This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Virtualized Delta moves for virtual knots and links

Takuji NAKAMURA Faculty of Education, University of Yamanashi, Takeda 4-4-37, Kofu, Yamanashi, 400-8510, Japan [email protected] Yasutaka NAKANISHI Department of Mathematics, Kobe University, Rokkodai-cho 1-1, Nada-ku, Kobe 657-8501, Japan [email protected] Shin SATOH Department of Mathematics, Kobe University, Rokkodai-cho 1-1, Nada-ku, Kobe 657-8501, Japan [email protected]  and  Kodai Wada Department of Mathematics, Kobe University, Rokkodai-cho 1-1, Nada-ku, Kobe 657-8501, Japan [email protected]
Abstract.

We introduce a local deformation called the virtualized Δ\Delta-move for virtual knots and links. We prove that the virtualized Δ\Delta-move is an unknotting operation for virtual knots. Furthermore we give a necessary and sufficient condition for two virtual links to be related by a finite sequence of virtualized Δ\Delta-moves.

Key words and phrases:
virtual knot, virtual link, virtualized Δ\Delta-move, odd writhe
2020 Mathematics Subject Classification:
57K12, 57K10
This work was supported by JSPS KAKENHI Grant Numbers JP20K03621, JP19K03492, JP22K03287, and JP23K12973.

1. Introduction

Although the crossing change is elemental among local deformations in classical knot theory, the virtualization replacing a real crossing with a virtual crossing is considered as a more elemental deformation in virtual knot theory; in fact, a crossing change is realized by two virtualizations.

In this paper, we will introduce an elemental version of a Δ\Delta-move. Here, the Δ\Delta-move is one of the important local deformations in classical knot theory. In fact, it is an unknotting operation for classical knots, and characterizes classical links with the same pairwise linking numbers.

Definition 1.1.

A virtualized Δ\Delta-move is a local deformation on a virtual link diagram as shown in Figure 1.1. We denote it by vΔv\Delta in figures.

\begin{overpic}[width=227.62204pt]{virtualized-delta-eps-converted-to.pdf} \put(63.0,25.0){$v\Delta$} \put(152.0,25.0){$v\Delta$} \end{overpic}
Figure 1.1. A virtualized Δ\Delta-move

By definition, we see that a Δ\Delta-move is realized by a combination of two virtualized Δ\Delta-moves and a generalized Reidemeister move VI. In this sense, we can say that the virtualized Δ\Delta-move is more elemental than the Δ\Delta-move.

Definition 1.2.

Two virtual links LL and LL^{\prime} are vΔv\Delta-equivalent to each other if their diagrams are related by a finite sequence of virtualized Δ\Delta-moves and generalized Reidemeister moves.

For virtual knots, we have the following.

Theorem 1.3.

Any virtual knot is vΔv\Delta-equivalent to the trivial knot; that is, the virtualized Δ\Delta-move is an unknotting operation for virtual knots.

Let uvΔ(K){\rm u}_{v\Delta}(K) be the minimal number of virtualized Δ\Delta-moves which needs to deform KK into the trivial knot.

Theorem 1.4.

For any integer m1m\geq 1, there are infinitely many virtual knots KK with uvΔ(K)=m{\rm u}_{v\Delta}(K)=m.

For an nn-component virtual link LL with n2n\geq 2, we will define invariants pi(L){0,1}p_{i}(L)\in\{0,1\} (i=1,,ni=1,\dots,n). Then we have the following.

Theorem 1.5.

For n2n\geq 2, two nn-component virtual links LL and LL^{\prime} are vΔv\Delta-equivalent to each other if and only if pi(L)=pi(L)p_{i}(L)=p_{i}(L^{\prime}) holds for any i=1,,ni=1,\dots,n.

This paper is organized as follows. In Section 2, we study several properties of virtualized Δ\Delta-moves for virtual knots, and prove Theorems 1.3 and 1.4. In Section 3, we study the behavior of virtualized Δ\Delta-moves for virtual links, and prove Theorem 1.5. In Section 4, we divide virtualized Δ\Delta-moves into four types, and prove that any one of them generates the other three.

2. Virtualized Δ\Delta-moves for virtual knots

Lemma 2.1.

A crossing change at a real crossing is realized by a combination of a virtualized Δ\Delta-move and generalized Reidemeister moves.

Proof.

This follows from Figure 2.1, where the symbol R\overset{\textrm{R}}{\longleftrightarrow} means a combination of generalized Reidemeister moves. ∎

\begin{overpic}[width=341.43306pt]{pf-lem-cc-R-eps-converted-to.pdf} \put(42.5,21.0){R} \put(164.0,21.0){$v\Delta$} \put(291.5,21.0){R} \end{overpic}
Figure 2.1. Proof of Lemma 2.1
Lemma 2.2.

A local deformation FD as shown in Figure 2.2 is realized by a combination of two virtualized Δ\Delta-moves and generalized Reidemeister moves.

\begin{overpic}[width=128.0374pt]{forbidden-detour-eps-converted-to.pdf} \put(57.0,26.5){FD} \end{overpic}
Figure 2.2. A local deformation FD
Proof.

This follows from Figure 2.3. ∎

\begin{overpic}[width=341.43306pt]{pf-lem-fd-R-eps-converted-to.pdf} \put(60.7,132.0){R} \put(142.7,132.0){$v\Delta$} \put(230.7,132.0){R} \put(18.0,47.0){R} \put(103.0,47.0){R} \put(185.0,47.0){$v\Delta$} \put(273.0,47.0){R} \end{overpic}
Figure 2.3. Proof of Lemma 2.2
Remark 2.3.

The local deformation FD in Figure 2.3 is called a forbidden detour move [4, 11] or a fused move [1].

Lemma 2.4.

A forbidden move is realized by a combination of virtualized Δ\Delta-moves and generalized Reidemeister moves.

Proof.

The sequence of Figure 2.4 shows that an upper forbidden move is realized by a combination of two crossing changes and a forbidden detour move, where the symbol cc\overset{\textrm{cc}}{\longleftrightarrow} means a combination of crossing changes. By Lemmas 2.1 and 2.2, we have the result for the case of an upper forbidden move. The case of a lower forbidden move is proved similarly. ∎

\begin{overpic}[width=284.52756pt]{pf-lem-forbidden-R-eps-converted-to.pdf} \put(58.5,26.0){cc} \put(133.3,26.0){FD} \put(217.0,26.0){cc} \end{overpic}
Figure 2.4. An upper forbidden move
Proof of Theorem 1.3.

Since the forbidden move is an unknotting operation for virtual knots [6, 8], we have the conclusion by Lemma 2.4. ∎

Remark 2.5.

It is proved in [11] that the forbidden detour move is also an unknotting operation for virtual knots. By using this result, Theorem 1.3 is a direct consequence of Lemma 2.2.

By Theorem 1.3, any two virtual knots KK and KK^{\prime} are vΔv\Delta-equivalent to each other. We denote by dvΔ(K,K){\rm d}_{v\Delta}(K,K^{\prime}) the minimal number of virtualized Δ\Delta-moves needed to deform a diagram of KK into that of KK^{\prime}. It is called the vΔv\Delta-distance between KK and KK^{\prime}. In particular, we denote dvΔ(K,O){\rm d}_{v\Delta}(K,O) by uvΔ(K){\rm u}_{v\Delta}(K), and call it the vΔv\Delta-unknotting number of KK, where OO is the trivial knot.

We briefly recall the definitions of the nn-writhe Jn(K)J_{n}(K) and the odd writhe J(K)J(K) of a virtual knot KK from [9]. Let GG be a Gauss diagram of KK and γ\gamma a chord of GG. If γ\gamma admits a sign ε\varepsilon, we assign ε\varepsilon and ε-\varepsilon to the terminal and initial endpoints of γ\gamma, respectively. The endpoints of γ\gamma divide the underlying oriented circle of GG into two arcs. Let α\alpha be the one of the two oriented arcs which runs from the initial endpoint of γ\gamma to the terminal. The index of γ\gamma is the sum of the signs of all the endpoints of chords on α\alpha, and denoted by Ind(γ){\rm Ind}(\gamma). For a nonzero integer nn, the nn-writhe Jn(K)J_{n}(K) of KK is the sum of the signs of all the chords γ\gamma with Ind(γ)=n{\rm Ind}(\gamma)=n, and the odd writhe J(K)J(K) of KK is defined to be n:oddJn(K)\sum_{n{\rm:odd}}J_{n}(K). We remark that J(K)J(K) is always even [3].

We give a lower bound for dvΔ(K,K){\rm d}_{v\Delta}(K,K^{\prime}) by using the odd writhes of KK and KK^{\prime} as follows.

Proposition 2.6.

Let KK and KK^{\prime} be virtual knots. Then we have the following.

  • (i)

    dvΔ(K,K)12|J(K)J(K)|{\rm d}_{v\Delta}(K,K^{\prime})\geq\frac{1}{2}|J(K)-J(K^{\prime})|.

  • (ii)

    uvΔ(K)12|J(K)|{\rm u}_{v\Delta}(K)\geq\frac{1}{2}|J(K)|.

Proof.

Let GG and GG^{\prime} be Gauss diagrams of KK and KK^{\prime}, respectively. It is sufficient to prove that if GG and GG^{\prime} are related by a virtualized Δ\Delta-move, then the odd writhes satisfy J(K)J(K){0,±2}J(K)-J(K^{\prime})\in\{0,\pm 2\}.

We may assume that GG^{\prime} is obtained from GG by removing three chords corresponding to three real crossings involved in a virtualized Δ\Delta-move. See Figure 2.5. It is known that among these three chords, the number of chords with index odd is equal to zero or two (cf. [7, 10]). Furthermore, the parity of the index of any other chord is preserved by the virtualized Δ\Delta-move; in fact, each pair of the three chords has two adjacent endpoints on the underlying circle. See the figure again. Therefore we have the conclusion. ∎

\begin{overpic}[width=170.71652pt]{pf-prop-distance-eps-converted-to.pdf} \put(79.0,26.0){$v\Delta$} \put(30.0,-15.0){$G$} \put(133.0,-15.0){$G^{\prime}$} \end{overpic}
Figure 2.5. A virtualized Δ\Delta-move on a Gauss diagram

It is known that the crossing change at a real crossing is not an unknotting operation for virtual knots (cf. [2, 5, 10]). If a virtual knot KK can be deformed into the trivial knot by a finite number of crossing changes, then we denote by u(K){\rm u}(K) the minimal number of such crossing changes. If KK cannot be unknotted by crossing changes, then we set u(K)={\rm u}(K)=\infty. Then we have the following by Lemma 2.1 immediately.

Lemma 2.7.

Any virtual knot KK satisfies uvΔ(K)u(K){\rm u}_{v\Delta}(K)\leq{\rm u}(K). \Box

In the following, we will construct two families of infinitely many virtual knots KK with uvΔ(K)=m{\rm u}_{v\Delta}(K)=m for any given integer m1m\geq 1, which have different properties for u(K){\rm u}(K). The following theorems induce Theorem 1.4 immediately.

Theorem 2.8.

For any integer m1m\geq 1, there are infinitely many virtual knots KK with uvΔ(K)=u(K)=m{\rm u}_{v\Delta}(K)={\rm u}(K)=m.

Proof.

For an integer s1s\geq 1, we consider a knot diagram and its Gauss diagram with 2m+2s12m+2s-1 real crossings (or chords) as shown in Figure 2.6. Let Ks(m)K_{s}(m) be the virtual knot presented by this diagram.

\begin{overpic}[width=312.9803pt]{pf-thm-infinite-eps-converted-to.pdf} \put(84.0,123.0){\rotatebox{-90.0}{$2m$ real crossings}} \put(7.7,-13.0){$2s-1$ real crossings} \put(319.0,96.0){\rotatebox{-90.0}{$2m$ chords}} \put(212.0,-13.0){$2s-1$ chords} \end{overpic}
Figure 2.6. A diagram of Ks(m)K_{s}(m) and its Gauss diagram

Since Ks(m)K_{s}(m) can be unknotted by mm crossing changes at mm real crossings among 2m2m half twists in the knot diagram, we have u(Ks(m))m{\rm u}(K_{s}(m))\leq m. By Lemma 2.7, we have

uvΔ(Ks(m))u(Ks(m))m.{\rm u}_{v\Delta}(K_{s}(m))\leq{\rm u}(K_{s}(m))\leq m.

On the other hand, the 2m2m horizontal chords of the Gauss diagram have the sign 1-1 and indices ±1\pm 1, and the 2s12s-1 vertical chords have index 0. Since J(Ks(m))=2mJ(K_{s}(m))=-2m holds, we have uvΔ(Ks(m))m{\rm u}_{v\Delta}(K_{s}(m))\geq m by Proposition 2.6(ii), and hence uvΔ(Ks(m))=u(Ks(m))=m{\rm u}_{v\Delta}(K_{s}(m))={\rm u}(K_{s}(m))=m.

Now it is enough to prove that the virtual knots Ks(m)K_{s}(m)’s are mutually distinct. By a straightforward calculation, the Jones polynomial fKs(m)(A)[A,A1]f_{K_{s}(m)}(A)\in{\mathbb{Z}}[A,A^{-1}] of Ks(m)K_{s}(m) is given by

fKs(m)(A)=A8m+(A4A8)(i=1mA8i)(A8s2+j=12s1(1)jA4j).f_{K_{s}(m)}(A)=A^{8m}+(A^{-4}-A^{-8})\left(\sum_{i=1}^{m}A^{8i}\right)\left(-A^{8s-2}+\sum_{j=1}^{2s-1}(-1)^{j}A^{4j}\right).

Since the maximal degree of fKs(m)(A)f_{K_{s}(m)}(A) is equal to 8m+8s68m+8s-6, we have Ks(m)Ks(m)K_{s}(m)\neq K_{s^{\prime}}(m) for any sss\neq s^{\prime}. ∎

Theorem 2.9.

For any integer m1m\geq 1, there are infinitely many virtual knots KK with uvΔ(K)=m{\rm u}_{v\Delta}(K)=m and u(K)={\rm u}(K)=\infty.

Proof.

For an integer s2s\geq 2, we consider a long virtual knot diagram and its Gauss diagram with 2s+32s+3 real crossings (or chords) a1,a2,a3a_{1},a_{2},a_{3} and b1,,b2sb_{1},\dots,b_{2s} as shown in Figure 2.7. Let TsT_{s} be the long virtual knot presented by this diagram, and Ks(m)K_{s}(m) the virtual knot obtained from the closure of the product of mm copies of TsT_{s}.

Since TsT_{s} can be unknotted by a single virtualized Δ\Delta-move involving the three crossings a1a_{1}, a2a_{2}, and a3a_{3} in the long knot diagram, we have uvΔ(Ks(m))m{\rm u}_{v\Delta}(K_{s}(m))\leq m.

On the other hand, since we have

Ind(a1)=1,Ind(a2)=2s,Ind(a3)=2s1, and Ind(bi)=2(1i2s),{\rm Ind}(a_{1})=1,\ {\rm Ind}(a_{2})=2s,\ {\rm Ind}(a_{3})=-2s-1,\mbox{ and }{\rm Ind}(b_{i})=2\ (1\leq i\leq 2s),

it follows from [9, Lemma 4.3] that

Jn(Ks(m))={mif n=2s,2msif n=2,mif n=1,2s1,0otherwise.J_{n}(K_{s}(m))=\begin{cases}-m&\text{if }n=2s,\\ 2ms&\text{if }n=2,\\ m&\text{if }n=1,-2s-1,\\ 0&\text{otherwise}.\end{cases}

This induces J(Ks(m))=J1(Ks(m))+J2s1(Ks(m))=2mJ(K_{s}(m))=J_{1}(K_{s}(m))+J_{-2s-1}(K_{s}(m))=2m. Therefore we have uvΔ(Ks(m))m{\rm u}_{v\Delta}(K_{s}(m))\geq m by Proposition 2.6(ii), and hence uvΔ(Ks(m))=m{\rm u}_{v\Delta}(K_{s}(m))=m. Furthermore, since J1(Ks(m))=m0=J1(Ks(m))J_{1}(K_{s}(m))=m\neq 0=J_{-1}(K_{s}(m)) holds, we have u(Ks(m))={\rm u}(K_{s}(m))=\infty by [9, Theorem 1.5].

Since J2(Ks(m))=2msJ_{2}(K_{s}(m))=2ms holds, we have Ks(m)Ks(m)K_{s}(m)\neq K_{s^{\prime}}(m) for any sss\neq s^{\prime}. ∎

\begin{overpic}[width=341.43306pt]{fig-infinite-example3-eps-converted-to.pdf} \put(33.5,-4.0){$a_{1}$} \put(50.0,-4.0){$a_{3}$} \put(42.0,30.0){$a_{2}$} \put(65.0,135.0){$b_{1}$} \put(65.0,81.0){$b_{2s-1}$} \put(65.0,53.5){$b_{2s}$} \put(218.0,77.0){$a_{1}$} \put(218.0,63.5){$a_{2}$} \put(249.0,63.5){$a_{3}$} \put(218.0,49.5){$b_{1}$} \put(218.0,36.0){$b_{2}$} \put(218.0,22.0){$b_{2s}$} \end{overpic}
Figure 2.7. A diagram of TsT_{s} and its Gauss diagram

3. Virtualized Δ\Delta-moves for virtual links

Lemma 3.1.

A local deformation as shown in Figure 3.1 is realized by a combination of a virtualized Δ\Delta-move and generalized Reidemeister moves.

\begin{overpic}[width=170.71652pt]{orientation-reversal-eps-converted-to.pdf} \end{overpic}
Figure 3.1. A local deformation considered in Lemma 3.1
Proof.

This follows from Figure 3.2. ∎

\begin{overpic}[width=341.43306pt]{pf-lem-or-R-eps-converted-to.pdf} \put(89.0,138.0){R} \put(227.0,138.0){R} \put(6.3,83.0){R} \put(145.5,84.0){$v\Delta$} \put(6.3,23.0){R} \put(148.8,23.0){R} \end{overpic}
Figure 3.2. Proof of Lemma 3.1
Lemma 3.2.

Let γ\gamma be a chord of a Gauss diagram GG.

  • (i)

    If a Gauss diagram GG^{\prime} is obtained from GG by reversing the orientation of γ\gamma, then GG and GG^{\prime} are related by a finite sequence of virtualized Δ\Delta-moves and Reidemeister moves.

  • (ii)

    If a Gauss diagram G′′G^{\prime\prime} is obtained from GG by changing the sign of γ\gamma, then GG and G′′G^{\prime\prime} are related by a finite sequence of virtualized Δ\Delta-moves and Reidemeister moves.

Proof.

(i) This follows from Lemma 3.1.

(ii) Let G′′′G^{\prime\prime\prime} be a Gauss diagram obtained from GG by reversing the orientation and changing the sign of γ\gamma. By Lemma 2.1, GG and G′′′G^{\prime\prime\prime} are related by a finite sequence of a virtualized Δ\Delta-move and Reidemeister moves. Furthermore G′′G^{\prime\prime} is obtained from G′′′G^{\prime\prime\prime} by reversing the orientation of γ\gamma. Therefore we have the conclusion by (i). ∎

Lemma 3.3.

Let γ\gamma and γ\gamma^{\prime} be chords of a Gauss diagram GG such that an endpoint of γ\gamma is adjacent to that of γ\gamma^{\prime}. If a Gauss diagram GG^{\prime} is obtained from GG by switching the positions of these consecutive endpoints, then GG and GG^{\prime} are related by a finite sequence of virtualized Δ\Delta-moves and Reidemeister moves.

Proof.

This follows from Lemmas 2.2 and 2.4. ∎

For n1n-1 integers a2,,an{0,1}a_{2},\dots,a_{n}\in\{0,1\} with n2n\geq 2, let H(a2,,an)=i=1nHiH(a_{2},\dots,a_{n})=\bigcup_{i=1}^{n}H_{i} be the Gauss diagram of an nn-component virtual link such that

  • (i)

    H(a2,,an)H(a_{2},\dots,a_{n}) has no self-chords,

  • (ii)

    there are no nonself-chords between HiH_{i} and HjH_{j} (2i<jn)(2\leq i<j\leq n),

  • (iii)

    if ai=0a_{i}=0, then there are no nonself-chords between H1H_{1} and HiH_{i},

  • (iv)

    if ai=1a_{i}=1, then there is a single nonself-chord between H1H_{1} and HiH_{i} which is oriented from H1H_{1} to HiH_{i} with a positive sign, and

  • (v)

    if ai=aj=1a_{i}=a_{j}=1 (2i<jn)(2\leq i<j\leq n), then we meet the endpoint of the chord between H1H_{1} and HiH_{i} before that between H1H_{1} and HjH_{j} along H1H_{1}.

Figure 3.3 shows the Gauss diagram H(1,0,1,1,0)H(1,0,1,1,0) with n=6n=6. Let M(a2,,an)M(a_{2},\dots,a_{n}) be the nn-component virtual link presented by H(a2,,an)H(a_{2},\dots,a_{n}).

\begin{overpic}[width=199.16928pt]{ex-H-eps-converted-to.pdf} \put(-15.0,67.0){$H_{1}$} \put(10.0,-13.0){$H_{2}$} \put(52.0,-13.0){$H_{3}$} \put(94.0,-13.0){$H_{4}$} \put(136.0,-13.0){$H_{5}$} \put(178.0,-13.0){$H_{6}$} \end{overpic}
Figure 3.3. The Gauss diagram H(1,0,1,1,0)H(1,0,1,1,0)
Proposition 3.4.

Let n2n\geq 2 be an integer. Any nn-component virtual link L=i=1nKiL=\bigcup_{i=1}^{n}K_{i} is vΔv\Delta-equivalent to M(a2,,an)M(a_{2},\dots,a_{n}) for some aia_{i}’s.

Proof.

Let G=i=1nGiG=\bigcup_{i=1}^{n}G_{i} be a Gauss diagram of LL. By using Lemma 3.3 and Reidemeister moves I, we can remove all the self-chords from GG. Therefore we may assume that GG has no self-chords (up to virtualized Δ\Delta-moves and Reidemeister moves).

If there is a nonself-chord between GiG_{i} and GjG_{j} (2i<jn)(2\leq i<j\leq n), then we can replace it with a pair of nonself-chords one of which connects between G1G_{1} and GiG_{i}, and the other between G1G_{1} and GjG_{j}. In fact, this is achieved by observing a sequence of link diagrams as shown in Figure 3.4. Therefore we may also assume that there are no nonself-chords between GiG_{i} and GjG_{j} (2i<jn)(2\leq i<j\leq n).

\begin{overpic}[width=227.62204pt]{pf-prop-form-R-eps-converted-to.pdf} \put(-12.0,53.0){$K_{j}$} \put(33.0,53.0){$K_{i}$} \put(13.0,-13.0){$K_{1}$} \put(54.0,35.5){R} \put(84.0,53.0){$K_{j}$} \put(112.0,53.0){$K_{i}$} \put(100.0,-13.0){$K_{1}$} \put(147.0,35.5){$v\Delta$} \put(179.0,53.0){$K_{j}$} \put(208.0,53.0){$K_{i}$} \put(196.0,-13.0){$K_{1}$} \end{overpic}
Figure 3.4. A sequence of link diagrams

By using Lemmas 3.2, 3.3, and Reidemeister moves II, we can reduce the number of nonself-chords between G1G_{1} and GiG_{i} (2in)(2\leq i\leq n) to zero or one.

Finally, any Gauss diagram GG of LL can be deformed into H(a2,,an)H(a_{2},\dots,a_{n}) for some aia_{i}’s by Lemmas 3.2 and 3.3. ∎

For n2n\geq 2, let L=K1KnL=K_{1}\cup\dots\cup K_{n} be an nn-component virtual link, and G=G1GnG=G_{1}\cup\dots\cup G_{n} a Gauss diagram of LL. For each i=1,,ni=1,\dots,n, the iith parity of LL is defined to be the parity of the number of endpoints of nonself-chords on GiG_{i}, and denoted by pi(L){0,1}p_{i}(L)\in\{0,1\}. Since a self-chord has two endpoints on the same component, pi(L)p_{i}(L) is coincident with the parity of the number of endpoints of self-/nonself-chords on GiG_{i}.

Lemma 3.5.

For any i=1,,ni=1,\dots,n, the iith parity pi(L)p_{i}(L) is an invariant of the vΔv\Delta-equivalence class of LL.

Proof.

Since the chord(s) involved in a Reidemeister move or a virtualized Δ\Delta-move have an even number of endpoints on each GiG_{i}, this move preserves pi(L)p_{i}(L). ∎

Remark 3.6.

For n2n\geq 2, any nn-component virtual link LL satisfies

p1(L)+p2(L)++pn(L)0(mod2).p_{1}(L)+p_{2}(L)+\dots+p_{n}(L)\equiv 0\ ({\rm mod}~{}2).

In fact, the number of endpoints of all self-/nonself-chords is even.

Proof of Theorem 1.5.

The only if part follows from Lemma 3.5.

We will prove the if part. By Proposition 3.4, LL and LL^{\prime} are vΔv\Delta-equivalent to M(a2,,an)M(a_{2},\dots,a_{n}) and M(b2,,bn)M(b_{2},\dots,b_{n}) for some aia_{i}’s and bib_{i}’s, respectively. By assumption and Lemma 3.5, it holds that

ai=pi(M(a2,,an))=pi(L)=pi(L)=pi(M(b2,,bn))=bia_{i}=p_{i}\bigl{(}M(a_{2},\dots,a_{n})\bigr{)}=p_{i}(L)=p_{i}(L^{\prime})=p_{i}\bigl{(}M(b_{2},\dots,b_{n})\bigr{)}=b_{i}

for any i=1,,ni=1,\dots,n. Therefore LL and LL^{\prime} are vΔv\Delta-equivalent to M(a2,,an)=M(b2,,bn)M(a_{2},\dots,a_{n})=M(b_{2},\dots,b_{n}). ∎

The following is an immediate consequence of the proof of Theorem 1.5.

Corollary 3.7.

For n2n\geq 2, a complete representative system of the vΔv\Delta-equivalence classes of nn-component virtual links is given by

{M(a2,,an)a2,,an{0,1}}.\left\{M(a_{2},\dots,a_{n})\mid a_{2},\dots,a_{n}\in\{0,1\}\right\}.

In particular, the number of vΔv\Delta-equivalence classes is equal to 2n12^{n-1}. \Box

4. Relations among four types of virtualized Δ\Delta-moves

We can divide virtualized Δ\Delta-moves into four types rvΔrv\Delta, vrΔvr\Delta, rvΔrv\Delta^{\prime}, and vrΔvr\Delta^{\prime} as shown in Figure 4.1. Here, the rvΔrv\Delta- and rvΔrv\Delta^{\prime}-moves replace three real crossings with virtual ones, and the vrΔvr\Delta- and vrΔvr\Delta^{\prime}-moves replace three virtual crossings with real ones.

\begin{overpic}[width=227.62204pt]{4types-eps-converted-to.pdf} \put(60.5,29.0){$rv\Delta$} \put(60.5,5.5){$vr\Delta$} \put(149.0,29.0){$rv\Delta^{\prime}$} \put(149.0,5.5){$vr\Delta^{\prime}$} \end{overpic}
Figure 4.1. Four types rvΔrv\Delta, vrΔvr\Delta, rvΔrv\Delta^{\prime}, and vrΔvr\Delta^{\prime}
Lemma 4.1.

For any X{rvΔ,vrΔ,rvΔ,vrΔ}X\in\{rv\Delta,vr\Delta,rv\Delta^{\prime},vr\Delta^{\prime}\}, a crossing change at a real crossing is realized by a combination of an XX-move and generalized Reidemeister moves.

Proof.

For X=rvΔX=rv\Delta, a crossing change at a real crossing is obtained from the deformation sequence in Figure 2.1 by replacing vΔ\stackrel{{\scriptstyle v\Delta}}{{\longleftrightarrow}} with rvΔ\stackrel{{\scriptstyle rv\Delta}}{{\longrightarrow}}. For X=vrΔX=vr\Delta, we may follow this sequence in reverse. For X=rvΔX=rv\Delta^{\prime} and vrΔvr\Delta^{\prime}, we may perform the crossing change at every real crossing in the above sequences for rvΔrv\Delta and vrΔvr\Delta, respectively. ∎

Lemma 4.2.

A local deformation SR as shown in Figure 4.2 is realized by a combination of rvΔrv\Delta-moves and generalized Reidemeister moves.

\begin{overpic}[width=170.71652pt]{sign-reversal-eps-converted-to.pdf} \put(78.5,16.0){SR} \end{overpic}
Figure 4.2. A local deformation SR
Proof.

Figure 4.3 indicates the proof. More precisely, the first deformation in this figure is obtained from the deformation sequence in Figures 3.2 by replacing vΔ\stackrel{{\scriptstyle v\Delta}}{{\longleftrightarrow}} with rvΔ\stackrel{{\scriptstyle rv\Delta}}{{\longrightarrow}}. The second deformation is a crossing change which is realized by an rvΔrv\Delta-move and generalized Reidemeister moves by Lemma 4.1. ∎

\begin{overpic}[width=284.52756pt]{pf-lem-sr-eps-converted-to.pdf} \put(190.0,16.0){cc} \end{overpic}
Figure 4.3. Proof of Lemma 4.2
Remark 4.3.

The local deformation SR in Figure 4.2 is called a sign reversal move [1].

Theorem 4.4.

For any XY{rvΔ,vrΔ,rvΔ,vrΔ}X\neq Y\in\{rv\Delta,vr\Delta,rv\Delta^{\prime},vr\Delta^{\prime}\}, a YY-move is realized by a combination of XX-moves and generalized Reidemeister moves.

Proof.

We use the notation “XYX\Rightarrow Y” if a YY-move is realized by a combination of XX-moves and generalized Reidemeister moves. Then we have the following.

  • rvΔvrΔrv\Delta\Rightarrow vr\Delta: The sequence in Figure 4.4 shows that a vrΔvr\Delta-move is realized by a combination of an rvΔrv\Delta-move, three sign reversal moves, and several generalized Reidemeister moves. Therefore we have rvΔvrΔrv\Delta\Rightarrow vr\Delta by Lemma 4.2.

  • vrΔvrΔvr\Delta\Rightarrow vr\Delta^{\prime}: The sequence in Figure 4.5 shows that a vrΔvr\Delta^{\prime}-move is realized by a vrΔvr\Delta-move and three crossing changes. Therefore we have vrΔvrΔvr\Delta\Rightarrow vr\Delta^{\prime} by Lemma 4.1.

  • vrΔrvΔvr\Delta^{\prime}\Rightarrow rv\Delta^{\prime}: We may follow any sequence for rvΔvrΔrv\Delta\Rightarrow vr\Delta with opposite crossing information in reverse.

  • rvΔrvΔrv\Delta^{\prime}\Rightarrow rv\Delta: We may follow any sequence for vrΔvrΔvr\Delta\Rightarrow vr\Delta^{\prime} with opposite crossing information in reverse.

Therefore we have the cycle

rvΔvrΔvrΔrvΔrvΔ,rv\Delta\Rightarrow vr\Delta\Rightarrow vr\Delta^{\prime}\Rightarrow rv\Delta^{\prime}\Rightarrow rv\Delta,

which induces the other eight cases immediately. ∎

\begin{overpic}[width=341.43306pt]{pf-thm-relation-R-eps-converted-to.pdf} \put(67.0,152.0){R} \put(204.0,152.0){$rv\Delta$} \put(7.0,48.0){SR} \put(157.0,48.0){R} \end{overpic}
Figure 4.4. A vrΔvr\Delta-move is realized by rvΔrv\Delta-moves
\begin{overpic}[width=256.0748pt]{pf-thm-relation2-R-eps-converted-to.pdf} \put(67.5,26.0){$vr\Delta$} \put(173.5,26.0){cc} \end{overpic}
Figure 4.5. A vrΔvr\Delta^{\prime}-move is realized by vrΔvr\Delta-moves

References

  • [1] B. Audoux, P. Bellingeri, J.-B. Meilhan, and E. Wagner, Extensions of some classical local moves on knot diagrams, Michigan Math. J. 67 (2018), no. 3, 647–672.
  • [2] J. S. Carter, S. Kamada, and M. Saito, Stable equivalence of knots on surfaces and virtual knot cobordisms, J. Knot Theory Ramifications 11 (2002), no. 3, 311–322.
  • [3] Z. Cheng, A polynomial invariant of virtual knots, Proc. Amer. Math. Soc. 142 (2014), no. 2, 713–725.
  • [4] A. S. Crans, B. Mellor, and S. Ganzell, The forbidden number of a knot, Kyungpook Math. J. 55 (2015), no. 2, 485–506.
  • [5] D. Hrencecin and L. H. Kauffman, On filamentations and virtual knots, Topology Appl. 134 (2003), 23–52.
  • [6] T. Kanenobu, Forbidden moves unknot a virtual knot, J. Knot Theory Ramifications 10 (2001), no. 1, 89–96.
  • [7] V. O. Manturov, Parity and projection from virtual knots to classical knots, J. Knot Theory Ramifications 22 (2013), no. 9, 1350044, 20 pp.
  • [8] S. Nelson, Unknotting virtual knots with Gauss diagram forbidden moves, J. Knot Theory Ramifications 10 (2001), no. 6, 931–935.
  • [9] S. Satoh and K. Taniguchi, The writhes of a virtual knot, Fund. Math. 225 (2014), no. 1, 327–342.
  • [10] V.Turaev, Virtual strings. Ann. Inst. Fourier (Grenoble) 54 (2004), no. 7, 2455–2525.
  • [11] S. Yoshiike and K. Ichihara, Forbidden detour number on virtual knot, Kyungpook Math. J. 61 (2021), no. 1, 205–212.