
    This follows from the fact that $\mathcal{P}(A)$ is locally posetal.

    Step 8: Item 4$\iff $Item 10
    This follows from the fact that $\mathcal{P}(A)$ is locally posetal.

    Step 9: Item 5$\iff $Item 11
    This follows from the fact that $\mathcal{P}(B)$ is locally posetal.

    Step 10: Item 6$\iff $Item 12
    We claim that Item 6 and Item 12 are equivalent:

    • Item 6$\implies $Item 12: Let $U,V\in \mathcal{P}(A)$ and consider the diagram

      By Remark 8.7.1.1.3, we have

      \begin{align*} R_{!}(U) & = R\mathbin {\diamond }U,\\ R_{!}(V) & = R\mathbin {\diamond }V. \end{align*}

      Now, if $R\mathbin {\diamond }U\subset R\mathbin {\diamond }V$, then $R_{!}(U)\subset R_{!}(V)$. By assumption, we then have $U\subset V$.

    • Item 12$\implies $Item 6: Consider the diagram

      and suppose that $R\mathbin {\diamond }S\subset R\mathbin {\diamond }T$. Note that, by assumption, given a diagram of the form
      if $R_{!}(U)=R\mathbin {\diamond }U\subset R\mathbin {\diamond }V=R_{!}(V)$, then $U\subset V$. In particular, for each $x\in X$, we may consider the diagram
      for which we have $R\mathbin {\diamond }S\mathbin {\diamond }[x]\subset R\mathbin {\diamond }T\mathbin {\diamond }[x]$, implying that we have

      \[ S(x)=S\mathbin {\diamond }[x]\subset T\mathbin {\diamond }[x]=T(x) \]

      for each $x\in X$, implying $S\subset T$.

    This finishes the proof.

    Step 11: Item 1$\iff $Item 13
    We defer this proof to Corollary 8.5.10.1.4.

    Step 12: Item 1$\iff $Item 14
    We defer this proof to Corollary 8.5.10.1.4.

    Step 13: Item 1$\iff $Item 15
    We defer this proof to Corollary 8.5.10.1.5.

    Let $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ be a relation. The following conditions are equivalent:

    1. 1.

      The relation $R$ is a monomorphism in $\mathsf{Rel}$.

    2. 2.

      For each $a\in A$ and each $U\in \mathcal{P}(A)$, if $R(a)\subset R(U)$, then $a\in U$.

    3. 3.

      For each $a\in A$, there exists some $b\in B$ such that $R^{-1}(b)=\left\{ a\right\} $.

    We will prove this by showing:

    Step 1: Item 1$\implies $Item 2
    We proceed in a few steps:

    • If $R$ is a monomorphism, then, by Item 2 of Proposition 8.5.10.1.2, the functor

      \[ R_{!}\colon (\mathcal{P}(A),\subset )\to (\mathcal{P}(B),\subset ) \]

      is full.

    • As a result, given $a\in A$ and $U\in \mathcal{P}(A)$ such that $R(a)\subset R(U)$, it follows that $\left\{ a\right\} \subset U$.

    • Thus, we have $a\in U$.

    Step 2: Item 2$\implies $Item 3
    We proceed in a few steps:

    • Let $a\in A$ and consider the subset $U=A\setminus \left\{ a\right\} $.

    • Since $a\not\in U$, we have $R(a)\centernot {\subset }R(U)$ by the contrapositve of Item 2.

    • As a result, there must exist some $b\in R(a)$ with $b\not\in R(U)$.

    • In particular, we have $a\in R^{-1}(b)$.

    • Moreover, the condition $b\not\in R(U)=R(A\setminus \left\{ a\right\} )$ means that, if $a'\in A\setminus \left\{ a\right\} $, then $a'\not\in R^{-1}(b)$.

    • Thus $R^{-1}(b)=\left\{ a\right\} $.

    Step 3: Item 3$\implies $Item 1
    We proceed in a few steps:

    • By the equivalence between Item 1 and Item 6 of Proposition 8.5.10.1.2, to show that $R$ is a monomorphism it suffices to prove that, for each $U,V\in \mathcal{P}(A)$, if $R(U)\subset R(V)$, then $U\subset V$.

    • So let $u\in U$ and assume $R(U)\subset R(V)$.

    • By assumption, there exists some $b\in B$ with $R^{-1}(b)=\left\{ u\right\} $.

    • In particular, $b\in R(U)$.

    • Since $R(U)\subset R(V)$, we also have $b\in R(V)$.

    • Thus, there exists some $v\in V$ with $b\in R(v)$.

    • However, $R^{-1}(b)=\left\{ u\right\} $, so we must in fact have $v=u$.

    • Therefore $u\in V$, showing that $U\subset V$.

    This finishes the proof.

    We claim that Item 3 of Proposition 8.5.10.1.3 is equivalent to Item 13 of Proposition 8.5.10.1.2:

    • Item 3 of Proposition 8.5.10.1.3$\implies $Item 13: By assumption, given $a\in A$, there exists some $b\in B$ such that $R^{-1}(b)=\left\{ a\right\} $. Invoking the axiom of choice, we may pick one such $b$ for each $a\in A$, giving us our desired function $f\colon A\to B$. All the requirements listed in Item 13 of Proposition 8.5.10.1.2 then follow by construction.

    • Item 13$\implies $Item 3 of Proposition 8.5.10.1.3: Given $a\in A$, we may pick $b=f(a)$, in which case $R^{-1}(f(a))$ will be equal to $\left\{ a\right\} $ by assumption.

    By Proposition 8.5.10.1.3, Item 3 of Proposition 8.5.10.1.3 is equivalent to Item 1 of Proposition 8.5.10.1.3. Since Item 1 of Proposition 8.5.10.1.3 is exactly the same condition as Item 1 of Proposition 8.5.10.1.2, the result follows.

    Indeed, we have

    \begin{align*} [R_{-1}\circ R_{!}](U) & \mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}R_{-1}(R_{!}(U))\\ & \mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}\left\{ a\in A\ \middle |\ R(a)\subset R(U)\right\} \end{align*}

    for each $U\in \mathcal{P}(A)$. As a result, the condition $R_{-1}\circ R_{!}=\operatorname {\mathrm{id}}_{\mathcal{P}(A)}$ becomes

    \[ \left\{ a\in A\ \middle |\ R(a)\subset R(U)\right\} =U, \]

    which holds precisely when Item 2 of Proposition 8.5.10.1.3 does. By Proposition 8.5.10.1.3, that in turn holds precisely if Item 1 of Proposition 8.5.10.1.3 holds. Since Item 1 of Proposition 8.5.10.1.3 is exactly the same condition as Item 1 of Proposition 8.5.10.1.2, the result follows.

    Item 1$\implies $Item 15
    We proceed by taking a specific choice of subset $U$:

    • Let $a$ be an arbitrary element of $A$. By our assumption, the condition $R^{-1}(R_{*}(U))=U$ must hold for the singleton set $U=\left\{ a\right\} $.

    • From $R^{-1}(R_{*}(\left\{ a\right\} ))=\left\{ a\right\} $, it follows that $a\in R^{-1}(R_{*}(\left\{ a\right\} ))$.

    • This means there must exist some $b\in R(a)$ such that $R^{-1}(b)\subset \left\{ a\right\} $.

    • The condition $b\in R(a)$ implies that $a\in R^{-1}(b)$. Therefore, $R^{-1}(b)$ is a non-empty subset of $\left\{ a\right\} $.

    • The only non-empty subset of $\left\{ a\right\} $ is $\left\{ a\right\} $ itself.

    • Thus, we must have $R^{-1}(b)=\left\{ a\right\} $.

    By Item 3 of Proposition 8.5.10.1.3, it then follows that $R$ is a monomorphism.

    Item 15$\implies $Item 1
    By Item 3 of Proposition 8.5.10.1.3, for each $a\in A$, there exists some $b\in B$ such that $R^{-1}(b)=\left\{ a\right\} $. We need to show that $R^{-1}(R_{*}(U))=U$ for any $U\in \mathcal{P}(A)$, which requires proving two set inclusions.

    • $R^{-1}(R_{*}(U))\subset U$: We proceed in a few steps:

      • Let $a\in R^{-1}(R_{*}(U))$.

      • By definition, there exists some $b\in R(a)$ such that $R^{-1}(b)\subset U$.

      • Since $b\in R(a)$ implies $a\in R^{-1}(b)$, it follows immediately that $a\in U$.

      • Thus, $R^{-1}(R_{*}(U))\subset U$.

    • $U\subset R^{-1}(R_{*}(U))$: Let $a\in U$. By assumption, there exists an element $b\in B$ such that $R^{-1}(b)=\left\{ a\right\} $. Thus $R^{-1}(b)\subset U$, so $a\in R^{-1}(R_{*}(U))$.

    Combining both inclusions gives $R^{-1}(R_{*}(U))=U$.

    The following conditions are equivalent and imply $R$ is a monomorphism, but the converse may fail. Thus, they are not equivalent to $R$ being a monomorphism:

    1. 1.

      The relation $R$ is total and injective.

    2. 2.

      The diagram

      commutes. In other words, we have

      \[ R^{-1}(R(a))=\left\{ a\right\} \]

      for each $a\in A$.

    3. 3.

      We have

      In other words, we have

      \[ U=\underbrace{\left\{ a\in A\ \middle |\ R(a)\cap R(U)\neq \text{Ø}\right\} }_{=R^{-1}(R_{!}(U))} \]

      for each $U\in \mathcal{P}(A)$.

    4. 4.

      We have

      In other words, we have

      \[ U=\underbrace{\left\{ a\in A\ \middle |\ R^{-1}(R(a))\subset U\right\} }_{=R_{-1}(R_{*}(U))} \]

      for each $U\in \mathcal{P}(A)$.

    Moreover, if $R$ is a monomorphism, then it satisfies the following condition, but the converse may fail:

    • (★)
    • For each $a,a'\in A$, if $R(a)=R(a')$, then $a=a'$.

    We proceed by showing:

    • Step 1: Item 1$\implies $Item 2.

    • Step 2: Item 2$\implies $Item 1.

    • Step 3: Item 2$\implies $Item 4.

    • Step 4: Item 4$\implies $Item 2.

    • Step 5: Item 1$\implies $Item 3.

    • Step 6: Item 3$\implies $Item 1.

    • Step 7: $R$ is a monomorphism $\centernot {\implies }$Item 1.

    • Step 8: Item 1$\implies $ $R$ is a monomorphism.

    • Step 9: Proof of $(\star )$.

    • Step 10: Counterexample to the converse of $(\star )$.

    Step 1: Item 1$\implies $Item 2
    Suppose that $R$ is total and injective. Given $a\in A$, we wish to show that $R^{-1}(R(a))=\left\{ a\right\} $.

    • Let $x\in R^{-1}(R(a))$. By definition, this means $R(x)\cap R(a)\neq \text{Ø}$.

    • Since the image sets are pairwise disjoint, this can only be true if $x=a$.

    • Therefore, $R^{-1}(R(a))\subset \left\{ a\right\} $.

    • Since $R$ is total, $R(a)$ is non-empty.

    • Thus $R(a)\cap R(a)\neq \text{Ø}$, which implies $a\in R^{-1}(R(a))$.

    • Therefore, $\left\{ a\right\} \subset R^{-1}(R(a))$.

    Combining both inclusions, we have $R^{-1}(R(a))=\left\{ a\right\} $.

    Step 2: Item 2$\implies $Item 1
    By definition,

    \[ R^{-1}(R(a))=\left\{ x\in A \mid R(x)\cap R(a)\neq \text{Ø}\right\} . \]

    The condition $R^{-1}(R(a))=\left\{ a\right\} $ implies two facts:

    • Totality. The element $a$ must belong to the set $\left\{ x\in A \mid R(x)\cap R(a)\neq \text{Ø}\right\} $. For this to be true, the condition must hold for $x=a$, so $R(a)\cap R(a)\neq \text{Ø}$. This is equivalent to $R(a)\neq \text{Ø}$. Since this must hold for all $a\in A$, the relation $R$ is total.

    • Injectivity. Any element $x\in A$ such that $x\neq a$ must not belong to the set. This means that for any $x\neq a$, we must have $R(x)\cap R(a)=\text{Ø}$. This means the image sets of distinct elements of $A$ are pairwise disjoint.

    Thus, $R$ is total and injective.

    Step 3: Item 2$\implies $Item 4
    We have

    \begin{align*} R^{-1}(R_{*}(U)) & = \left\{ a\in A\ \middle |\ R^{-1}(R(a))\subset U\right\} \\ & = \left\{ a\in A\ \middle |\ \left\{ a\right\} \subset U\right\} \\ & = U. \end{align*}

    Step 4: Item 4$\implies $Item 2
    Let $a\in A$.

    • First, let $U=\left\{ a\right\} $. By assumption, we have

      \[ \left\{ a\right\} =\left\{ a'\in A\ \middle |\ R^{-1}(R(a'))\subset \left\{ a\right\} \right\} . \]

      Since $a$ is in the set on the left-hand side, it must also be in the set on the right-hand side. Thus $R^{-1}(R(a'))\subset \left\{ a\right\} $ must be true.

    • Next, consider the complement $U=A\setminus \left\{ a\right\} $. By assumption, we have

      \[ A\setminus \left\{ a\right\} =\left\{ a'\in A\ \middle |\ R^{-1}(R(a'))\subset A\setminus \left\{ a\right\} \right\} \]

      Since $a$ is not in the set on the left-hand side, it cannot be in the set on the right-hand side. Thus $R^{-1}(R(a))\centernot {\subset }A\setminus \left\{ a\right\} $.

    • The statement $R^{-1}(R(a))\centernot {\subset }A\setminus \left\{ a\right\} $ implies that there exists an element $x\in R^{-1}(R(a))$ such that $x\not\in A\setminus \left\{ a\right\} $. The only such element is $a$, so we must have $a\in R^{-1}(R(a))$.

    • Combining these two results, namely $R^{-1}(R(a))\subset \left\{ a\right\} $ and $a\in R^{-1}(R(a))$, we conclude that $R^{-1}(R(a))=\left\{ a\right\} $, as we wished to show.

    Step 5: Item 1$\implies $Item 3
    Assume that $R$ is total and injective. Let $S(U)\mathrel {\smash {\overset {\mathclap {\scriptscriptstyle \text{def}}}=}}\left\{ a\in A \mid R(a)\cap R(U)\neq \text{Ø}\right\} $. We need to show that $U = S(U)$ for any $U \in \mathcal{P}(A)$ by proving double inclusion.

    • $U\subset S(U)$: Let $u\in U$.

      • Since $R$ is total, we have $R(u)\neq \text{Ø}$.

      • By definition, $R(U)=\bigcup _{x\in U}R(x)$, so $R(u)\subset R(U)$.

      • Therefore, $R(u)\cap R(U)=R(u)$.

      • Since $R(u)\neq \text{Ø}$, we have $R(u)\cap R(U)\neq \text{Ø}$.

      • By the definition of $S(U)$, it follows that $u\in S(U)$.

    • $S(U)\subset U$: Let $a\in S(U)$.

      • By assumption, $R(a)\cap R(U)\neq \text{Ø}$.

      • This means $R(a)\cap \bigcup _{u\in U}R(u)\neq \text{Ø}$.

      • Using the distributivity of intersection over union (Chapter 4: Constructions With Sets, Item 8 of Proposition 4.3.6.1.2), this is equivalent to

        \[ \bigcup _{u\in U}(R(a)\cap R(u))\neq \text{Ø}. \]
      • For this union of sets to be non-empty, at least one of the sets in the union must be non-empty. Thus, there must exist some $u\in U$ such that $R(a)\cap R(u)\neq \text{Ø}$.

      • Since $R$ is injective, the images of distinct elements are disjoint. Thus, for the intersection $R(a)\cap R(u)$ to be non-empty, we must therefore have $a=u$.

      • Since $u\in U$, it follows that $a\in U$.

    As both inclusions hold, we conclude that $U=S(U)$.

    Step 6: Item 3$\implies $Item 1
    Assume that for every $U\in \mathcal{P}(A)$, we have $U=\left\{ a\in A\ \middle |\ R(a)\cap R(U)\neq \text{Ø}\right\} $. We must show that $R$ is both total and injective.

    • Totality: Let $a\in A$. We must show that $R(a)\neq \text{Ø}$.

      • Consider the singleton set $U=\left\{ a\right\} $.

      • By assumption, $U=\left\{ x\in A\ \middle |\ R(x)\cap R(U)\neq \text{Ø}\right\} $.

      • Since $a\in U$, we must have $R(a)\cap R(U)\neq \text{Ø}$.

      • Substituting $U=\left\{ a\right\} $, we get $R(a)\cap R(\left\{ a\right\} )\neq \text{Ø}$.

      • Since $R(\left\{ a\right\} )=R(a)$, this simplifies to $R(a)\cap R(a)=R(a)\neq \text{Ø}$.

      • Thus $R(a)\neq \text{Ø}$ for all $a\in A$, showing $R$ to be total.

    • Injectivity: Let $a,a'\in A$ such that $a\neq a'$. We must show that $R(a)\cap R(a')=\text{Ø}$.

      • Consider the singleton set $U=\left\{ a\right\} $.

      • By assumption, $U=\left\{ x\in A\ \middle |\ R(x)\cap R(U)\neq \text{Ø}\right\} $.

      • Since $a\neq a'$, we have $a'\not\in U$.

      • Therefore, $a'$ cannot satisfy the membership condition for $U$. This means $R(a')\cap R(U)=\text{Ø}$.

      • Substituting $U=\left\{ a\right\} $, we get $R(a')\cap R(\left\{ a\right\} )=\text{Ø}$, which simplifies to $R(a')\cap R(a)=\text{Ø}$.

      • As this holds for any pair of distinct elements, the relation $R$ is injective.

    This completes the proof.

    Step 7: $R$ Is a Monomorphism $\centernot {\implies }$Item 1
    The relation depicted in Explanation 8.5.10.1.1 isn’t total and injective, but it is a monomorphism in $\mathsf{Rel}$.

    Step 8: Item 1$\implies $ $R$ Is a Monomorphism
    Consider the diagram
    where $R\mathbin {\diamond }S=R\mathbin {\diamond }T$, and let $x\in X$ and $a\in A$ such that $x\sim _{S}a$.

    • Since $R$ is total and $a\in A$, there exists some $b\in B$ such that $a\sim _{R}b$.

    • In this case, we have $x\sim _{R\mathbin {\diamond }S}b$, and since $R\mathbin {\diamond }S=R\mathbin {\diamond }T$, we have also $x\sim _{R\mathbin {\diamond }T}b$.

    • Thus there must exist some $a'\in A$ such that $x\sim _{T}a'$ and $a'\sim _{R}b$.

    • However, since $a\sim _{R}b$ and $a'\sim _{R}b$, we must have $a=a'$ since $R$ is injective.

    • Thus $x\sim _{T}a$ as well.

    • A similar argument shows that if $x\sim _{T}a$, then $x\sim _{S}a$.

    • Thus $S=T$, showing $R$ to be a monomorphism.

    This finishes the proof.

    Step 9: Proof of $(\star )$
    This follows from Item 2 of Proposition 8.5.10.1.2.

    Step 10: Counterexample to the converse of $(\star )$
    Let $A=\left\{ a_{1},a_{2},a_{3}\right\} $, let $B=\left\{ b_{1},b_{2}\right\} $, and let $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ be the relation defined by

    \begin{align*} R(a_{1}) & = \left\{ b_{1}\right\} ,\\ R(a_{2}) & = \left\{ b_{2}\right\} ,\\ R(a_{3}) & = \left\{ b_{1},b_{2}\right\} . \end{align*}

    Since $R(a_{1})$, $R(a_{2})$, and $R(a_{3})$ are all distinct, $R$ satisfies the condition in $(\star )$.

    Taking $U=\left\{ a_{1},a_{2}\right\} $ and $V=\left\{ a_{3}\right\} $, however, shows that $R(U)=R(V)$ holds while $U\neq V$, so $R$ can’t be a monomorphism.

    Taking the contrapositive of Item 2 of Proposition 8.5.10.1.3 and letting $U=\left\{ a'\right\} $ shows that the subset

    \[ \left\{ R(a)\in \mathcal{P}(B)\ \middle |\ a\in A\right\} \]

    of $\mathcal{P}(B)$ forms an antichain in $\mathcal{P}(B)$. The converse however, fails.

    Every 1-morphism in $\boldsymbol {\mathsf{Rel}}$ is representably faithful.

    A relation $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ will be representably faithful in $\boldsymbol {\mathsf{Rel}}$ iff, for each $X\in \operatorname {\mathrm{Obj}}(\boldsymbol {\mathsf{Rel}})$, the functor

    \[ R_{*}\colon \mathbf{Rel}(X,A)\to \mathbf{Rel}(X,B) \]

    given by postcomposition by $R$ is faithful. This happens iff the morphism

    \[ R_{*|S,T}\colon \operatorname {\mathrm{Hom}}_{\mathbf{Rel}(X,A)}(S,T)\to \operatorname {\mathrm{Hom}}_{\mathbf{Rel}(X,B)}(R\mathbin {\diamond }S,R\mathbin {\diamond }T) \]

    is injective for each $S,T\in \operatorname {\mathrm{Obj}}(\mathbf{Rel}(X,A))$.

    However, since $\boldsymbol {\mathsf{Rel}}$ is locally posetal, the Hom-set $\operatorname {\mathrm{Hom}}_{\mathbf{Rel}(X,A)}(S,T)$ is either empty or a singleton. As a result, the map $R_{*|S,T}$ will necessarily be injective in either of these cases.

    Let $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ be a relation. The following conditions are equivalent:

    1. 1.

      The morphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ is a monomorphism in $\mathsf{Rel}$.

  • 2.

    The 1-morphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ is representably full in $\boldsymbol {\mathsf{Rel}}$.

  • 3.

    The 1-morphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ is representably fully faithful in $\boldsymbol {\mathsf{Rel}}$.

  • 4.

    The 1-morphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ is pseudomonic in $\boldsymbol {\mathsf{Rel}}$.

  • 5.

    The 1-morphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ is representably essentially injective in $\boldsymbol {\mathsf{Rel}}$.

  • 6.

    The 1-morphism $R\colon A\mathrel {\rightarrow \kern -9.5pt\mathrlap {|}\kern 6pt}B$ is representably conservative in $\boldsymbol {\mathsf{Rel}}$.

  • We will prove this by showing:

    Step 1: Item 1$\iff $Item 2
    The condition that $R$ is representably full corresponds precisely to Item 12 of Proposition 8.5.10.1.2, so this follows by Proposition 8.5.10.1.2.

    Step 2: Item 2$\iff $Item 3
    This follows from Step 1 and Proposition 8.5.10.1.8.

    Step 3: Item 3$\iff $Item 4$\iff $Item 5$\iff $Item 6
    Since $\boldsymbol {\mathsf{Rel}}$ is locally posetal, the conditions in Item 4, Item 5, and Item 6 all collapse to the one of Item 3.


Noticed something off, or have any comments? Feel free to reach out!


You can also use the contact form below: