
In 3D reconstruction tasks, camera parameter matrix estimation is usually used to present the single view of an object, which is not necessary when mapping the 3D point to 2D image. The single view reconstruction task should care more about the quality of reconstruction instead of the alignment. So in this paper, we propose an implicit field knowledge distillation model (IFKD) to reconstruct 3D objects from the single view. Transformations are performed on 3D points instead of the camera and keep the camera coordinate identified with the world coordinate, so that the extrinsic matrix can be omitted. Besides, a knowledge distillation structure from 3D voxel to the feature vector is established to further refine the feature description of 3D objects. Thus, the details of a 3D model can be better captured by the proposed model. This paper adopts ShapeNet Core dataset to verify the effectiveness of the IFKD model. Experiments show that IFKD has strong advantages in IOU and other core indicators compared with the camera matrix estimation methods, which verifies the feasibility of the new proposed mapping method.
Citation: Jianyuan Wang, Huanqiang Xu, Xinrui Hu, Biao Leng. IFKD: Implicit field knowledge distillation for single view reconstruction[J]. Mathematical Biosciences and Engineering, 2023, 20(8): 13864-13880. doi: 10.3934/mbe.2023617
[1] | Cameron Browne . Immune response in virus model structured by cell infection-age. Mathematical Biosciences and Engineering, 2016, 13(5): 887-909. doi: 10.3934/mbe.2016022 |
[2] | Khalid Hattaf, Noura Yousfi . Dynamics of SARS-CoV-2 infection model with two modes of transmission and immune response. Mathematical Biosciences and Engineering, 2020, 17(5): 5326-5340. doi: 10.3934/mbe.2020288 |
[3] | Ting Guo, Zhipeng Qiu . The effects of CTL immune response on HIV infection model with potent therapy, latently infected cells and cell-to-cell viral transmission. Mathematical Biosciences and Engineering, 2019, 16(6): 6822-6841. doi: 10.3934/mbe.2019341 |
[4] | Chentong Li, Jinhu Xu, Jiawei Liu, Yicang Zhou . The within-host viral kinetics of SARS-CoV-2. Mathematical Biosciences and Engineering, 2020, 17(4): 2853-2861. doi: 10.3934/mbe.2020159 |
[5] | A. M. Elaiw, A. S. Shflot, A. D. Hobiny . Stability analysis of general delayed HTLV-I dynamics model with mitosis and CTL immunity. Mathematical Biosciences and Engineering, 2022, 19(12): 12693-12729. doi: 10.3934/mbe.2022593 |
[6] | Xuejuan Lu, Lulu Hui, Shengqiang Liu, Jia Li . A mathematical model of HTLV-I infection with two time delays. Mathematical Biosciences and Engineering, 2015, 12(3): 431-449. doi: 10.3934/mbe.2015.12.431 |
[7] | Jiawei Deng, Ping Jiang, Hongying Shu . Viral infection dynamics with mitosis, intracellular delays and immune response. Mathematical Biosciences and Engineering, 2023, 20(2): 2937-2963. doi: 10.3934/mbe.2023139 |
[8] | Suqi Ma . Low viral persistence of an immunological model. Mathematical Biosciences and Engineering, 2012, 9(4): 809-817. doi: 10.3934/mbe.2012.9.809 |
[9] | Abdessamad Tridane, Yang Kuang . Modeling the interaction of cytotoxic T lymphocytes and influenza virus infected epithelial cells. Mathematical Biosciences and Engineering, 2010, 7(1): 171-185. doi: 10.3934/mbe.2010.7.171 |
[10] | A. M. Elaiw, Raghad S. Alsulami, A. D. Hobiny . Global dynamics of IAV/SARS-CoV-2 coinfection model with eclipse phase and antibody immunity. Mathematical Biosciences and Engineering, 2023, 20(2): 3873-3917. doi: 10.3934/mbe.2023182 |
In 3D reconstruction tasks, camera parameter matrix estimation is usually used to present the single view of an object, which is not necessary when mapping the 3D point to 2D image. The single view reconstruction task should care more about the quality of reconstruction instead of the alignment. So in this paper, we propose an implicit field knowledge distillation model (IFKD) to reconstruct 3D objects from the single view. Transformations are performed on 3D points instead of the camera and keep the camera coordinate identified with the world coordinate, so that the extrinsic matrix can be omitted. Besides, a knowledge distillation structure from 3D voxel to the feature vector is established to further refine the feature description of 3D objects. Thus, the details of a 3D model can be better captured by the proposed model. This paper adopts ShapeNet Core dataset to verify the effectiveness of the IFKD model. Experiments show that IFKD has strong advantages in IOU and other core indicators compared with the camera matrix estimation methods, which verifies the feasibility of the new proposed mapping method.
Middle East respiratory syndrome (MERS) is a viral respiratory disease caused by Middle East respiratory syndrome coronavirus (MERS-CoV). The intermediate host of MERS-CoV is probably the dromedary camel, a zoonotic virus [1]. Most MERS cases are acquired by human-to-human transmission. There is no vaccine or specific treatment available, and approximately 35% of patients with MERS-CoV infection have died [2]. There has been extensive works on infectious disease models and viral infection models associated with MERS that can help in disease control and provide strategies for potential drug treatments [3,4,5,6,7,8].
Dipeptidyl peptidase-4 (DPP4) plays an important role in viral infection [2]. Based on classic viral infection models developed in [9,10,11], a four-dimensional ordinary differential equation model is proposed and studied in [8]. The model in [8] describes the interaction mechanisms among uninfected cells, infected cells, DPP4 and MERS-CoV.
Recently, taking into account periodic factors such as diurnal temperature differences and periodic drug treatment, the model in [8] has been further extended a periodic case in [12], and then the existence of positive periodic solutions is studied by using the theorem in [13].
It is well-known that CTL immune responses play a very critical role in controlling viral load and the concentration of infected cells. Thus, many scholars have considered CTL immune responses in various viral infection models and have achieved many excellent research results [14,15,16,17,18]. CTL cells can kill virus-infected cells and are important for the control and clearance of MERS-CoV infections [19]. Inspired by the above research works, we consider the following periodic MERS-CoV infection model with CTL immune response:
{˙T(t)=λ(t)−β(t)D(t)v(t)T(t)−d(t)T(t),˙I(t)=β(t)D(t)v(t)T(t)−d1(t)I(t)−p(t)I(t)Z(t),˙v(t)=d1(t)M(t)I(t)−c(t)v(t),˙D(t)=λ1(t)−β1(t)β(t)D(t)v(t)T(t)−γ(t)D(t),˙Z(t)=q(t)I(t)Z(t)−b(t)Z(t). | (1.1) |
In model (1.1), T(t), I(t), v(t), D(t) and Z(t) represent the concentrations of uninfected cells, infected cells, free virus, DPP4 on the surface of uninfected cells and CTL cells at time t, respectively. CTL cells increase at a rate bilinear rate q(t)I(t)Z(t) by the viral antigen of the infected cells and decay at rate b(t)Z(t); infected cells are killed by the CTL immune response at rate p(t)I(t)Z(t). Except for p(t), q(t) and b(t), all the remaining parameters of model (1.1) have the same biological meanings as in [12].
Throughout the paper, it is assumed that the functions λ(t), β(t), d(t), d1(t), p(t), M(t), c(t), λ1(t), γ(t), q(t) and b(t) are positive, continuous and ω periodic (ω>0); the function β1(t) is non-negative, continuous and ω periodic.
From point of view in both biology and mathematics, it is one of the most significant topics to study the existence of periodic oscillations of a system (see, for example, [12,20,21,22,23,24,25,26] and the references therein).
In the next section, some sufficient criteria are given for the existence of positive periodic oscillations of model (1.1). It should be mentioned here that, in the proofs of the main results in the following section, a new technique is developed to obtain a lower bound of the state variable Z(t) characterizing CTL immune response in model (1.1).
For some function f(t) which is continuous and ω-periodic on R, let us define the following notations:
fU=maxt∈[0,ω]f(t),fl=mint∈[0,ω]f(t),ˆf=1ω∫ω0f(t)dt. |
Moreover, for convenience, let us give the following parameters:
R∗=ˆλβlexp{L3+L4}^d1exp{M2}(βlexp{L3+L4}+dU)>1,ω∗=ˆb2ˆλˆq,δ∗=^d12ˆp(R∗−1),M1=ln(λUdl),M2=ln(ˆbˆq+2ˆλω),M3=ln(^(d1M)ˆc)+M2+2ˆcω,M4=ln(λU1γl),M5=ln(ˆβexp{M1+M3+M4}ˆpexp{L2})+2ˆbω,L1=ln(λlβUexp{M3+M4}+dU),L2=ln(ˆbˆq−2ˆλω),L3=ln((^d1M)ˆc)+L2−2ˆcω,L4=ln(λl1(β1β)Uexp{M1+M3}+γU),L5=ln(δ∗)−2ˆbω. |
The following theorem is the main result of this paper.
Theorem 2.1. If R∗>1 and ω<ω∗, then model (1.1) has at least one positive ω-periodic solution.
Proof. Making the change of variables T(t)=exp{u1(t)}, I(t)=exp{u2(t)}, v(t)=exp{u3(t)}, D(t)=exp{u4(t)}, Z(t)=exp{u5(t)}, then model (1.1) can be rewritten as
{˙u1(t)=λ(t)exp{u1(t)}−β(t)exp{u3(t)+u4(t)}−d(t),˙u2(t)=β(t)exp{u1(t)+u3(t)+u4(t)}exp{u2(t)}−d1(t)−p(t)exp{u5(t)},˙u3(t)=d1(t)M(t)exp{u2(t)}exp{u3(t)}−c(t),˙u4(t)=λ1(t)exp{u4(t)}−β1(t)β(t)exp{u1(t)+u3(t)}−γ(t),˙u5(t)=q(t)exp{u2(t)}−b(t). | (2.1) |
Thus, we only need to consider model (2.1).
Let us set
X=Y={u=(u1(t),u2(t),u3(t),u4(t),u5(t))T∈C(R,R5)|u(t)=u(t+ω)} |
with the norm
||u||=maxt∈[0,ω]|u1(t)|+maxt∈[0,ω]|u2(t)|+maxt∈[0,ω]|u3(t)|+maxt∈[0,ω]|u4(t)|+maxt∈[0,ω]|u5(t)|. |
It can be shown that X and Y are Banach spaces. Define
Nu=[λ(t)exp{u1(t)}−β(t)exp{u3(t)+u4(t)}−d(t)β(t)exp{u1(t)+u3(t)+u4(t)}exp{u2(t)}−d1(t)−p(t)exp{u5(t)}d1(t)M(t)exp{u2(t)}exp{u3(t)}−c(t)λ1(t)exp{u4(t)}−β1(t)β(t)exp{u1(t)+u3(t)}−γ(t)q(t)exp{u2(t)}−b(t)]:=[N1(t)N2(t)N3(t)N4(t)N5(t)](u∈X), |
Lu=˙u(u∈DomL),Pu=1ω∫ω0u(t)dt(u∈X),Qu=1ω∫ω0u(t)dt(u∈Y), |
here DomL={u∈X,˙u∈X}. It easily has that KerL={u∈X|u∈R5} and ImL={u∈Y|∫ω0u(t)dt=0}. Further, it is clear that ImL is closed in Y and dimKerL=codimImL=5. Hence, L is a Fredholm mapping with index zero.
For μ∈(0,1), let us consider the equation Lu=μNu, i.e.,
{˙u1(t)=μ[λ(t)exp{u1(t)}−β(t)exp{u3(t)+u4(t)}−d(t)],˙u2(t)=μ[β(t)exp{u1(t)+u3(t)+u4(t)}exp{u2(t)}−d1(t)−p(t)exp{u5(t)}],˙u3(t)=μ[d1(t)M(t)exp{u2(t)}exp{u3(t)}−c(t)],˙u4(t)=μ[λ1(t)exp{u4(t)}−β1(t)β(t)exp{u1(t)+u3(t)}−γ(t)],˙u5(t)=μ[q(t)exp{u2(t)}−b(t)]. | (2.2) |
For any solution u=(u1(t),u2(t),u3(t),u4(t),u5(t))T∈X of (2.2), it has
{∫ω0[λ(t)exp{u1(t)}−β(t)exp{u3(t)+u4(t)}−d(t)]dt=0,∫ω0[β(t)exp{u1(t)+u3(t)+u4(t)}exp{u2(t)}−d1(t)−p(t)exp{u5(t)}]dt=0,∫ω0[d1(t)M(t)exp{u2(t)}exp{u3(t)}−c(t)]dt=0,∫ω0[λ1(t)exp{u4(t)}−β1(t)β(t)exp{u1(t)+u3(t)}−γ(t)]dt=0,∫ω0[q(t)exp{u2(t)}−b(t)]dt=0. | (2.3) |
From the first two equations in (2.2), it has
˙u1(t)exp{u1(t)}=μ[λ(t)−β(t)exp{u1(t)+u3(t)+u4(t)}−d(t)exp{u1(t)}], |
and
˙u2(t)exp{u2(t)}=μ[β(t)exp{u1(t)+u3(t)+u4(t)}−d1(t)exp{u2(t)}−p(t)exp{u2(t)+u5(t)}]. |
Hence, by integrating the above two equations on [0,ω], it has
∫ω0[λ(t)−β(t)exp{u1(t)+u3(t)+u4(t)}−d(t)exp{u1(t)}]dt=0 | (2.4) |
and
∫ω0[β(t)exp{u1(t)+u3(t)+u4(t)}−d1(t)exp{u2(t)}−p(t)exp{u2(t)+u5(t)}]dt=0. | (2.5) |
Note that I(t):=exp{u2(t)} satisfies
˙I(t)=˙u2(t)exp{u2(t)}=μ[β(t)exp{u1(t)+u3(t)+u4(t)}−d1(t)−p(t)exp{u2(t)+u5(t)}]. |
Then, from (2.4) and (2.5), it has
∫ω0|˙I(t)|dt≤μ∫ω0[β(t)exp{u1(t)+u3(t)+u4(t)}+d1(t)+p(t)exp{u2(t)+u5(t)}]dt≤2∫ω0β(t)exp{u1(t)+u3(t)+u4(t)}dt≤2ˆλω. | (2.6) |
From the third and the fifth equations of (2.2), it has
∫ω0|˙u3(t)|dt≤μ[∫ω0d1(t)M(t)exp{u2(t)}exp{u3(t)}dt+∫ω0c(t)dt]<2ˆcω,∫ω0|˙u5(t)|dt≤μ[∫ω0q(t)exp{u2(t)}dt+∫ω0b(t)dt]<2ˆbω. | (2.7) |
Note that u∈X, there exist ξi,ηi∈[0,ω] (i=1,2,3,4,5), such that
ui(ξi)=mint∈[0,ω]ui(t),ui(ηi)=maxt∈[0,ω]ui(t)(i=1,2,3,4,5). |
From (2.2), ˙u1(η1)=0 and ˙u4(η4)=0, it has
λ(η1)exp{u1(η1)}−β(η1)exp{u3(η1)+u4(η1)}−d(η1)=0,λ1(η4)exp{u4(η4)}−β1(η4)β(η4)exp{u1(η4)+u3(η4)}−γ(η4)=0, |
which imply that
u1(t)≤u1(η1)≤ln(λ(η1)d(η1))≤ln(λUdl)=M1,u4(t)≤u4(η4)≤ln(λ1(η4)γ(η4))≤ln(λU1γl)=M4. | (2.8) |
From the last equation of (2.3), it has
∫ω0q(t)exp{u2(ξ2)}dt≤ˆbω≤∫ω0q(t)exp{u2(η2)}dt, |
which implies that
I(ξ2)=exp{u2(ξ2)}≤ˆbˆq≤exp{u2(η2)}=I(η2). |
Then, from (2.6) and ω<ω∗, it has
I(t)≤I(ξ2)+∫ω0|˙I(t)|dt≤ˆbˆq+2ˆλω,I(t)≥I(η2)−∫ω0|˙I(t)|dt≥ˆbˆq−2ˆλω=2ˆλ(ω∗−ω)>0. |
Thus, it has
u2(t)≤ln(ˆbˆq+2ˆλω)=M2,u2(t)≥ln(ˆbˆq−2ˆλω)=L2. | (2.9) |
From the third equation of (2.3), it has
∫ω0d1(t)M(t)exp{M2}exp{u3(ξ3)}dt≥ˆcω≥∫ω0d1(t)M(t)exp{L2}exp{u3(η3)}dt, |
which implies that
u3(ξ3)≤ln(^(d1M)ˆc)+M2,u3(η3)≥ln(^(d1M)ˆc)+L2. |
Then, from (2.7), it has
u3(t)≤u3(ξ3)+∫ω0|˙u3(t)|dt≤ln(^(d1M)ˆc)+M2+2ˆcω=M3,u3(t)≥u3(η3)−∫ω0|˙u3(t)|dt≥ln(^(d1M)ˆc)+L2−2ˆcω=L3. | (2.10) |
From the second equation of (2.3), it has
ˆpexp{u5(ξ5)}ω≤∫ω0[β(t)exp{M1+M3+M4}exp{L2}−d1(t)]dt≤exp{M1+M3+M4}exp{L2}ˆβω, |
which implies that
u5(ξ5)≤ln(ˆβexp{M1+M3+M4}ˆpexp{L2}):=l5. |
Then, from (2.7), it has
u5(t)≤u5(ξ5)+∫ω0|˙u5(t)|dt≤l5+2ˆbω=M5. |
From ˙u1(ξ1)=0, ˙u4(ξ4)=0, (2.8) and (2.10), it has
exp{u1(ξ1)}=λ(ξ1)β(ξ1)exp{u3(ξ1)+u4(ξ1)}+d(ξ1)≥λlβUexp{M3+M4}+dU,exp{u4(ξ4)}=λ1(ξ4)β1(ξ4)β(ξ4)exp{u1(ξ4)+u3(ξ4)}+γ(ξ4)≥λl1(β1β)Uexp{M1+M3}+γU. |
Thus, it has
u1(t)≥u1(ξ1)≥ln(λlβUexp{M3+M4}+dU)=L1,u4(t)≥u4(ξ4)=ln(λl1(β1β)Uexp{M1+M3}+γU)=L4. | (2.11) |
Let us give an estimate of the lower bound of the state variable u5(t) related to CTL immune response. It should be mentioned here that a completely different method from that in [12] has been used.
Claim A If R∗>1 and ω<ω∗, then
exp{u5(η5)}≥δ∗. |
If Claim A is not true, then it has that, for any t, exp{u5(t)}≤exp{u5(η5)}<δ∗. Hence, it has from (2.3), (2.9)–(2.11) that
0=∫ω0[β(t)exp{u1(t)+u3(t)+u4(t)}exp{u2(t)}−d1(t)−p(t)exp{u5(t)}]dt≥∫ω0[β(t)exp{u1(t)+L3+L4}exp{M2}−d1(t)−p(t)exp{u5(η5)}]dt≥βlexp{L3+L4}exp{M2}∫ω0exp{u1(t)}dt−(ˆd1+ˆpδ∗)ω, |
which implies that
∫ω0d(t)exp{u1(t)}dt≤dU∫ω0exp{u1(t)}dt≤dU(ˆd1+ˆpδ∗)exp{M2}βlexp{L3+L4}ω:=Ψω. | (2.12) |
Adding (2.4) and (2.5) together, it has
∫ω0[λ(t)−d(t)exp{u1(t)}]dt=∫ω0[d1(t)exp{u2(t)}+p(t)exp{u2(t)+u5(t)}]dt≤∫ω0exp{M2}[d1(t)+p(t)exp{u5(η5)}]dt≤exp{M2}(^d1+ˆpδ∗)ω, |
which implies that
∫ω0d(t)exp{u1(t)}dt≥[ˆλ−exp{M2}(^d1+ˆpδ∗)]ω=Ψω+[ˆλ−Ψ−exp{M2}(^d1+ˆpδ∗)]ω=Ψω+[ˆλ−exp{M2}(1+dUβlexp{L3+L4})(^d1+ˆpδ∗)]ω=Ψω+^d1exp{M2}(1+dUβlexp{L3+L4})(R∗−1−ˆp^d1δ∗)ω=Ψω+^d12exp{M2}(1+dUβlexp{L3+L4})(R∗−1)ω>Ψω, |
which is a contradiction to (2.12). Thus, the claim holds.
From Claim A and (2.7), it has
u5(t)≥u5(η5)−∫ω0|˙u5(t)|dt≥ln(δ∗)−2ˆbω=L5. | (2.13) |
Now, for convenience, let us define
¯R∗=(ˆλ−^d1ˆbˆq)ˆβ^(d1M)ˆd^d1ˆc^λ1^(β1β)ˆλ^(d1M)ˆbˆdˆcˆq+ˆγ,Zmax=ˆqˆpˆb(ˆλ−^d1ˆbˆq). |
Note that if R∗>1, then it has
~R∗:=(ˆλ−^d1exp{M2})βlexp{L3+L4}dU^d1exp{M2}>1, |
which implies that
Zmax>ˆqˆpˆb[ˆλ−^d1(ˆbˆq+2ˆλω)]=ˆqˆpˆb(ˆλ−^d1exp{M2})>0,¯R∗≥(ˆλ−^d1ˆbˆq)ˆβ^(d1M)ˆd^d1ˆcλl1(β1β)Uexp{M3+M1}+γU≥~R∗>1. |
Let (u1,u2,u3,u4,u5)T∈R5 be the solution of the following equations:
{ˆλexp{u1}−ˆβexp{u3+u4}−ˆd=0,ˆβexp{u1+u3+u4}exp{u2}−ˆd1−ˆpexp{u5}=0,^(d1M)exp{u2}exp{u3}−ˆc=0,ˆλ1exp{u4}−^(β1β)exp{u1+u3}−ˆγ=0,ˆqexp{u2}−ˆb=0.. | (2.14) |
Define Γ:[0,Zmax]→R, via
Γ(x)=ˆβ^(d1M)ˆc^λ1Γ1(x)^(β1β)Γ1(x)^(d1M)ˆbˆqˆc+ˆγ−ˆd1−ˆpx, |
where
Γ1(x)=ˆλˆd−^d1ˆbˆdˆq−ˆpˆbˆdˆqx=ˆpˆbˆdˆq(Zmax−x). |
Equation (2.14) can be rewritten as
exp{u2}=ˆbˆq,exp{u3}=^(d1M)ˆcexp{u2}=^(d1M)ˆbˆqˆc,exp{u1}=ˆλˆd−^d1exp{u2}ˆd−ˆpexp{u2}ˆdexp{u5}=Γ1(exp{u5}),exp{u4}=^λ1^(β1β)exp{u1+u3}+ˆγ=^λ1^(β1β)Γ1(exp{u5})^(d1M)ˆbˆqˆc+ˆγ,ˆβ^(d1M)ˆcexp{u1+u4}−ˆd1−ˆpexp{u5}=Γ(exp{u5})=0. |
It is obvious that if there is a solution (u1,u2,u3,u4,u5)T∈R5 for (2.14), it must have 0<exp{u5}<Zmax. In addition, note that Γ(x) is monotonically decreasing with respect to x on [0,Zmax]. It has from Γ(Zmax)=−^d1−ˆpZmax<0 and
Γ(0)=ˆβ^(d1M)ˆc^λ1Γ1(0)^(β1β)Γ1(0)^(d1M)ˆbˆqˆc+ˆγ−^d1>ˆβ^(d1M)ˆc^λ1(ˆλˆd−^d1ˆbˆdˆq)^(β1β)ˆλˆd^(d1M)ˆbˆqˆc+ˆγ−^d1=^d1(¯R∗−1)>0 |
that there exists a unique positive constant x=Z∗∈(0,Zmax) such that Γ(Z∗)=0.
The above discussions show that, if R∗>1, (2.14) has a unique solution (u∗1,u∗2,u∗3,u∗4,u∗5)T, here u∗i=ln(ei) (i=1,2,3,4,5),
e1=Γ1(Z∗)>0,e2=ˆbˆq>0,e3=^(d1M)ˆbˆqˆc>0,e4=^λ1^(β1β)Γ1(Z∗)^(d1M)ˆbˆqˆc+ˆγ>0,e5=Z∗>0. |
Let us define the following set
Ω={u∈X|||u||<U1=1+5∑i=1(max{|Mi|,|Li|}+|u∗i|)}⊂X. |
Moreover, by similar arguments as in [12], it has that N is L-compact on ¯Ω.
Now, let us compute the Leray-Schauder degree deg{QN,∂Ω∩KerL,(0,0,0,0,0)T}:=Δ as follows,
Δ=sign|−ˆλe10−ˆβe3e4−ˆβe3e40ˆβe1e3e4e2−ˆβe1e3e4e2ˆβe1e3e4e2ˆβe1e3e4e2−ˆpe50^(d1M)e2e3−^(d1M)e2e300−^(β1β)e1e30−^(β1β)e1e3−^λ1e400ˆqe2000|=sign{−^(d1M)ˆpˆqe22e5e3(ˆλ^λ1e1e4−ˆβ^(β1β)e1e23e4)}=sign{−^(d1M)ˆpˆqe22e5e3e1e4[ˆde1(^(β1β)e1e3e4+ˆγe4)+ˆβˆγe1e3e24]}=−1≠0, |
where ˆλ=ˆβe1e3e4+ˆde1 and ^λ1=^(β1β)e1e3e4+ˆγe4 are used.
Finally, it has those all the conditions of the continuation theorem in [13] (also see, for example, Lemma 2.1 in [12]) are satisfied. This proves that, if ω<ω∗ and R∗>1, model (2.1) has at least one ω-periodic solution.
Let us consider the following classical viral infection dynamic model [9] with CTL immune response:
{˙T(t)=λ(t)−β(t)v(t)T(t)−d(t)T(t),˙I(t)=β(t)v(t)T(t)−d1(t)I(t)−p(t)I(t)Z(t),˙v(t)=d1(t)M(t)I(t)−c(t)v(t),˙Z(t)=q(t)I(t)Z(t)−b(t)Z(t),(A) |
where, all the coefficients are the same with that in model (1.1).
Define R1:[0,ω∗]→R, via
R1(x)=ˆλβl[^(d1M)ˆcexp{−2ˆcx}(ˆbˆq−2ˆλx)]^d1(ˆbˆq+2ˆλx){dU+βl[^(d1M)ˆcexp{−2ˆcx}(ˆbˆq−2ˆλx)]}. |
Obviously, R1(x) is monotonically decreasing on [0,ω∗] and
R1(0)=ˆλβl^(d1M)ˆq^d1(dUˆcˆq+βl^(d1M)ˆb),R1(ω∗)=0. |
Therefore, if R1(0)>1, then there exists a unique constant ω∗∗∈(0,ω∗) such that R1(ω∗∗)=1, R1(x)>1 for 0≤x<ω∗∗ and R1(x)<1 for ω∗∗<x≤ω∗.
For model (A), it is not difficult to derive the following result.
Theorem 2.2. If R1(ω)>1 and ω<ω∗ (i.e. R1(0)>1 and ω<ω∗∗<ω∗), then model (A) has at least one positive ω-periodic solution.
Remark 2.1. If all the coefficients in model (A) take constants values, i.e., λ(t)≡λ>0, β(t)≡β>0, d(t)≡d>0, d1(t)≡d1>0, p(t)≡p>0, M(t)≡M>0, c(t)≡c>0, q(t)≡q>0 and b(t)≡b>0, then model (A) becomes the classical model which is first proposed by Nowak and Bangham in [9]. the condition ω<ω∗∗ in Theorem 2.2 is naturally satisfied. Furthermore, it has R1(0)=(λβMq)/(dcq+βd1Mb):=R1. From \emph{[9]}, it has that the condition R1>1 implies the existence of a unique positive equilibrium. This shows that the conditions and conclusion in Theorem 2.2 are reasonable.
In summary, Theorem 2.1 in the paper successfully extends the main result in [12]) to a MERS-CoV viral infection model with CTL immune response. In the proof of Theorem 2.1, we use a very different method from that in [9] to obtain the lower bound (ln(δ∗)−2ˆbω) of the state variable u5(t). Furthermore, as a special case, Theorem 2.2 gives sufficient conditions for the existence of positive periodic solution of model (A). Model (A) is a natural extension of the classical model in [9]. As the end of the paper, let us give a example to summarize the applications of Theorem 2.1. Let us choose the coefficients in model (1.1) as follows (for the values of some parameters, please refer to [7,27] for the case of some autonomous models), λ(t)=45(1+0.1sin(4πt)), β(t)=1.4×10−8(1+0.1cos(4πt)), d(t)=0.001(1+0.5cos(4πt)), d1(t)=0.056(1+0.5cos(4πt)), p(t)=0.00092(1+0.5cos(4πt)), M(t)=100000, c(t)=2.1(1+0.3cos(4πt)), λ1(t)=10(1+0.1sin(4πt)), β1(t)=0.001, γ(t)=0.01(1+0.1cos(4πt)), q(t)=0.005(1+0.5sin(4πt)), b(t)=0.5(1+0.4cos(4πt)). Then, with the help of Maple mathematical software, it has ω=0.5<ω∗≈1.111111, M1≈11.502875, M2≈4.976734, M3≈14.965318, M4≈7.108426, M5≈18.976215, L1≈−0.383569, L2=4.007333, L3≈9.795917, L4≈0.623402, L5≈1.387881, R∗≈1.2170332>1. From Theorem 2.1, it follows that model (1.1) has at least one positive ω (ω=0.5)-periodic solution. Figure 1 gives the corresponding numerical simulation, and the initial value is chosen as (T(0),I(0),v(0),D(0),Z(0))T=(12.5,100,265000,995.4,423)T.
This paper is supported by National Natural Science Foundation of China (No.11971055) and Beijing Natural Science Foundation (No.1202019).
The authors declare there is no conflict of interest.
[1] | M. Li, H. Zhang, D2im-net: Learning detail disentangled implicit fields from single images, in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, (2021), 10246–10255. https://doi.org/10.1109/CVPR46437.2021.01011 |
[2] | S. Saito, Z. Huang, R. Natsume, S. Morishima, A. Kanazawa, H. Li, Pifu: Pixel-aligned implicit function for high-resolution clothed human digitization, in Proceedings of the IEEE/CVF International Conference on Computer Vision, (2019), 2304–2314. https://doi.org/10.1109/ICCV.2019.00239 |
[3] | T. Groueix, M. Fisher, V. G. Kim, B. C. Russell, M. Aubry, A papier-mâché approach to learning 3d surface generation, in Proceedings of the IEEE conference on computer vision and pattern recognition, (2018), 216–224. https://doi.org/10.1109/CVPR.2018.00030 |
[4] |
C. B. Choy, D. Xu, J. Gwak, K. Chen, S. Savarese, 3d-r2n2: A unified approach for single and multi-view 3d object reconstruction, Eur. Conf. Comput. Vision, 9912 (2016), 628–644. https://doi.org/10.1007/978-3-319-46484-8_38 doi: 10.1007/978-3-319-46484-8_38
![]() |
[5] | G. Yang, X. Huang, Z. Hao, M. Liu, S. Belongie, B. Hariharan, Pointflow: 3d point cloud generation with continuous normalizing flows, in Proceedings of the IEEE/CVF International Conference on Computer Vision, (2019), 4541–4550. https://doi.org/10.1109/ICCV.2019.00464 |
[6] | G. Chen, W. Choi, X. Yu, T. Han, M. Chandraker, Learning efficient object detection models with knowledge distillation, Adv. Neural Inform. Proc. Syst., (2017), 30. |
[7] | J. J. Park, P. Florence, J. Straub, R. Newcombe, S. Lovegrove, Deepsdf: Learning continuous signed distance functions for shape representation, in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, (2019), 165–174. https://doi.org/10.1109/CVPR.2019.00025 |
[8] | G. Hinton, O. Vinyals, J. Dean, Distilling the knowledge in a neural network, preprint, arXiv: 1503.02531. |
[9] | N. Wang, Y. Zhang, Z. Li, Y. Fu, W. Liu, Y. Jiang, Pixel2mesh: Generating 3d mesh models from single rgb images. in Proceedings of the European conference on computer vision (ECCV), (2018), 52–67. https://doi.org/10.1007/978-3-030-01252-6_4 |
[10] | H. Xie, H. Yao, X. Sun, S. Zhou, S. Zhang, Pix2vox: Context-aware 3d reconstruction from single and multi-view images, preprint, arXiv: 1901.11153. |
[11] |
H. Xie, H. Yao, S. Zhang, S. Zhou, W. Sun, Pix2vox++: Multi-scale context-aware 3d object reconstruction from single and multiple images, Int. J. Comput. Vision, 128 (2020), 2919–2935. https://doi.org/10.1007/s11263-020-01347-6 doi: 10.1007/s11263-020-01347-6
![]() |
[12] | L. P. Tchapmi, V. Kosaraju, H. Rezatofighi, I. Reid, S. Savarese, Topnet: Structural point cloud decoder, in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, (2019), 383–392. https://doi.org/10.1109/CVPR.2019.00047 |
[13] | L. Mescheder, M. Oechsle, M. Niemeyer, S. Nowozin, A. Geiger, Occupancy networks: Learning 3d reconstruction in function space. in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, 4460–4470. https://doi.org/10.1109/CVPR.2019.00459 |
[14] | Z. Chen, A. Tagliasacchi, H. Zhang, Bsp-net: Generating compact meshes via binary space partitioning, preprint, arXiv: 1911.06971. |
[15] | Q. Xu, W. Wang, D. Ceylan, R. Mech, U. Neumann, Disn: Deep implicit surface network for high-quality single-view 3d reconstruction, Adv. Neural Inform. Process. Syst., (2019), 32. |
[16] | J. Bechtold, M. Tatarchenko, V. Fischer, T. Brox, Fostering generalization in single-view 3d reconstruction by learning a hierarchy of local and global shape priors, in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, (2021), 15880–15889. https://doi.org/10.1109/CVPR46437.2021.01562 |
[17] |
Y. Duan, H. Zhu, H. Wang, L. Yi, R. Nevatia, L. J. Guibas, Curriculum deepsdf, Eur. Conf. Comput. Vision, (2020), 51–67. https://doi.org/10.1007/978-3-030-58598-3_4 doi: 10.1007/978-3-030-58598-3_4
![]() |
[18] | A. Romero, N. Ballas, S. E. Kahou, A. Chassang, C. Gatta, Y. Bengio, Fitnets: Hints for thin deep nets, preprint, arXiv: 1412.6550. |
[19] | Q. Li, S. Jin, J. Yan, Mimicking very efficient network for object detection, in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), (2017), 7341–7349. https://doi.org/10.1109/CVPR.2017.776 |
[20] | T. Wang, L. Yuan, X. Zhang, J. Feng, Distilling object detectors with fine-grained feature imitation, preprint, arXiv: 1906.03609. |
[21] |
W. E. Lorensen, H. E. Cline, Marching cubes: A high resolution 3d surface construction algorithm, ACM SIGGRAPH Comput. Graphics, 21 (1987), 163–169. https://doi.org/10.1145/37402.37422 doi: 10.1145/37402.37422
![]() |
[22] | O. Ronneberger, P. Fischer, T. Brox, U-net: Convolutional networks for biomedical image segmentation, Medical Image Computing and Computer-Assisted Intervention, Springer, 9351 (2015), 234–241. https://doi.org/10.1007/978-3-319-24574-4_28 |
[23] | T. Hastie, R. Tibshirani, J. Friedman, The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Springer, 2009. |
[24] | Z. Chen, H. Zhang, Learning implicit fields for generative shape modeling, in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, (2019), 5932-5941. https://doi.org/10.1109/CVPR.2019.00609 |
[25] | C. Häne, S. Tulsiani, J. Malik, Hierarchical surface prediction for 3d object reconstruction, preprint, arXiv: 1704.00710. |
[26] | B. Xu, N. Wang, T. Chen, M. Li, Empirical evaluation of rectified activations in convolutional network, preprint, arXiv: 1505.00853. |
[27] | J. Chibane, T. Alldieck, G. Pons-Moll, Implicit functions in feature space for 3d shape reconstruction and completion, in 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition, (2020), 6968–6979. https://doi.org/10.1109/CVPR42600.2020.00700 |
[28] |
J. Jin, A. G. Patil, Z. Xiong, H. Zhang, Dr-kfs: A differentiable visual similarity metric for 3d shape reconstruction, Eur. Conf. Comput. Vision, (2020), 295–311. https://doi.org/10.1007/978-3-030-58589-1_18 doi: 10.1007/978-3-030-58589-1_18
![]() |
[29] |
B. Yang, S. Wang, A. Markham, N. Trigoni, Robust attentional aggregation of deep feature sets for multi-view 3d reconstruction, Int. J. Comput. Vision, 128 (2020), 53–73. https://doi.org/10.1007/s11263-019-01217-w doi: 10.1007/s11263-019-01217-w
![]() |
[30] | M. Jaderberg, K. Simonyan, A. Zisserman, K. Kavukcuoglu, Spatial transformer networks, in Proceedings of the 28th International Conference on Neural Information Processing Systems, 2 (2015), 2017–2025. https://doi.org/10.5555/2969442.2969465 |
1. | Tuersunjiang Keyoumu, Wanbiao Ma, Ke Guo, Global Stability of a MERS-CoV Infection Model with CTL Immune Response and Intracellular Delay, 2023, 11, 2227-7390, 1066, 10.3390/math11041066 | |
2. | Tongqian Zhang, Xinna Xu, Xinzeng Wang, Dynamic analysis of a cytokine-enhanced viral infection model with time delays and CTL immune response, 2023, 170, 09600779, 113357, 10.1016/j.chaos.2023.113357 | |
3. | Jinlong Lv, Wanbiao Ma, Delay induced stability switch in a mathematical model of CD8 T-cell response to SARS-CoV-2 mediated by receptor ACE2, 2024, 34, 1054-1500, 10.1063/5.0187872 | |
4. | Ahmed M. Elaiw, Amani S. Alsulami, Aatef D. Hobiny, Global properties of delayed models for SARS-CoV-2 infection mediated by ACE2 receptor with humoral immunity, 2024, 9, 2473-6988, 1046, 10.3934/math.2024052 | |
5. | Mengfan Tan, Guijie Lan, Chunjin Wei, Dynamic analysis of HIV infection model with CTL immune response and cell-to-cell transmission, 2024, 156, 08939659, 109140, 10.1016/j.aml.2024.109140 |