Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions .cspell.json
Original file line number Diff line number Diff line change
Expand Up @@ -171,6 +171,7 @@
"groupoid",
"groupoids",
"Haus",
"hausdorff",
"Hertweck",
"Heyting",
"homotopic",
Expand Down
17 changes: 7 additions & 10 deletions database/data/functors/discrete_topology.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ name: discrete topology functor
notation: $D$
domain: Set
codomain: Top
description: This functor maps a set $X$ to the discrete topological space $D(X) \coloneqq (X, P(X))$ in which every subset is open.
description: This functor maps a set $X$ to the discrete topological space $D(X) \coloneqq (X, P(X))$ in which every subset is open. It is a typical example of a fully faithful functor that preserves finite but does not preserve infinite products.
nlab_link: https://ncatlab.org/nlab/show/discrete+and+indiscrete+topology
left_adjoint: null

Expand All @@ -15,24 +15,21 @@ related:
- indiscrete_topology

satisfied_properties:
- property: fully faithful
proof: This is trivial.

- property: left adjoint
proof: 'This functor is left adjoint to the forgetful functor $U_{\Top} : \Top \to \Set$.'

- property: left-invertible
proof: The forgetful functor provides a left inverse.

- property: full
proof: This is trivial.

- property: preserves finite products
proof: It is easy to check that a finite product of discrete spaces is again discrete.

- property: preserves equalizers
proof: This follows since a subspace of a discrete space is again discrete.

unsatisfied_properties:
- property: preserves products
proof: An infinite product of discrete spaces is usually not discrete, e.g. $\{0,1\}^{\IN}$.

- property: dominant
proof: A subspace of a discrete space is again discrete. Hence, if $Y$ is any non-discrete space such as $\IR$ with the Euclidean topology, there is no set $X$ with a regular (or even split) monomorphism $Y \hookrightarrow D(X)$.

- property: preserves products
proof: An infinite product of discrete spaces is usually not discrete, e.g. $\{0,1\}^{\IN}$.
34 changes: 34 additions & 0 deletions database/data/functors/empty_sets.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,34 @@
id: empty_sets
name: empty functor to the category of sets
notation: $!_{\Set}$
domain: '0'
codomain: Set
description: 'Every category $\C$ has a unique functor $!_{\C} : \varnothing \to \C$. Here, we specify $\C = \Set$, but most of the properties do not depend on the choice of $\C$, as long as $\C$ is non-empty. This is the simplest example of a functor to $\Set$ that is both continuous and cocontinuous, but is neither representable nor a left or right adjoint.'
nlab_link: null
left_adjoint: null

tags:
- set theory

related:
- trivial_sets

satisfied_properties:
- property: fully faithful
proof: This is trivial.

- property: continuous
proof: 'For every category $\C$ the unique functor $!_{\C} : \varnothing \to \C$ is continuous, since no diagram in $\varnothing$ has a limit.'

- property: cocontinuous
proof: 'For every category $\C$ the unique functor $!_{\C} : \varnothing \to \C$ is cocontinuous, since no diagram in $\varnothing$ has a colimit.'

unsatisfied_properties:
- property: left-invertible
proof: There is no functor $\Set \to \varnothing$ at all.

- property: dominant
proof: This is trivial.

- property: representable
proof: There cannot be any representing object.
6 changes: 2 additions & 4 deletions database/data/functors/forget_abelian.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -15,17 +15,15 @@ related:
- forget_group
- forget_inverses
- forget_torsion_free
- forget_commutative

satisfied_properties:
- property: full
- property: fully faithful
proof: This is trivial.

- property: right adjoint
proof: The abelianization functor $\Grp \to \Ab$, $G \mapsto G^{\ab}$ provides a left adjoint.

- property: left-invertible
proof: The abelianization functor $\Grp \to \Ab$, $G \mapsto G^{\ab}$ provides a left inverse.

- property: preserves coequalizers
proof: 'This follows from the concrete construction of coequalizers, but it can also be proven as follows: Assume that $f,g : A \rightrightarrows B$ have coequalizer $p : B \to C$ in $\Ab$. If $G$ is any group and $h : B \to G$ is a homomorphism with $hf = hg$, consider the image factorization $h = i h''$ of $h$. We have $h'' f = h'' g$, and the image $\im(h)$ is abelian. Hence, there is a unique homomorphism $k : C \to \im(h)$ with $kp = h''$. Hence, $ik : C \to G$ satisfies $ikp = h$. Uniqueness follows since we already know that the forgetful functor is preserves epimorphisms.'

Expand Down
41 changes: 41 additions & 0 deletions database/data/functors/forget_commutative.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,41 @@
id: forget_commutative
name: forgetful functor from commutative rings to rings
notation: $U_{\CRing,\Ring}$
domain: CRing
codomain: Ring
description: This is the inclusion functor $\CRing \hookrightarrow \Ring$ that maps a commutative ring to itself, regarded merely as a ring.
nlab_link: https://ncatlab.org/nlab/show/forgetful+functor
left_adjoint: null # TODO: add the left adjoint to the database

tags:
- algebra
- forgetful

related:
- forget_abelian

satisfied_properties:
- property: fully faithful
proof: This is trivial.

- property: right adjoint
proof: 'The functor $\Ring \to \CRing$, $R \mapsto R/\langle xy - yx : x,y \in R \rangle$ provides a left adjoint.'

- property: preserves initial objects
proof: In both categories, the ring $\IZ$ is initial.

- property: finitary
proof: 'Since the forgetful functor $U_{\Ring} : \Ring \to \Set$ is finitary and conservative, it suffices to prove that the composition $U_{\Ring} \circ U_{\CRing,\Ring} : \CRing \to \Set$ is finitary. This is just the forgetful functor $U_{\CRing} : \CRing \to \Set$ and is therefore finitary.'

- property: preserves coequalizers
proof: 'First of all, the functor preserves regular epimorphisms, since they are surjective homomorphisms in both categories. Now assume that $f,g : A \rightrightarrows B$ have coequalizer $p : B \to C$ in $\CRing$. If $T$ is any ring and $h : B \to T$ is a homomorphism with $hf = hg$, consider the image factorization $h = i h''$ of $h$. We have $h'' f = h'' g$, and the image $\im(h)$ is commutative. Hence, there is a unique homomorphism $k : C \to \im(h)$ with $kp = h''$. Hence, $ik : C \to G$ satisfies $ikp = h$. Uniqueness follows since we already know that the forgetful functor preserves regular epimorphisms.'

- property: preserves epimorphisms
proof: This is not formal. See <a href="https://math.stackexchange.com/questions/5133488" target="_blank">MSE/5133488</a> for a proof.

unsatisfied_properties:
- property: dominant
proof: This is trivial.

- property: preserves binary coproducts
proof: The coproduct of $\IZ[X]$ and $\IZ[Y]$ in $\CRing$ is the polynomial ring $\IZ[X,Y]$ in which $X,Y$ commute. Their coproduct in $\Ring$ is the non-commutative polynomial ring $\IZ\langle X,Y \rangle$ in which $X,Y$ do not commute.
1 change: 1 addition & 0 deletions database/data/functors/forget_finite.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,7 @@ tags:

related:
- id_Set
- forget_finite_group

satisfied_properties:
- property: fully faithful
Expand Down
42 changes: 42 additions & 0 deletions database/data/functors/forget_finite_abelian_group.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,42 @@
id: forget_finite_abelian_group
name: forgetful functor from finite abelian groups to abelian groups
notation: $U_{\FinAb, \Ab}$
domain: FinAb
codomain: Ab
description: 'This is the inclusion functor $\FinAb \hookrightarrow \Ab$ that maps a finite abelian group to itself, regarded as an abelian group that has "forgotten" that it is finite. It provides an example of a fully faithful functor that is neither finitary nor cofinitary.'
nlab_link: https://ncatlab.org/nlab/show/forgetful+functor
left_adjoint: null

tags:
- algebra
- forgetful

related:
- forget_finite_group
- forget_torsion

satisfied_properties:
- property: fully faithful
proof: This is trivial.

- property: exact
proof: This follows from the construction of finite limits and colimits in $\FinAb$.

- property: preserves products
proof: The proof is identical to the proof that the <a href="/functor/forget_finite_group">inclusion functor $\FinGrp \hookrightarrow \Grp$</a> preserves products.

- property: preserves coproducts
proof: The proof is a minor variation of the proof that products are preserved, using the fact that finite coproducts and finite products coincide. Namely, one can show that a family of finite abelian groups has a coproduct in $\FinAb$ if and only if almost all groups are trivial.

unsatisfied_properties:
- property: dominant
proof: This is trivial.

- property: finitary
proof: We can take the same counterexample as for the <a href="/functor/forget_finite_group">inclusion functor $\FinGrp \hookrightarrow \Grp$</a>, which only involves abelian groups.

- property: cofinitary
proof: We can take the same counterexample as for the <a href="/functor/forget_finite_group">inclusion functor $\FinGrp \hookrightarrow \Grp$</a>, which only involves abelian groups.

- property: left-invertible
proof: We can use the same proof as for the <a href="/functor/forget_finite_group">inclusion functor $\FinGrp \hookrightarrow \Grp$</a>.
67 changes: 67 additions & 0 deletions database/data/functors/forget_finite_group.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,67 @@
id: forget_finite_group
name: forgetful functor from finite groups to groups
notation: $U_{\FinGrp, \Grp}$
domain: FinGrp
codomain: Grp
description: 'This is the inclusion functor $\FinGrp \hookrightarrow \Grp$. It can also be viewed as a forgetful functor that forgets the property of being finite. Among other things, it provides an example of a fully faithful functor that is neither finitary nor cofinitary.'
nlab_link: https://ncatlab.org/nlab/show/forgetful+functor
left_adjoint: null

tags:
- algebra
- forgetful

related:
- forget_group
- forget_finite
- forget_torsion
- forget_finite_abelian_group

satisfied_properties:
- property: fully faithful
proof: This is trivial.

- property: left exact
proof: This follows from the construction of finite limits in $\FinGrp$.
check_redundancy: false

- property: preserves coequalizers
proof: This follows from the construction of coequalizers in $\FinGrp$.

- property: preserves products
proof: >-
It is sufficient to prove that if a family of finite groups $(G_i)_{i \in I}$ has a product $P$ in $\FinGrp$, then almost all $G_i$ are trivial; then the product reduces to a finite product, which is clearly preserved. Note that it is not enough to prove that the $\Grp$-product is infinite; this argument would be circular, since it assumes that the $\FinGrp$-product is equal to the $\Grp$-product.
Assume that there are infinitely many indices $i_1,i_2,\dotsc$ for which $G_{i_n}$ is non-trivial. Let $N \geq 1$. There is a canonical homomorphism $P \to \prod_{n=1}^{N} G_{i_n}$ to the finite product (which definitely exists), and by using trivial homomorphisms, we see that it is actually a split epimorphism. Therefore,
$$\textstyle\card(P) \geq \prod_{i=1}^{N} \card(G_{i_n}) \geq \prod_{i=1}^{N} 2 = 2^N.$$
Since this holds for all $N$, we obtain a contradiction to the finiteness of $P$.

- property: preserves coproducts
proof: >-
It is sufficient to prove that if a family of finite groups $(G_i)_{i \in I}$ has a coproduct $C$ in $\FinGrp$, then at most one $G_j$ is non-trivial; then the coproduct is simply $G_j$, and it is clearly preserved. Note that it is not enough to prove that the $\Grp$-coproduct is infinite; this argument would be circular, since it assumes that the $\FinGrp$-coproduct is equal to the $\Grp$-coproduct.
Let $\coprod_{i \in I} G_i$ denote the coproduct of $(G_i)_{i \in I}$ in $\Grp$, which can be constructed as the free product. Assume that two groups in the family are non-trivial, say $G_0$ and $G_1$ for some indices $0,1 \in I$.
We claim that the canonical homomorphism $G_0 \sqcup G_1 \to C$ is a monomorphism. If $x$ is a non-trivial element in the free product $G_0 \sqcup G_1$, which is well-known to be residually finite, there is some finite group $H$ and a homomorphism $\varphi : G_0 \sqcup G_1 \to H$ with $\varphi(x) \neq 1$. The homomorphism $\varphi$ corresponds to two homomorphisms $\varphi_0 : G_0 \to H$ and $\varphi_1 : G_1 \to H$. For all $i \neq 0,1$, let $\varphi_i : G_i \to H$ be the trivial homomorphism. The universal property of $C$ yields a unique homomorphism $\psi : C \to H$ that extends each $\varphi_i$. It maps the image of $x$ in $C$ to $\varphi(x) \neq 1$, so that this image is non-trivial, as claimed.
However, $G_0 \sqcup G_1$ is an infinite group (which can be seen from the usual element description of a free product), so it cannot embed into the finite group $C$. This contradiction finishes the proof.

unsatisfied_properties:
- property: dominant
proof: This is trivial.

- property: left-invertible
proof: 'Assume that there is a functor $F : \Grp \to \FinGrp$ with natural isomorphisms $F(G) \cong G$ for finite groups $G$. Consider the group $A := F(\prod_{n \geq 1} \IZ/n)$. For every $n \geq 1$ there is a split monomorphism from $\IZ/n$ to $\prod_{n \geq 1} \IZ/n$. It induces a split monomorphism from $F(\IZ/n) \cong \IZ/n$ to $A$, so that $\card(A) \geq n$. But then $A$ cannot be finite.'

- property: finitary
proof: >-
Let $Q$ be any non-trivial abelian divisible torsion group, such as $Q = \IQ / \IZ$. Let $\I$ be the poset of finite subgroups of $Q$ and let $D : \I \to \FinGrp$ be the diagram $D(A)=A$. Of course, the colimit of $U \circ D : \I \to \Grp$ is $Q$, where $U : \FinGrp \hookrightarrow \Grp$ is the inclusion functor. However, it turns out that the colimit of $D$ also exists and is the trivial group $1$. To prove this, let $G$ be any finite group. Then we compute:
$$\Hom(D,\Delta(G)) \cong \Hom(U \circ D, \Delta(U(G))) \cong \Hom(Q,U(G))$$
But every homomorphism $Q \to U(G)$ is trivial because its image is a finite divisible abelian group. Thus, $\Hom(Q,U(G)) \cong \Hom(1,G)$, proving that $1$ is a colimit of $D$. This shows that the colimit of $D$ is not preserved by $U$.

For example, when $Q = \IQ / \IZ$, the diagram is isomorphic to $(\IN_{>0},\mid)^{\op} \to \FinGrp$, mapping $n$ to the cyclic group $\IZ/n$, and its (trivial) colimit is not preserved by $U$.

- property: cofinitary
proof: >-
Let $p$ be a prime, and consider the sequence of projection homomorphisms
$$\cdots \to \IZ/p^3 \to \IZ/p^2 \to \IZ/p.$$
Its limit in $\Grp$ is the additive group $\IZ_p$ of $p$-adic integers. However, its limit in $\FinGrp$ exists and is the trivial group. This is because a cone in $\FinGrp$ over that sequence with vertex $G \in \FinGrp$ can be identified with a homomorphism $G \to \IZ_p$ in $\Grp$, which must be trivial since its image is a finite torsion-free group.

- property: right exact
proof: See <a href="https://mathoverflow.net/questions/513646" target="_blank">MO/513646</a>.
1 change: 1 addition & 0 deletions database/data/functors/forget_group.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@ related:
- forget_ring
- forget_abelian
- forget_group_pointed
- forget_finite_group

satisfied_properties:
- property: representable
Expand Down
38 changes: 38 additions & 0 deletions database/data/functors/forget_hausdorff.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
id: forget_hausdorff
name: forgetful functor from Hausdorff spaces to topological spaces
notation: $U_{\Haus,\Top}$
domain: Haus
codomain: Top
description: This is the inclusion functor $\Haus \hookrightarrow \Top$ that maps a Hausdorff space to itself. It can also be viewed as a forgetful functor, since Hausdorff spaces "forget" that they are Hausdorff.
nlab_link: https://ncatlab.org/nlab/show/forgetful+functor
left_adjoint: null # TODO: add the Hausdorff reflection functor

tags:
- topology
- forgetful

related:
- forget_topology

satisfied_properties:
- property: fully faithful
proof: This is trivial.

- property: right adjoint
proof: The Hausdorff reflection (see <a href="https://ncatlab.org/nlab/show/Hausdorff+space#HausdorffReflectionViaHomsIntoAllHausdorffSpaces" target="_blank">nLab</a>) provides a left adjoint $\Top \to \Haus$.

- property: preserves coproducts
proof: This follows easily from the elementary fact that the $\Top$-coproduct of a family of Hausdorff spaces is again Hausdorff.

- property: preserves regular epimorphisms
proof: This follows from the classifications of regular epimorphisms in $\Haus$ and $\Top$; in both cases, they are surjective quotient maps.

unsatisfied_properties:
- property: preserves epimorphisms
proof: The inclusion $\IQ \hookrightarrow \IR$ is an epimorphism in $\Haus$ (since it has dense image), but not an epimorphism in $\Top$ (since it is not surjective).

- property: dominant
proof: A subspace of a Hausdorff space is again Hausdorff, and not every space is Hausdorff.

- property: finitary
proof: The inclusion functor certainly preserves monomorphisms, and in fact all limits. We know that monomorphisms in <a href="/category/Top">$\Haus$</a> are not stable under filtered colimits, whereas they are in <a href="/category/Top">$\Top$</a>. Thus, the claim follows from Lemma 2 <a href="/content/subcategories">here</a>.
5 changes: 1 addition & 4 deletions database/data/functors/forget_inverses.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ related:
- forget_torsion_free

satisfied_properties:
- property: full
- property: fully faithful
proof: It is a standard fact from group theory that a multiplicative map already preserves inverses.

- property: right adjoint
Expand All @@ -27,9 +27,6 @@ satisfied_properties:
- property: left adjoint
proof: 'This functor is left adjoint to the functor $(-)^{\times} : \Mon \to \Grp$ that sends a monoid $M$ to its group of units $M^{\times} \coloneqq \{(a,b) \in M^2 : ab = ba = 1 \}$. In fact, if $M$ is a monoid and $G$ is a group, there is a bijection $\Hom(U_{\Grp,\Mon}(G),M) \to \Hom(G,M^{\times})$ given by mapping $\varphi$ to $g \mapsto (\varphi(g),\varphi(g^{-1}))$.'

- property: left-invertible
proof: 'The group of units functor $(-)^{\times} : \Mon \to \Grp$ provides a left inverse.'

unsatisfied_properties:
- property: dominant
proof: Every submonoid of a group is cancellative. Thus, if $M$ is a non-cancellative monoid such as $(\IN,\cdot,1)$, there cannot be a group $G$ with a (split) monomorphism $M \hookrightarrow G$.
34 changes: 34 additions & 0 deletions database/data/functors/forget_torsion.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,34 @@
id: forget_torsion
name: forgetful functor from torsion abelian groups to abelian groups
notation: $U_{\TorsAb, \Ab}$
domain: TorsAb
codomain: Ab
description: 'This is the inclusion functor $\TorsAb \hookrightarrow \Ab$. It can also be viewed as a forgetful functor that forgets the property of being torsion. It is a typical example of a fully faithful functor that preserves finite products but does not preserve infinite products.'
nlab_link: https://ncatlab.org/nlab/show/forgetful+functor
left_adjoint: null # we only have the torsion functor Ab -> Ab in the database, not Ab -> TorsAb

tags:
- algebra
- forgetful

related:
- forget_abelian
- forget_finite_abelian_group
- forget_torsion_free

satisfied_properties:
- property: fully faithful
proof: This is trivial.

- property: left exact
proof: This follows from the construction of finite limits in $\TorsAb$.

- property: left adjoint
proof: 'This functor is left adjoint to the torsion subgroup functor $T : \Ab \to \TorsAb$, since for every torsion abelian group $A$ and every abelian group $B$, every homomorphism $A \to B$ restricts to a homomorphism $A \to T(B)$.'

unsatisfied_properties:
- property: dominant
proof: This is trivial.

- property: preserves products
proof: The product of the family $(\IZ/n)_{n \geq 1}$ in $\TorsAb$ is $T(\prod_{n \geq 1} \IZ/n)$, where $T(-)$ denotes the torsion subgroup and $\prod$ denotes the usual product in $\Ab$. This is not equal to $\prod_{n \geq 1} \IZ/n$ since, for example, $([1],[1],\dotsc)$ is not a torsion element.
3 changes: 2 additions & 1 deletion database/data/functors/forget_torsion_free.yaml
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
id: forget_torsion_free
name: forgetful functor from torsion-free abelian groups
name: forgetful functor from torsion-free abelian groups to abelian groups
notation: $U_{\TorsFreeAb, \Ab}$
domain: TorsFreeAb
codomain: Ab
Expand All @@ -14,6 +14,7 @@ tags:
related:
- forget_abelian
- forget_inverses
- forget_torsion

satisfied_properties:
- property: fully faithful
Expand Down
Loading