
In this study, we develop a nonmonotone variable metric Barzilai-Borwein method for minimizing the sum of a smooth function and a convex, possibly nondifferentiable, function. At each step, the descent direction is obtained by taking the difference between the minimizer of the scaling proximal function and the current iteration point. An adaptive nonmonotone line search is proposed for determining the step length along this direction. We also show that the limit point of the iterates sequence is a stationary point. Numerical results with parallel magnetic resonance imaging, Poisson, and Cauchy noise deblurring demonstrate the effectiveness of the new algorithm.
Citation: Xiao Guo, Chuanpei Xu, Zhibin Zhu, Benxin Zhang. Nonmonotone variable metric Barzilai-Borwein method for composite minimization problem[J]. AIMS Mathematics, 2024, 9(6): 16335-16353. doi: 10.3934/math.2024791
[1] | Tahair Rasham, Muhammad Nazam, Hassen Aydi, Abdullah Shoaib, Choonkil Park, Jung Rye Lee . Hybrid pair of multivalued mappings in modular-like metric spaces and applications. AIMS Mathematics, 2022, 7(6): 10582-10595. doi: 10.3934/math.2022590 |
[2] | Ahlam Almulhim . Signed double Italian domination. AIMS Mathematics, 2023, 8(12): 30895-30909. doi: 10.3934/math.20231580 |
[3] | Abel Cabrera Martínez, Iztok Peterin, Ismael G. Yero . Roman domination in direct product graphs and rooted product graphs. AIMS Mathematics, 2021, 6(10): 11084-11096. doi: 10.3934/math.2021643 |
[4] | Ana Klobučar Barišić, Antoaneta Klobučar . Double total domination number in certain chemical graphs. AIMS Mathematics, 2022, 7(11): 19629-19640. doi: 10.3934/math.20221076 |
[5] | Linyu Li, Jun Yue, Xia Zhang . Double total domination number of Cartesian product of paths. AIMS Mathematics, 2023, 8(4): 9506-9519. doi: 10.3934/math.2023479 |
[6] | Bana Al Subaiei, Ahlam AlMulhim, Abolape Deborah Akwu . Vertex-edge perfect Roman domination number. AIMS Mathematics, 2023, 8(9): 21472-21483. doi: 10.3934/math.20231094 |
[7] | S. Gajavalli, A. Berin Greeni . On strong geodeticity in the lexicographic product of graphs. AIMS Mathematics, 2024, 9(8): 20367-20389. doi: 10.3934/math.2024991 |
[8] | Mingyu Zhang, Junxia Zhang . On Roman balanced domination of graphs. AIMS Mathematics, 2024, 9(12): 36001-36011. doi: 10.3934/math.20241707 |
[9] | Chang Liu, Jianping Li . Sharp bounds on the zeroth-order general Randić index of trees in terms of domination number. AIMS Mathematics, 2022, 7(2): 2529-2542. doi: 10.3934/math.2022142 |
[10] | Abdullah Shoaib, Tahair Rasham, Giuseppe Marino, Jung Rye Lee, Choonkil Park . Fixed point results for dominated mappings in rectangular b-metric spaces with applications. AIMS Mathematics, 2020, 5(5): 5221-5229. doi: 10.3934/math.2020335 |
In this study, we develop a nonmonotone variable metric Barzilai-Borwein method for minimizing the sum of a smooth function and a convex, possibly nondifferentiable, function. At each step, the descent direction is obtained by taking the difference between the minimizer of the scaling proximal function and the current iteration point. An adaptive nonmonotone line search is proposed for determining the step length along this direction. We also show that the limit point of the iterates sequence is a stationary point. Numerical results with parallel magnetic resonance imaging, Poisson, and Cauchy noise deblurring demonstrate the effectiveness of the new algorithm.
Throughout this paper, all graphs considered are finite, undirected, loopless and without multiple edges. We refer the reader to [3,18] for terminology and notation in graph theory.
Let G=(V,E) be a graph of order n with vertex set V(G) and edge set E(G). The open neighborhood of a vertex v in G is NG(v)=N(v)={u∈V(G)|uv∈E(G)}, and the closed neighborhood of v is NG[v]=N[v]=N(v)∪{v}. The degree of a vertex v in the graph G is dG(v)=d(v)=|N(v)|. Let δ(G)=δ and Δ(G)=Δ denote the minimum and maximum degree of a graph G, respectively. Denote by G[H] the induced subgraph of G induced by H with H⊂V(G). A vertex v in G is a leaf if d(v)=1. A vertex u is a support vertex if u has a leaf neighbor. Denote L(u)={v|uv∈E(G),d(v)=1}.
A subset M⊆E(G) is called a matching in G if no two elements are adjacent in G. A vertex v is said to be M-saturated if some edges of M are incident with v, otherwise, v is M-unsaturated. If every vertex of G is M-saturated, the matching M is perfect. M is a maximum matching if G has no matching M′ with |M′|>|M|. Let R be a subgraph of G, M be a matching of G, v∈V(R), uv∈M. We say the vertex v is RI (resp. RO) if u∈V(R) (resp. u∉V(R)).
A set VC of vertices in a graph G is a vertex cover of G if all the edges are touched by the vertices in VC. A vertex cover VC of G is minimal if no proper subset of it is a vertex cover of G. A minimal vertex cover of maximum cardinality is called a VC-set. In 2001, Mishra et al. [13] denote by MAX-MIN-VC the problem of finding a VC-set of G. Bazgan et al. [2] showed that MAX-MIN-VC is APX-complete for cubic graphs.
A set PD⊆V(G) in a graph G is a paired dominating set if every vertex v∉PD is adjacent to a vertex in PD and the subgraph induced by PD contains a perfect matching. Paired domination was proposed in 1996 [9] and was studied for example in [4,5,6,12,16,17]. A paired dominating set PD of G is minimal if there is no proper subset PD′⊂PD which is a paired dominating set of G. A minimal paired dominating set with maximum cardinality is called a Γpr(G)-set. The upper paired domination number of G is the cardinality of a Γpr(G)-set of G. Denote by Upper-PDS the problem of finding a Γpr(G)-set of G. Upper paired domination was introduced by Dorbec et al. in [7]. They investigated the relationship between the upper total domination and upper paired domination numbers of a graph. Later, they established bounds on upper paired domination number for connected claw-free graphs[10]. Denote Pr(v)={u|u∉PD,N(u)∩PD={v},uv∈E(G)}, where PD is a minimal paired dominating set of G.
Recently, Michael et al. showed that Upper-PDS is NP-hard for split graphs and bipartite graphs, and APX-completeness of Upper-PDS for bipartite graphs with Δ=4 in [11]. In order to improve the results in [11], we show that Upper-PDS is APX-complete for bipartite graphs with Δ=3.
The class APX is the set of NP-optimization problems that allow polynomial-time approximation algorithms with approximation ratio bounded by a constant.
First, we recall the notation of L-reduction [1,15]. Given two NP-optimization problems H and G and polynomial time transformation f from instances of H to instances of G, we say that f is an L-reduction if there are positive constants α and β such that for every instance x of H:
(i) optG(f(x))≤αoptH(x);
(ii) for every feasible solution y of f(x) with objective value mG(f(x),y)=a, we can find a solution y′ of x with mH(x,y′)=b in polynomial time such that |optH(x)−b|≤β|optG(f(x))−a|.
To show that a problem P∈APX is APX-complete, it's enough to show that there is an L-reduction from some APX-complete problems to P.
Denote by MAX-MIN-VC the problem of finding a maximum minimal vertex cover of G. Note that, Minimum Domination problem is APX-complete even for bipartite graphs with maximum degree 3 [14], and Minimum Independent Domination problem [8] is the complement problem of MAX-MIN-VC in a graph G. We can obtain an L-reduction from Minimum Domination problem to Minimum Independent Domination problem by replacing every edge uv with a path Puv=uabcv with α=7, β=1. It's clear that Minimum Independent Domination problem and MAX-MIN-VC are in APX. Thus, Minimum Independent Domination problem is APX-complete even for bipartite graphs with maximum degree 3, so is MAX-MIN-VC (by Theorem 7 in [2]).
In this section, we show Upper-PDS for bipartite graphs with maximum degree 3 is APX-complete by providing an L-reduction f from MAX-MIN-VC for bipartite graphs with maximum degree 3.
We formalize the optimization problems as follows.
Lemma 1. [11] Upper-PDS can be approximated with a factor of 2Δ for graphs without isolated vertices and with maximum degree Δ.
Therefore, Upper-PDS is in APX.
Let G=(V,E) be a bipartite graph with |E|=m, Δ(G)=3.
For each edge xy∈E(G), let Hxy be the graph which is shown in Figure 1. Let T1={ a,...,a5, b,...,b5, r,...,r6, s,...,s6, c,..,c3, u,u1,v,v1}, T2={p,...,p6, q,...,q6, d,...,d3,w,z}, T3={h,..,h5}, T4={t,...,t5}, V(Hxy)=V(T1)∪V(T2)∪V(T3)∪V(T4)∪{w1,z1,x,y}, |V(Hxy)|=70.
Construct G′ by replacing each edge xy∈E(G) with the graph Hxy.
It's clear, Δ(G′)=3 and G′ is a bipartite graph.
Let Sp={p,p1,...,p6}, Sq={q,q1,...,q6}, Sa={a,a1,...,a5}, Sb={b,b1,...,b5}, Sr={r,r1,...,r6}, Ss={s,s1,...,s6}, Sc={c,c1,c2,c3}, Sd={d,d1,d2,d3}.
Let xy=e∈E(G), H′e=H′xy=Hxy−{x,y}, |V(H′xy)|=68.
Let PD be a paired dominating set of G′, uv∈E(G). We say H′uv is [I,O] if u is HIuv and v is HOuv, or if v is HIuv and u is HOuv. We say H′uv is [I,0] if u is HIuv and v∉PD, or if u∉PD and v is HIuv. Analogously, H′uv could be [0,0] ([I,I] or [O,O] or [O,0]).
Note that
|T1∩PD|=|Sa∩PD|+|Sb∩PD|+|Sc∩PD|+|Sr∩PD|+|Ss∩PD|+|{u,u1,v,v1}∩PD|, | (2.1) |
|T2∩PD|=|Sp∩PD|+|Sq∩PD|+|Sd∩PD|+|{w,z}∩PD|, | (2.2) |
|V(H′xy)∩PD|=|T1∩PD|+|T2∩PD|+|T3∩PD|+|T4∩PD|+|{w1,z1}∩PD|. | (2.3) |
The following lemma is immediate.
Lemma 2. Let PD be a minimal paired dominating set of G, M be a perfect matching of G[PD]. If v,u are support vertices, uv∈E(G), x∈L(v), y∈L(u), then |{x,y}∩PD|≤1.
Lemma 3. Let PD be a minimal paired dominating set of G′, M be a perfect matching of G′[PD]. For each Hxy, we have
(a) |Sc∩PD|=4 if and only if r,s∉PD, Pr(c3)≠∅ or Pr(c)≠∅.
(b) |Sd∩PD|=4 if and only if p,q∉PD, Pr(d3)≠∅ or Pr(d)≠∅.
(c) |T3∩PD|≤4 with equality if and only if (i) or (ii) holds,
(i) Pr(h1)≠∅ if h∉PD,
(ii) N(h)∩PD={h1} or Pr(h)≠∅ if h∈PD.
And if h is G[T3]O, |T3∩PD|=3.
(d) |T4∩PD|≤4 with equality if and only if (i) or (ii) holds,
(i) Pr(t1)≠∅ if t∉PD,
(ii) N(t)∩PD={t1} or Pr(t)≠∅ if t∈PD.
And if t is G[T4]O, |T4∩PD|=3.
(e) |Sa∩PD|≤4 with equality if and only if (i) or (ii) holds,
(i) Pr(a1)≠∅ if a∉PD,
(ii) N(a)∩PD={a1} or Pr(a)≠∅ if a∈PD.
And if a is G[Sa]O, |Sa∩PD|=3.
(f) |Sb∩PD|≤4 with equality if and only if (i) or (ii) holds,
(i) Pr(b1)≠∅ if b∉PD,
(ii) N(b)∩PD={b1} or Pr(b)≠∅ if b∈PD.
And if b is G[Sb]O, |Sb∩PD|=3.
(g) 3≤|Sr∩PD|≤4. And if r∈PD and r is G[Sr]O, |Sr∩PD|=3.
(h) 3≤|Ss∩PD|≤4. And if s∈PD and s is G[Sr]O, |Ss∩PD|=3.
(i) 3≤|Sp∩PD|≤4. And if p∈PD and p is G[Sp]O, |Sp∩PD|=3.
(j) 3≤|Sq∩PD|≤4. And if q∈PD and q is G[Sq]O, |Sq∩PD|=3.
(k) |T2∩PD|≤13 with equality if and only if |{w,z}∩PD|=1.
(l) |T1∩PD|≤22 with equality if and only if {u,v}⊆PD.
(m) If {u,v}∩PD=∅, |T1∩PD|≤18.
Proof. (a) W.l.o.g. we consider r∈PD. If rc∈M, |Sc∩PD|≠4. Otherwise, c1c2,c3s∈M and let PD′=PD∖{c1,c2}, M′=M∖{c1c2}. Then, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction. If rc∉M, |Sc∩PD|≠4. Otherwise, cc1,c2c3∈M and let PD′=PD∖{c,c1}, M′=M∖{cc1}. Then, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction.
If Pr(c3)=∅ and Pr(c)=∅, let PD′=PD∖{c,c3} and M′=M∖{cc1,c2c3}∪{c1c2}. Then, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction.
(b) The proof is analogous to that of (a), and the proof is omitted.
(c) Clearly, |T3∩PD|≤4. If |T3∩PD|=4, {h1,h2,h3,h4}⊆PD or {h,h1,h2,h3}⊆PD. If {h1,h2,h3,h4}⊆PD, Pr(h1)≠∅. Otherwise, let PD′=PD∖{h4,h1}, M′=M∖{h3h4,h1h2}∪{h2h3}. Then, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction. If {h,h1,h2,h3}⊆PD, N(h)∩PD={h1} or Pr(h)≠∅. Otherwise, let PD′=PD∖{h,h1}, M′=M∖{hh1}. Then, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction.
If h is G[T3]O, and since |T3∖{h}∩PD| is even, we have |T3∩PD|=3.
(d)–(f) We obtain the conclusions with a similar proof of (c).
(g) Clearly, 3≤|Sr∩PD|≠6. If |Sr∩PD|=5, we obtain Sr∩PD={r,r1,r2,r3,r4} or Sr∩PD={r,r2,r3,r4,r5} by Lemma 2. Therefore, PD is not a minimal paired dominating set, a contradiction. Thus, 3≤|Sr∩PD|≤4.
If r is G[Sr]O, and since |Sr∖{r}∩PD| is even, we have |Sr∩PD|=3.
(k) Since |Sd∩PD|≤4, |T2∩PD|≤14 by (i)–(j) and Eq (2.1).
If |{w,z}∩PD|=0, |T2∩PD|≤12.
If |{w,z}∩PD|=1, |T2∩PD|≤13.
Then we consider |{w,z}∩PD|=2. If w is G[T2]I or z is G[T2]I, we may assume w is G[T2]I. We obtain wp∈M, |Sp∩PD|=3 by (i), |Sd∩PD|≤3 by (b). Therefore, |T2∩PD|≤12 by Eq (2.1). If w,z are G[T2]O, |Sd∩PD|≤3 by (b). Since |T2∩PD| is even, |T2∩PD|≤12 by Eq (2.1).
Thus, |T2∩PD|≤13 with equality if and only if |{w,z}∩PD|=1, see Figure 2 (a).
(h)–(j) Using similar arguments of (g), the conclusions follow.
(l)–(m) We discuss the following cases.
Case 1. |{u,v}∩PD|=2.
In this case, we have |{u1,v1}∩PD|≥1, |Sa∩PD|≥3 and |Sc∩PD|≥3, otherwise, |T1∩PD|≤22 by (e)–(h) and Eq (2.2).
W.l.o.g. we assume u1∈PD.
First, we assume that |Sa∩PD|=4. We obtain aa1,uu1∈M, {r,c,r1}∩PD=∅ by (e). Thus, |Sc∩PD|≤3. Then, we consider |Sc∩PD|=3, that is, c3s∈M. By (h), we have |Ss∩PD|=3. Therefore, |T1∩PD|≤22 by Eq (2.2).
Then, we consider |Sa∩PD|=3. Therefore, v1∈PD, otherwise, |T1∩PD|≤22 by Eq (2.2).
We have |Sb∩PD|=4, otherwise, |T1∩PD|≤22 by Eq (2.2). By (f), {s,s1,c3}∩PD=∅. Thus, |Sc∩PD|≤3. Therefore, |T1∩PD|≤22 by Eq (2.2), see Figure 2 (b).
Case 2. |{u,v}∩PD|=1.
W.l.o.g. we assume u∈PD.
We have |{u1,v1}∩PD|≥1 and |Sa∩PD|≥3, otherwise, |T1∩PD|≤21 by Eq (2.2).
Case 2.1 u1∈PD.
If |Sa∩PD|=4, {r,r1,c}∩PD=∅ by (e). Then |Sc∩PD|≤3. If |Sc∩PD|=3, we have c3s∈M, |Ss∩PD|≤3 by (h). Therefore, |T1∩PD|≤21 by Eq (2.2). If |Sc∩PD|=2, |T1∩PD|≤21 by Eq (2.2).
Now we consider |Sa∩PD|=3. If v1∉PD, |T1∩PD|≤21 by Eq (2.2). Thus, v1∈PD, that is, v1b∈M. Therefore, |Sb∩PD|=3 by (f), |T1∩PD|≤21 by Eq (2.2).
Case 2.2 u1∉PD.
If v1∈PD, v1b∈M. Therefore, |Sb∩PD|=3 by (f), |T1∩PD|≤21 by Eq (2.2). Thus, v1∉PD, and |T1∩PD|≤21 by Eq (2.2).
Case 3. |{u,v}∩PD|=0.
In this case, |T1∩PD| is even.
Case 3.1 |{u1,v1}∩PD|≥1.
W.l.o.g. we assume u1∈PD. Then u1a∈M, |Sa∩PD|=3 by (e). If v1∉PD, b∈PD. By (a), |Sc∩PD|≤3. Therefore, |T1∩PD|≤19 by Eq (2.2). If v1∈PD, we obtain v1b∈M, |Sb∩PD|=3 by (f). |Sc∩PD|≤3 by (a). Therefore, |T1∩PD|≤19 by Eq (2.2).
Case 3.2 |{u1,v1}∩PD|=0.
In this case, a,b∈PD. By (a), |Sc∩PD|≤3. Therefore, |T1∩PD|≤19 by Eq (2.2).
Note that |T1∩PD| is even, so |T1∩PD|≤18, see Figure 2 (c).
Thus, (l) and (m) hold.
Lemma 4. Let PD be a minimal paired dominating set of G′.
(a) |V(H′xy)∩PD|≤43.
(b) If {xw1,yz1}⊂M, |V(H′xy)∩PD|≤42.
(c) If {xw1,yz1}∩M=∅ and {w1,z1}⊆PD, |V(H′xy)∩PD|≤42.
(d) If xw1∉M(G), w1∈PD and y∉PD, then |V(H′xy)∩PD|≤42.
Proof. (a) By Lemma 3 and Eq (2.3),
|V(H′xy)∩PD|=|T1∩PD|+|T2∩PD|+|T3∩PD|+|T4∩PD|+|{w1,z1}∩PD|≤22+13+4+4+2=45. |
We consider that {w1,z1}∩PD≠∅, |T4∩PD|≥3 and |T3∩PD|≥3, otherwise, |V(H′xy)∩PD|≤43. Then, w.l.o.g. we assume that w1∈PD.
If |T4∩PD|=4, {tt1,t2t3}⊆M or {t1t2,t3t4}⊆M.
If {tt1,t2t3}⊆M, Pr(t)≠∅ or N(t)∩PD={t1}. If Pr(t)≠∅, u∈Pr(t). By Lemma 3 (m), |V(H′xy)∩PD|≤43. If N(t)∩PD={t1}, we have u,w1∉PD, |T1∩PD|≤21 by Lemma 3 (l). Then we obtain z∈PD, otherwise, |V(H′xy)∩PD|≤43 by Lemma 3 (k) and Eq (2.3). If |T3∩PD|=4, we have v∉PD, therefore, |V(H′xy)∩PD|≤43 by Eq (2.3). If |T3∩PD|=3, |V(H′xy)∩PD|≤43 by Eq (2.3).
If {t1t2,t3t4}⊆M, {w,u}∩PD=∅. By Lemma 3 (l), |T1∩PD|≤21, and v∈PD. If z∉PD, |V(H′xy)∩PD|≤43 by Lemma 3 (k) and Eq (2.3). If z∈PD, |T3∩PD|≤3 by Lemma 3 (c). Therefore, |V(H′xy)∩PD|≤43 by Eq (2.3).
If |T4∩PD|=3, we consider |T1∩PD|=22, and {u,v,z1}∈PD. We have |T3∩PD|≠4 by Lemma 3 (c). Therefore, |V(H′xy)∩PD|≤43 by Eq (2.3).
(b)–(d) Since |V(H′xy)∩PD|≤43, and, |V(H′xy)∩PD| is even in those cases, so |V(H′xy)∩PD|≤42.
Lemma 5. Let PD be a minimal paired dominating set of G′, M be a perfect matching of G′[PD].
(a) If {x,w1,y,z1}⊂PD, xw1∈M(G) and yz1∉M, we have |V(H′xy)∩PD|≤41.
(b) If {x,y}∩PD=∅, |V(H′xy)∩PD|≤40.
Proof. (a) In this case, we have z∈PD and zz1∈M.
Since |V(H′xy)∩PD| is odd, it's sufficient to show |V(H′xy)∩PD|≤42. We only consider {u,v}∩PD≠∅ by Lemma 3 (m).
Case 1. |T4∩PD|=4.
In this case, we have {t1t2,t3t4}⊆M or {tt1,t2t3}⊆M. If {t1t2,t3t4}⊆M (or {tt1,t2t3}⊆M), we obtain u,w∉PD, v∈PD. Since z,v∈PD, |T3∩PD|≤3 by Lemma 3 (c). If |T3∩PD|=2, |V(H′xy)∩PD|≤42 by Eq (2.3). If |T3∩PD|=3, hv∈M. Thus, {q,q1,d3}∩PD=∅, otherwise, let PD′=PD∖{z,z1}, M′=M∖{zz1}. Then, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction. Since zz1∈M, we obtain that |T2∩PD| is odd. So |T1∩PD|≤12. Therefore, |V(H′xy)∩PD|≤42 by Eq (2.3), see Figure 3 (a).
Case 2. |T4∩PD|=3.
If tw∈M, u∉PD or {p,p1,d}∩PD=∅. Otherwise, let PD′=PD∖{t,w}, M′=M∖{tw}. Then, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction. If {p,p1,d}∩PD=∅, |Sd∩PD|≤3. W have |Sd∩PD|=3, d3q∈M, otherwise, |V(H′xy)∩PD|≤42 by Eq (2.3). Thus |Sq∩PD|≤3 by Lemma 3 (j) and Eq (2.3), and |V(H′xy)∩PD|≤42. If u∉PD, v∈PD. Thus, |T3∩PD|≤3 by Lemma 3 (c) and Eq (2.3), and |V(H′xy)∩PD|≤42.
If tu∈M, |T3∩PD|≤3 by Lemma 3 (c). We have |T3∩PD|=3, hv∈M, otherwise, |V(H′xy)∩PD|≤42 by Eq (2.3). Let PD′=PD∖{t,h}, M′=M∖{tu,hv}∪{uv}. Therefore, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction.
Case 3. |T4∩PD|=2.
Now we only consider |T1∩PD|=22, and {u,v}⊂PD. By Lemma 3 (c) and Eq (2.3), |V(H′xy)∩PD|≤42.
(b) Since |V(H′xy)∩PD| is even, it's sufficient to show |V(H′xy)∩PD|≤41.
Case 1. |{z1,w1}∩PD|=0.
We obtain {z,w}⊆PD, |T2∩PD|≤12 by Lemma 3 (k). If |T4∩PD|≤3, |V(H′xy)∩PD|≤41 by Eq (2.3), see Figure 3 (b). If |T4∩PD|=4, t∈PD and Pr(t)≠∅ by Lemma 3(d). So, {u,v,u1}∩PD=∅. By Lemma 3 (m) and Eq (2.3), |V(H′xy)∩PD|≤40.
Case 2. |{z1,w1}∩PD|=1.
W.l.o.g. we assume w1∈PD. Thus, ww1∈M, z∈PD, |T2∩PD|≤12 by Lemma 3 (k). If |T4∩PD|=2, |V(H′xy)∩PD|≤41 by Eq (2.3). If |T4∩PD|=4, we obtain Pr(t)={u} for t∈PD, {u,u1,v}∩PD=∅. By Lemma 3 (m) and Eq (2.3), |V(H′xy)∩PD|≤40. If |T4∩PD|=3, tu∈M. And v∈PD, otherwise |T1∩PD|≤21 by Lemma 3 (l), |V(H′xy)∩PD|≤41 by Eq (2.3). Thus, |T4∩PD|≠4 by Lemma 3 (d). Afterwards, |V(H′xy)∩PD|≤41 by Eq (2.3).
Case 3. |{z1,w1}∩PD|=2.
Thus, ww1∈M, zz1∈M, |T2∩PD|≤12 by Lemma 3 (k).
If |T4∩PD|=4, t∈PD and {u,u1,v}∩PD=∅. By Lemma 3 (m) and Eq (2.3), we have |V(H′xy)∩PD|≤40.
If |T4∩PD|=3, we have tu∈M, and |T3∩PD|≤3 by Lemma 3 (c). If |T3∩PD|=2, |V(H′xy)∩PD|≤41 by Eq (2.3). If |T3∩PD|=3, hv∈M. Let PD′=PD∖{t,h}, M′=M∖{tu,hv}∪{uv}. Therefore, PD′ is a paired dominating set and PD is not a minimal paired dominating set, a contradiction.
If |T4∩PD|=2, we only consider |T1∩PD|=22. Thus, u,v∈PD. By Lemma 3 (c), |T3∩PD|≤3. Therefore, |V(H′xy)∩PD|≤41 by Eq (2.3).
Corollary 6. Let PD be a minimal paired dominating set of G′. If |V(H′uv)∩PD|=43 if and only if |{u,v}∩PD|=1, and, u or v is HIuv.
Lemma 7. If VC1 is a minimal vertex cover of G, there exists a minimal paired dominating set PD1 of G′ with |PD1|=42m+2|VC|.
Proof. A minimal paired dominating set PD1 can be constructed by the following manner:
For each vertex x∈VC1, we have |N(x)∩VC1|<d(x)≤3. So there exists at least one edge xx1 with x1∉VC1 in G, and maybe exist edges xx2 or xx3.
Therefore, for the edge xx1, put i into PD′ for i∈{x,w1, p2,p3,p4,p5, d,d1,d2,d3, q2,q3,q4,q5, z,z1, h,h2,h3, v, b1,b2,b3,b4, s2,s3,s4,s5, c,c1,c2,c3, r2,r3,r4,r5, a,a1,a2,a3, t1,t2,t3,t4}. Put j into M for j∈{xw1, p5p4,p3p2,dd1,d2d3, q2q3,q4q5, zz1, hv, h2h3, b1b2,b3b4, s2s3,s4s5, cc1,c2c3, r2r3,r4r5, aa1,a2a3, t1t2,t3t4}. See Figure 4 (a).
For edges xx2,xx3, put i into PD′ for i∈{x, p2,p3,p4,p5, d,d1,d2,d3, q2,q3,q4,q5, z,z1, u,v, h2,h3, b1,b2,b3,b4, s2,s3,s4,s5, c,c1,c2,c3, r2,r3,r4,r5, a1,a2,a3,a4, t1,t2,t3,t4}. Put j into M for j∈{ p5p4,p3p2,dd1,d2d3, q2q3,q4q5, zz1, h2h3, uv, b1b2,b3b4, s2s3,s4s5, cc1,c2c3, r2r3,r4r5, a1a2,a3a4, tt1,t2t3}. See Figure 4 (b).
Let PD1=PD′∪VC1. Since vertex x is M-saturated in PD1. Therefore, PD1 is a paired dominating set of G′.
Since N(w)∩PD1={w1}, then PD1∖{w1} is not a dominating set of G′. So PD1 is a minimal paired dominating set of G′. And |PD1|=|VC1|+|VC1|×43+(m−|VC1|)×42. Therefore, |PD1|=2|VC1|+42m.
Let PD be a minimal paired dominating set of G′. Algorithm 1 is to obtain a minimal vertex cover VC of G, and it terminates in polynomial time.
Algorithm 1 CONST-VC(G′,PD) |
Input: A graph G′ with a minimal paired dominating set PD Output: A graph G with a minimal vertex cover VC 1: VC=PD 2: for every Hxy⊆G′ do 3: Delete vertices in H′xy 4: Add an edge between x and y {obtained the graph G} 5: VC=VC∖V(H′xy) 6: end for 7: VC′=VC 8: De=∅ {Mo is the set of vertex which is removed from VC.} 9: In=∅ {In is the set of vertex which is added into VC.} 10: Mo=∅ {De is the set of vertex which is added into VC at first, then removed from VC.} 11: while |N[v]∩VC|=d(v)+1 do 12: VC=VC∖{v},Mo=Mo∪{v} 13: end while 14: while uv∈E(G) and u,v∉VC do 15: VC=VC∪{u},In=In∪{u} 16: for w∈N(u) do 17: if |N[w]∩VC|=d(w)+1 then 18: VC=VC∖{w},De=De∪{w} 19: end if 20: end for 21: end while 22: return VC |
Lemma 8. If PD is a minimal paired dominating set of G′ and VC is a minimal vertex cover of G obtained by Algorithm 1, |VC|≥|PD|−42m−|VC|.
Proof. Let M be the perfect matching of G[PD], me=V(H′xy)∩PD where e=xy∈E(G), Me=⋃e∈E(G)me, Le=V(G)∖(Mo∪In∪De).
In Algorithm 1, we have:
Claim 9. (a) If v is put into Mo by the while loop (lines 11 to 13) or De (line 18), v will not be put into In later.
(b) For every vertex v∈V(G), v will be put into Mo (or De or In) at most once.
(c) Mo∩De=∅, Mo∩In=∅.
(d) If v∈De, there exists a vertex w∈N(v)∩In.
(e) If vertex v∈De∩In, we have v∉VC′, that is, v is put into In at first and then into De.
(f) If u,v∈De∪Mo, N(v)∩N(u)∩Mo∩De=∅.
(g) If v∈De∖In, there exists a vertex u∈N(v)∩(In∖De), u∉VC′. And |N(u)∩De|≤2. What's more, there exists a vertex w∈N(u)∖VC′. If w∈In∖De, |(N(u)∪N(w))∩(De∖In)|≤3.
Proof. (a) After v is put into De (or Mo), every w∈N(v) has a neighbor v which does not belong to VC, so w will not be put into De. Therefore, v will not be put into In later.
(b)–(d) By (a), it is immediate.
(e) By (a) and (c), it is immediate.
(f) Suppose v is put into De∪Mo. By (a), w∈N(v) will not be put into De∪Mo.
(g) For vertex v∈De∖In, by (d) and (f), let u∈N(v)∩(In∖De), and u∉VC′, |N(u)∩De|≤2.
Since u∈In∖De, there exists a vertex w∈N(u)∖VC′.
Since 1≤|N(u)∩(De∖In)|≤2, |N(w)∩(De∖In)|≤2. If w∈In∖De, we may assume u is put into In at first. Then N(u)∩(De∖In)|≤1, otherwise, w will not be put into In later. Therefore, |(N(u)∪N(w))∩(De∖In)|≤3.
Thus,
|VC|=|PD|−|Me|−|Mo|−|De|+|In|. | (2.4) |
To show that |Me|+|Mo|+|De|−|In|≤42m+|VC|, we use the following strategy.
Discharging procedure:
In the graph G′, we set the initial charge of every vertex v to be s(v)=1 for v∈Mo∪Me∪(De∖In), s(v)=−1 for v∈In∖De, s(v)=0 otherwise, s(H′uv)=∑x∈V(H′uv)s(x), s(G′)=∑v∈V(G′)s(v).
Obviously,
∑v∈V(G′)s(v)=|Me|+|Mo|+|De|−|In|. | (2.5) |
We use the discharging procedure, leading to a final charge s′, defined by applying the following rules:
Rule 1: For the vertex v∈Mo, v is M-saturated. Therefore, v is HIuv for u. If u is HIuv, s(v) transmits 1 charge to s(u). If u is HOuv, s(v) transmits 1 charge to s(H′uv) which is [I,O].
Rule 2: For each s(H′uv)=43, by Corollary 6, s(H′uv) transmits 1 charge to u∈VC′.
Rule 3: For the vertex v∈De∖In, by Claim 9 (g), there exists a vertex u∈N(v)∩(In∖De), and a vertex w∈N(u)∖VC′ and |N(u)∩De|≤2. If |N(u)∩De|=2, s(v) transmits 1 charge to s(u) and transmits 1 charge to s(H′uw) which is [0,0]. If |N(u)∩De|=1, s(v) transmits 2 charge to s(u).
After discharging, we have:
Claim 10. (a) s′(v)≤0 for v∈Mo∪(De∖In)∪(Le∖VC)∪(In∩De).
(b) For each H′xy, s′(H′xy)≤42.
(c) s′(v)≤1 for v∈(In∖De)∪(Le∩VC).
Proof. (a) If v∈Mo, by Claim 9 (f), v will not receive any charge by Rules 1 and 3. Since N[v]∩VC′=N[v]. By Lemmas 4 and 5, v will not receive any charge by Rule 2. Therefore, s′(v)=0.
If v∈De∖In, v∈VC′. By Claim 9 (f), N(v)∩Mo=∅. Thus, v will not receive any charge by Rules 1 and 3. Since v is HIuv for u. By Lemmas 4 and 5, if u∈VC′, v will not receive any charge by Rule 2. If u∉VC′, v will receive 1 charge at most by Rule 2. Afterwards, by Rule 3, v will transmit 2 charge to others, so s′(v)≤0.
If v∈Le∖VC, v will not receive any charge by Rules 1, 2 and 3.
If v∈In∩De, v∉VC′ by Claim 9 (e). Thus, v will not receive any charge by Rules 1 and 2. By Claim 9 (f), v∈De, N(v)∩De=∅. Thus, v will not receive any charge by Rule 3.
(b) If H′uw is [I,I] or [O,O] or [I,0] or [O,0], s(H′uw) will not receive any charge by Rules 1, 2 and 3. If H′uw is [0,0], s(H′uw) will not receive any charge by Rules 1 and 2.
If H′uw is [0,0], by Claim 9 (g), |(N(u)∪N(w))∩(De∖In)|≤3. Thus, s(H′uw) will receive 2 charge at most from s(x) where x∈N(v)∖{w} by Rule 3.
And if s(H′uw)=43, by Corollary 6, there exists a vertex u∈VC′ and u is HIuw. Therefore, s′(H′uw)=42 by Rule 2.
Thus, by Lemmas 4 and 5, s′(H′uw)≤42.
(c) If v∈In∖De, v∉VC′, v will receive any charge by Rules 1 and 2. And there exists a vertex w∈N(v) w∉VC′ and w∉De∖In. So v will receive 2 charge at most by Rule 3, s′(v)≤−1+2=1.
If v∈Le∩VC, v will receive any charge by Rule 3. By Lemmas 4, 5 and Corollary 6, H′uv is [I,0] if s(H′uv)=43. Since v can be M-saturated once, v will receive 1 charge at most by Rules 1 and 2. Thus, s′(v)≤0+1=1.
By Claim 10,
|Me|+|Mo|+|De|−|In|=∑uv∈E(G)s(H′uv)+∑v∈Mos(v)+∑v∈De∖Ins(v)−∑v∈In∖Des(v)=∑uv∈E(G)s′(H′uv)+∑v∈Mos′(v)+∑v∈De∖Ins′(v)+∑v∈In∖Des′(v)+∑v∈In∩Des′(v)+∑v∈Le∖VCs′(v)∑v∈Le∩VCs′(v)≤42m+|In∖De|+|Le∩VC|≤42m+|VC|. |
Thus, by Eq (2.4),
|VC|=|PD|−|Me|−|Mo|−|De|+|In|≥|PD|−42m−|VC|. |
Let PD∗ be a Γpr(G′)-set of G′, and be the Input of Algorithm 1. Then we obtain the Output VC by Algorithm 1.
Since
|VC∗|≥mΔ=m3. |
By Lemma 8,
|VC|≥|PD∗|−42m−|VC|≥|PD∗|−42×3|VC|−|VC||VC|≥|PD∗|−127|VC| |
Let VC∗ be a VC-set of G. Since |VC|≤|VC∗|,
|PD∗|≤128|VC|≤128|VC∗| | (2.6) |
By Lemma 7, |PD∗|≥|PD1|=42m+2|VC∗|. By Lemma 8,
|PD|−|VC|≤|VC|+42m≤|VC∗|+42m≤|PD∗|−|VC∗|. |
Thus,
|VC∗|−|VC|≤|PD∗|−|PD| | (2.7) |
Therefore, by Eq (2.6) and Eq (2.7), f is an L-reduction with α=128, β=1.
Upper-PDS for bipartite graphs is proved to be APX-complete with maximum degree 4 and still open with maximum degree 3. In this paper, we show that Upper-PDS for bipartite graphs with maximum degree 3 is APX-complete by providing an L-reduction f from MAX-MIN-VC for bipartite graphs to it.
This work was supported by the National Key R & D Program of China (No. 2018YFB1005100), the Guangzhou Academician and Expert Workstation (No. 20200115-9) and the Innovation Ability Training Program for Doctoral student of Guangzhou University (No. 2019GDJC-D01).
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
[1] |
E. J. Candes, J. Romberg, T. Tao, Robust uncertainty principles: exact signal reconstruction from highly incomplete frequency information, IEEE Trans. Inf. Theory, 52 (2006), 489–509. https://doi.org/10.1109/TIT.2005.862083 doi: 10.1109/TIT.2005.862083
![]() |
[2] |
E. T. Hale, W. Yin, Y. Zhang, Fixed-point continuation for l1-minimization: methodology and convergence, SIAM J. Optim., 19 (2008), 1107–1130. https://doi.org/10.1137/070698920 doi: 10.1137/070698920
![]() |
[3] |
Y. Li, C. Li, W. Yang, W. Zhang, A new conjugate gradient method with a restart direction and its application in image restoration, AIMS Math., 8 (2023), 28791–28807. https://doi.org/10.3934/math.20231475 doi: 10.3934/math.20231475
![]() |
[4] |
A. Chambolle, An algorithm for total variation minimization and applications, J. Math. Imag. Vis., 20 (2004), 89–97. https://doi.org/10.1023/B:JMIV.0000011325.36760.1e doi: 10.1023/B:JMIV.0000011325.36760.1e
![]() |
[5] |
L. Rudin, S. Osher, E. Fatemi, Nonlinear total variation based noise removal algorithms, Phys. D, 60 (1992), 259–268. https://doi.org/10.1016/0167-2789(92)90242-F doi: 10.1016/0167-2789(92)90242-F
![]() |
[6] |
N. Artsawang, Accelerated preconditioning Krasnosel'skii-Mann method for efficiently solving monotone inclusion problems, AIMS Math., 8 (2023), 28398–28412. https://doi.org/10.3934/math.20231453 doi: 10.3934/math.20231453
![]() |
[7] |
P. H. Calamai, J. J. Moré, Projeeted gradient methods for linearly constralned problems, Math. Program., 39 (1987), 93–116. https://doi.org/10.1007/BF02592073 doi: 10.1007/BF02592073
![]() |
[8] | D. P. Bertsekas, Nonlinear programming, Athena scientific, Belmont, 1999. |
[9] |
J. Barzilai, J. M. Borwein, Two point step size gradient methods, IMA J. Numer. Anal., 8 (1988), 141–148. https://doi.org/10.1093/imanum/8.1.141 doi: 10.1093/imanum/8.1.141
![]() |
[10] |
E. G. Birgin, J. M. Martínez, M. Raydan, Nonmonotone spectral projected gradient methods on convex sets, SIAM J. Optim., 10 (2000), 1196–1211. https://doi.org/10.1137/S1052623497330963 doi: 10.1137/S1052623497330963
![]() |
[11] |
Y. H. Dai, R. Fletcher, Projected Barzilai-Borwein methods for large-scale box-constrained quadratic programming, Numer. Math., 100 (2005), 21–47. https://doi.org/10.1007/s00211-004-0569-y doi: 10.1007/s00211-004-0569-y
![]() |
[12] |
B. Zhang, Z. Zhu, S. Li, A modified spectral conjugate gradient projection algorithm for total variation image restoration, Appl. Math. Lett., 27 (2014), 26–35. https://doi.org/10.1016/j.aml.2013.08.006 doi: 10.1016/j.aml.2013.08.006
![]() |
[13] |
I. Daubechies, M. Defrise, C. De Mol, An iterative thresholding algorithm for linear inverse problems with a sparsity constraint, Commun. Pure Appl. Math., 57 (2004), 1413–1457. https://doi.org/10.1002/cpa.20042 doi: 10.1002/cpa.20042
![]() |
[14] |
A. Beck, M. Teboulle, A fast iterative shrinkage-thresholding algorithm for linear inverse problems, SIAM J. Imag. Sci., 2 (2009), 183–202. https://doi.org/10.1137/080716542 doi: 10.1137/080716542
![]() |
[15] |
E. T. Hale, W. Yin, Y. Zhang, Fixed-point continuation applied to compressed sensing: implementation and numerical experiments, J. Comput. Math., 28 (2010), 170–194. https://doi.org/10.4208/jcm.2009.10-m1007 doi: 10.4208/jcm.2009.10-m1007
![]() |
[16] |
S. J. Wright, R. D. Nowak, T. Figueiredo, Sparse reconstruction by separable approximation, IEEE Trans. Signal Process., 57 (2009), 2479–2493. https://doi.org/10.1109/TSP.2009.2016892 doi: 10.1109/TSP.2009.2016892
![]() |
[17] |
W. W. Hager, D. T. Phan, H. Zhang, Gradient-based methods for sparse recovery, SIAM J. Imag. Sci., 4 (2011), 146–165. https://doi.org/10.1137/090775063 doi: 10.1137/090775063
![]() |
[18] |
Y. Huang, H. Liu, A Barzilai-Borwein type method for minimizing composite functions, Numer. Algorithms, 69 (2015), 819–838. https://doi.org/10.1007/s11075-014-9927-8 doi: 10.1007/s11075-014-9927-8
![]() |
[19] |
W. Cheng, Z. Chen, D. Li, Nonmonotone spectral gradient method for sparse recovery, Inverse Probl. Imag., 9 (2015), 815–833. https://doi.org/10.3934/ipi.2015.9.815 doi: 10.3934/ipi.2015.9.815
![]() |
[20] |
Y. Xiao, S. Y. Wu, L. Qi, Nonmonotone Barzilai-Borwein gradient algorithm for l1-regularized nonsmooth minimization in compressive sensing, J. Sci. Comput., 61 (2014), 17–41. https://doi.org/10.1007/s10915-013-9815-8 doi: 10.1007/s10915-013-9815-8
![]() |
[21] |
P. Tseng, S. Yun, A coordinate gradient descent method for nonsmooth separable minimization, Math. Program., 117 (2009), 387–423. https://doi.org/10.1007/s10107-007-0170-0 doi: 10.1007/s10107-007-0170-0
![]() |
[22] |
E. Chouzenoux, J. C. Pesquet, A. Repetti, Variable metric forward-backward algorithm for minimizing the sum of a differentiable function and a convex function, J. Optim. Theory Appl., 162 (2014), 107–132. https://doi.org/10.1007/s10957-013-0465-7 doi: 10.1007/s10957-013-0465-7
![]() |
[23] |
P. L. Combettes, B. C. Vû, Variable metric forward-backward splitting with applications to monotone inclusions in duality, Optimization, 63 (2014), 1289–1318. https://doi.org/10.1080/02331934.2012.733883 doi: 10.1080/02331934.2012.733883
![]() |
[24] |
S. Bonettini, I. Loris, F. Porta, M. Prato, Variable metric inexact line-search-based methods for nonsmooth optimization, SIAM J. Optim., 26 (2016), 891–921. https://doi.org/10.1137/15M1019325 doi: 10.1137/15M1019325
![]() |
[25] |
S. Rebegoldi, L. Calatroni, Scaled, inexact and adaptive generalized FISTA for strongly convex optimization, SIAM J. Optim., 32 (2022), 2428–2459. https://doi.org/10.1137/21M1391699 doi: 10.1137/21M1391699
![]() |
[26] |
S.Bonettini, M. Prato, S. Rebegoldi, A new proximal heavy ball inexact line-search algorithm, Comput. Optim. Appl., 2024. https://doi.org/10.1007/s10589-024-00565-9 doi: 10.1007/s10589-024-00565-9
![]() |
[27] |
, S. Bonettini, P. Ochs, M. Prato, S. Rebegoldi, An abstract convergence framework with application to inertial inexact forward-backward methods, Comput. Optim. Appl., 84 (2023), 319–362. https://doi.org/10.1007/s10589-022-00441-4 doi: 10.1007/s10589-022-00441-4
![]() |
[28] |
H. Liu, T. Wang, Z. Liu, A nonmonotone accelerated proximal gradient method with variable stepsize strategy for nonsmooth and nonconvex minimization problems, J. Glob. Optim., 2024. https://doi.org/10.1007/s10898-024-01366-4 doi: 10.1007/s10898-024-01366-4
![]() |
[29] |
Y. Chen, W. W. Hager, M. Yashtini, X. Ye, H. Zhang, Bregman operator splitting with variable stepsize for total variation image reconstruction, Comput. Optim. Appl., 54 (2013), 317–342. https://doi.org/10.1007/s10589-012-9519-2 doi: 10.1007/s10589-012-9519-2
![]() |
[30] |
Y. Chen, W. W. Hager, F. Huang, D. T. Phan, X. Ye, W. Yin, Fast algorithms for image reconstruction with application to partially parallel MR imaging, SIAM J. Imag. Sci., 5 (2012), 90–118. https://doi.org/10.1137/100792688 doi: 10.1137/100792688
![]() |
[31] |
S. Bonettini, M. Prato, New convergence results for the scaled gradient projection method, Inverse Probl., 31 (2015), 095008. https://doi.org/10.1088/0266-5611/31/9/095008 doi: 10.1088/0266-5611/31/9/095008
![]() |
[32] |
F. Porta, M. Prato, L. Zanni, A new steplength selection for scaled gradient methods with application to image deblurring, J. Sci. Comput., 65 (2015), 895–919. https://doi.org/10.1007/s10915-015-9991-9 doi: 10.1007/s10915-015-9991-9
![]() |
[33] |
L. Grippo, F. Lampariello, S. Lucidi, A nonmonotone line search technique for Newton's method, SIAM J. Numer. Anal., 23 (1986), 707–716. https://doi.org/10.1137/0723046 doi: 10.1137/0723046
![]() |
[34] |
K. Amini, M. Ahookhosh, H. Nosratipour, An inexact line search approach using modified nonmonotone strategy for unconstrained optimization, Numer. Algorithms, 66 (2014), 49–78. https://doi.org/10.1007/s11075-013-9723-x doi: 10.1007/s11075-013-9723-x
![]() |
[35] | R. T. Rockafellar, Convex analysis, Princeton University Press, 2015. |
[36] |
Y. Ouyang, Y. Chen, G. Lan, J. E. Pasiliao, An accelerated linearized alternating direction method of multipliers, SIAM J. Imag. Sci., 8 (2015), 644–681. https://doi.org/10.1137/14095697X doi: 10.1137/14095697X
![]() |
[37] |
X. Ye, Y. Chen, F. Huang, Computational acceleration for MR image reconstruction in partially parallel imaging, IEEE Trans. Med. Imag., 30 (2011), 1055–1063. https://doi.org/10.1109/TMI.2010.2073717 doi: 10.1109/TMI.2010.2073717
![]() |
[38] |
S. Bonettini, I. Loris, F. Porta, M. Prato, S. Rebegoldi, On the convergence of a linesearch based proximal-gradient method for nonconvex optimization, Inverse Probl., 33 (2017), 055005. https://doi.org/10.1088/1361-6420/aa5bfd doi: 10.1088/1361-6420/aa5bfd
![]() |
[39] |
S. Bonettini, M. Prato, S. Rebegoldi, Convergence of inexact forward-backward algorithms using the forward-backward envelope, SIAM J. Optim., 30 (2020), 3069–3097. https://doi.org/10.1137/19M1254155 doi: 10.1137/19M1254155
![]() |
[40] |
M. K. Riahi, I. A. Qattan, On the convergence rate of Fletcher-Reeves nonlinear conjugate gradient methods satisfying strong Wolfe conditions: application to parameter identification in problems governed by general dynamics, Math. Methods Appl. Sci., 45 (2022), 3644–3664. https://doi.org/10.1002/mma.8009 doi: 10.1002/mma.8009
![]() |
[41] |
X. Li, Q. L. Dong, A. Gibali, PMiCA-Parallel multi-step inertial contracting algorithm for solving common variational inclusions, J. Nonlinear Funct. Anal., 2022 (2022), 7. https://doi.org//10.23952/jnfa.2022.7 doi: 10.23952/jnfa.2022.7
![]() |
[42] |
L. O. Jolaoso, Y. Shehu, J. Yao, R. Xu, Double inertial parameters forward-backward splitting method: applications to compressed sensing, image processing, and SCAD penalty problems, J. Nonlinear Var. Anal., 7 (2023), 627–646. https://doi.org/10.23952/jnva.7.2023.4.10 doi: 10.23952/jnva.7.2023.4.10
![]() |
Algorithm 1 CONST-VC(G′,PD) |
Input: A graph G′ with a minimal paired dominating set PD Output: A graph G with a minimal vertex cover VC 1: VC=PD 2: for every Hxy⊆G′ do 3: Delete vertices in H′xy 4: Add an edge between x and y {obtained the graph G} 5: VC=VC∖V(H′xy) 6: end for 7: VC′=VC 8: De=∅ {Mo is the set of vertex which is removed from VC.} 9: In=∅ {In is the set of vertex which is added into VC.} 10: Mo=∅ {De is the set of vertex which is added into VC at first, then removed from VC.} 11: while |N[v]∩VC|=d(v)+1 do 12: VC=VC∖{v},Mo=Mo∪{v} 13: end while 14: while uv∈E(G) and u,v∉VC do 15: VC=VC∪{u},In=In∪{u} 16: for w∈N(u) do 17: if |N[w]∩VC|=d(w)+1 then 18: VC=VC∖{w},De=De∪{w} 19: end if 20: end for 21: end while 22: return VC |
Algorithm 1 CONST-VC(G′,PD) |
Input: A graph G′ with a minimal paired dominating set PD Output: A graph G with a minimal vertex cover VC 1: VC=PD 2: for every Hxy⊆G′ do 3: Delete vertices in H′xy 4: Add an edge between x and y {obtained the graph G} 5: VC=VC∖V(H′xy) 6: end for 7: VC′=VC 8: De=∅ {Mo is the set of vertex which is removed from VC.} 9: In=∅ {In is the set of vertex which is added into VC.} 10: Mo=∅ {De is the set of vertex which is added into VC at first, then removed from VC.} 11: while |N[v]∩VC|=d(v)+1 do 12: VC=VC∖{v},Mo=Mo∪{v} 13: end while 14: while uv∈E(G) and u,v∉VC do 15: VC=VC∪{u},In=In∪{u} 16: for w∈N(u) do 17: if |N[w]∩VC|=d(w)+1 then 18: VC=VC∖{w},De=De∪{w} 19: end if 20: end for 21: end while 22: return VC |