
The standard treatments of surgery, radiation, and chemotherapy in head and neck squamous cell carcinomas (HNSCC) causes disturbance to normal surrounding tissues, systemic toxicities and functional problems with eating, speaking, and breathing. With early detection, many of these cancers can be effectively treated, but treatment should also focus on retaining the function of the proximal nerves, tissues and vasculature surrounding the tumor. With current research focused on understanding pathogenesis of these cancers in a molecular level, targeted therapy using monoclonal antibodies (MoAbs), can be modified and directed towards tumor genes, proteins and signal pathways with the potential to reduce unfavorable side effects of current treatments. This review will highlight the current MoAb therapies used in HNSCC, and discuss ongoing research efforts to develop novel treatment agents with potential to improve efficacy, increase overall survival (OS) rates and reduce toxicities.
Citation: Vivek Radhakrishnan, Mark S. Swanson, Uttam K. Sinha. Monoclonal Antibodies as Treatment Modalities in Head and Neck Cancers[J]. AIMS Medical Science, 2015, 2(4): 347-359. doi: 10.3934/medsci.2015.4.347
[1] | Raimund Bürger, Christophe Chalons, Rafael Ordoñez, Luis Miguel Villada . A multiclass Lighthill-Whitham-Richards traffic model with a discontinuous velocity function. Networks and Heterogeneous Media, 2021, 16(2): 187-219. doi: 10.3934/nhm.2021004 |
[2] | Maya Briani, Emiliano Cristiani . An easy-to-use algorithm for simulating traffic flow on networks: Theoretical study. Networks and Heterogeneous Media, 2014, 9(3): 519-552. doi: 10.3934/nhm.2014.9.519 |
[3] | Mauro Garavello, Roberto Natalini, Benedetto Piccoli, Andrea Terracina . Conservation laws with discontinuous flux. Networks and Heterogeneous Media, 2007, 2(1): 159-179. doi: 10.3934/nhm.2007.2.159 |
[4] | Raimund Bürger, Kenneth H. Karlsen, John D. Towers . On some difference schemes and entropy conditions for a class of multi-species kinematic flow models with discontinuous flux. Networks and Heterogeneous Media, 2010, 5(3): 461-485. doi: 10.3934/nhm.2010.5.461 |
[5] | Helge Holden, Nils Henrik Risebro . Follow-the-Leader models can be viewed as a numerical approximation to the Lighthill-Whitham-Richards model for traffic flow. Networks and Heterogeneous Media, 2018, 13(3): 409-421. doi: 10.3934/nhm.2018018 |
[6] | Adriano Festa, Simone Göttlich, Marion Pfirsching . A model for a network of conveyor belts with discontinuous speed and capacity. Networks and Heterogeneous Media, 2019, 14(2): 389-410. doi: 10.3934/nhm.2019016 |
[7] | Christophe Chalons, Paola Goatin, Nicolas Seguin . General constrained conservation laws. Application to pedestrian flow modeling. Networks and Heterogeneous Media, 2013, 8(2): 433-463. doi: 10.3934/nhm.2013.8.433 |
[8] | Raimund Bürger, Stefan Diehl, M. Carmen Martí, Yolanda Vásquez . A difference scheme for a triangular system of conservation laws with discontinuous flux modeling three-phase flows. Networks and Heterogeneous Media, 2023, 18(1): 140-190. doi: 10.3934/nhm.2023006 |
[9] | Giuseppe Maria Coclite, Lorenzo di Ruvo, Jan Ernest, Siddhartha Mishra . Convergence of vanishing capillarity approximations for scalar conservation laws with discontinuous fluxes. Networks and Heterogeneous Media, 2013, 8(4): 969-984. doi: 10.3934/nhm.2013.8.969 |
[10] | Wen Shen . Traveling wave profiles for a Follow-the-Leader model for traffic flow with rough road condition. Networks and Heterogeneous Media, 2018, 13(3): 449-478. doi: 10.3934/nhm.2018020 |
The standard treatments of surgery, radiation, and chemotherapy in head and neck squamous cell carcinomas (HNSCC) causes disturbance to normal surrounding tissues, systemic toxicities and functional problems with eating, speaking, and breathing. With early detection, many of these cancers can be effectively treated, but treatment should also focus on retaining the function of the proximal nerves, tissues and vasculature surrounding the tumor. With current research focused on understanding pathogenesis of these cancers in a molecular level, targeted therapy using monoclonal antibodies (MoAbs), can be modified and directed towards tumor genes, proteins and signal pathways with the potential to reduce unfavorable side effects of current treatments. This review will highlight the current MoAb therapies used in HNSCC, and discuss ongoing research efforts to develop novel treatment agents with potential to improve efficacy, increase overall survival (OS) rates and reduce toxicities.
Traffic flow models based on scalar conservation laws with continuous flux functions are widely used in the literature. For a general presentation of the models, we refer to the books [11,12,23] and the references therein. Extensions to road traffic networks have been also established. We mention in particular the contributions [6,15], where the authors introduce the coupled network problem and show the existence of solutions. Within this article, we are concerned with the special case of scalar conservation laws with discontinuous flux in the unknown that are motivated in the traffic flow theory by the observation of a gap between the free flow and the congested flow regime [4,5,8]. This phenomenon generates an interesting dynamical behavior called zero waves, i.e., waves with infinite (negative) speed but zero wave strength, and has been investigated in recent years either from a theoretical or numerical point of view, see for instance [2,19,20,22,24] or more generally [1,3,7,13].
To the best of our knowledge, the study of scalar conservation laws with discontinuous flux functions on networks is still missing in the traffic flow literature. However, in the context of supply chains with discontinuous flux such considerations have been already done [10,14]. We remark that supply chain models differ essentially from traffic flow models due to simpler dynamics and different coupling conditions.
In this work, we aim to derive a traffic network model, where the dynamics on each road are governed by a scalar conservation law with discontinuous flux function in the unknown. For simplicity, we restrict to piecewise linear flux functions. Special emphasis is put on the coupling at junction points to ensure a unique admissible weak solution. In particular, we focus on dispersing junctions where the number of incoming roads does not exceed the number of outgoing roads and merging with two incoming and one outgoing road. The latter type of junction can be extended to the case of multiple incoming roads and a single outgoing one. In order to construct a suitable numerical scheme that is not based on regularization techniques we adapt the splitting algorithm originally introduced in [22]. Therein, the discontinuous flux is decomposed into a Lipschitz continuous flux and a Heaviside flux such that a two-point monotone flux scheme, e.g., Godunov, can be employed in an appropriate manner. This algorithm has been studied in [22] for the case of a single road only. However, in the network case, multiple roads with possibly disjunctive flux functions need to be considered at a junction point to ensure mass conservation. Hence, the key challenge is to determine the correct flux through the junction in an appropriate manner. Therefore, a detailed case distinction in accordance with the theoretical investigations is provided for the different types of junctions. The numerical results validate the proposed algorithm for some relevant network problems.
The paper is organized as follows: in Section 2 we discuss the basic model and Riemann problems which permit to derive an exact solution. We extend the modeling framework to networks in Section 3 and focus on the coupling conditions. In Section 4, we introduce how the splitting algorithm [22] can be extended to also deal with the different types of junctions. Finally, we present a suitable discretization and numerical simulations in Section 5.
In this section, we briefly recall the case of the Lighthill-Whitham-Richards (LWR) model [18,21] on a single edge with a flux function having a single decreasing jump at
Following [22], we consider the scalar conservation law Eq (2.1),
$ {ut+f(u)x=0,(x,t)∈(a,b)×(0,T)=:ΠT,u(x,0)=u0(x)∈[0,umax],x∈(a,b),u(a,t)=r(t)∈[0,umax],t∈(0,T),u(b,t)=s(t)∈[0,umax],F(t)∈˜f(s(t)),t∈(0,T). $
|
(2.1) |
More precisely the flux function is defined as follows Eq (2.2),
$ f(u)={f1(u) ˆ=Flux1,ifu∈[0,u∗],f2(u) ˆ=Flux2,ifu∈(u∗,umax]. $
|
(2.2) |
We denote
$ α:=f(u∗−)−f(u∗+). $
|
(2.3) |
As usual we require for the flux function
As in [22], the multivalued version of
$ ˜f(u)={f(u),u∈[0,u∗),[f(u∗+),f(u∗−)],u=u∗,f(u),u∈(u∗,umax]. $
|
(2.4) |
Finally, we have to discuss the imposed boundary conditions at
$ F(t)={f(u∗−),ifthetrafficaheadofx=bisfree−flowing,f(u∗+),ifthetrafficaheadofx=biscongested. $
|
(2.5) |
The state of traffic ahead of
Remark 2.1. [22,Remark 1.3] We note that for the boundary condition at the left end the state of traffic ahead of
The following assumptions are important for the proof of existence and uniqueness of solutions.
Assumption 2.2. [22,Assumption 1.1] The initial data satisfies
A weak solution is intended in the following sense:
Definition 2.3 [22,Definition 1.1] A function
$ v(x,t) \in \tilde f(u(x,t)) \rm{ a.e.} $ |
such that for each
$ \int_0^T \int_a^b (u\psi_t + v\psi_x){\rm{d}}x{\rm{d}}t + \int_a^bu_0(x)\psi(x,0){\rm{d}}x = 0. $ |
As usual, weak solutions do not lead to a unique solution and additional criteria are necessary to rule out physically incorrect solutions. In particular, the discontinuity of the flux prohibits from directly using the classical approaches. Note that in [22] an adapted version of Oleinik's entropy condition [9] is used to single out the correct solution, while in [24] the convex hull construction [17] is used to construct solutions to Riemann problems.
Here, we will concentrate on the convex hull construction. For completeness we will shortly recall the solutions to Riemann problems considered in [24] as they are essential in order to construct a Riemann solver at a junction.
We consider a Riemann problem with initial data
We consider the following flux function by Eq (2, 6),
$ f(u)={d1u+d0,x≤u∗(Flux1),e1u+e0,x>u∗(Flux2) $
|
(2.6) |
with the regularized flux function given by Eq (2.7),
$ fϵ(u)={f1(u)=d1u+d0,0≤u≤u∗,fϵmid(u)=−1ϵ(f1(u∗)−fϵ2(u∗+ϵ))(u−u∗)+f1(u∗),u∗<u<u∗+ϵ,fϵ2(u)=e1u+e0,u∗+ϵ<u≤umax. $
|
(2.7) |
We define
Case 1: Either
This case corresponds to the classical case of solving Riemann problems, where the solution consists of a single rarefaction wave or shock, see [17].
Case 2:
By using the smallest convex hull approach the solution consists of a contact line following
$ s=f(uL)−f(u∗)uL−u∗<0, $
|
where we recall that
$ u(x,t)={uL,ifx<st,u∗,ifst≤x≤d1t,uR,ifx>d1t. $
|
(2.8) |
Case 3:
Here, the solution is given by a shock connecting
$ s = \frac{f(u^*+)-f(u_L)}{u^*-u_L} < 0. $ |
Note that due to
$ u(x,t)={uL,ifx<st,u∗,ifst≤x≤e1t,uR,ifx>e1t. $
|
(2.9) |
Case 4:
In this case, we get only one shock connecting
$ u(x,t)={uL,ifx<st,uR,ifx≥st. $
|
(2.10) |
Remark 2.4. As aforementioned in [19] Riemann solutions for piecewise quadratic discontinuous flux functions are derived. They also cover the case of a piecewise linear flux function if the quadratic terms are zero. For a general quadratic discontinuous flux, the solutions are more involved since no contact discontinuities occur.
Up to now, we have not addressed the case, where one of the boundary conditions equals the critical density
Next, we focus on networks where we allow for discontinuous flux functions. The key idea is to consider the regularized flux function
We start with a short introduction to the network setting. For more details on traffic flow network models we refer the reader to [11] and the references therein.
Let
We call the couple
In order to derive the network solution, we restrict to the description of a single junction
$ A = \left( {β1,1⋯β1,m⋮⋮⋮βn,1⋯βn,m } \right). $
|
To conserve the mass we assume
Definition 3.1 (Weak solution at a junction). Let
$ n∑i=1(∫T0∫Iini(uini(x,t)∂tϕini(x,t)+wini(x,t)∂xϕini(x,t))dxdt)+m∑j=1(∫T0∫Ioutj(uoutj(x,t)∂tϕoutj(x,t)+woutj(x,t)∂xϕoutj(x,t))dxdt)=0, $
|
for every collection of test functions
$ \phi_i^{{\rm{in}}}(b_i^{{\rm{in}}}, \cdot) = \phi_j^{{\rm{out}}}(a_j^{{\rm{out}}}, \cdot), \quad \partial_x \phi_i^{{\rm{in}}}(b_i^{{\rm{in}}}, \cdot) = \partial_x \phi_j^{{\rm{out}}}(a_j^{{\rm{out}}}, \cdot), $ |
for
Additionally, in order to get unique solutions, we will consider the following concept of admissible solutions, which adapts ref. [11] (rule (A) and (B), p. 81) to the discontinuous setting:
Definition 3.2 (Admissbile Weak Solution). We call
1.
2.
3.
4.
In particular, the maximization of the inflow with respect to the distributions parameters and the technical assumption [11] guarantee the uniqueness of solutions for a continuous flux.
If
For solving the maximization problems imposed by the definition 3.2 so-called supply and demand functions can be used, see [16]. The demand describes the maximal flux the incoming road wants to send. In contrast, the supply describes the maximal flux the outgoing road is able to absorb. The definition of the supply and demand functions of the regularized function is straightforward. As
Definition 3.3. For a network with flux function
$ D(u)={f(u),u∈[0,u∗),f(u∗−),u∈[u∗,umax]. $
|
(3.1) |
On the contrary, the supply reads as
$ S(u)={f(u∗−),u∈[0,u∗),f(u),u∈(u∗,umax] $
|
(3.2) |
and
$ S(u∗)={f(u∗−),freeflowing,f(u∗+),congested. $
|
(3.3) |
Remark 3.4. If we consider the regularized flux function
In order to show existence and uniqueness in the discontinuous case we need to define an additional function. For a regularized flux function we notice that for every flux value, we get two different density values, see left picture in Figure 4. As different density values lead to different solutions, we need to be able to distinguish them and choose the correct solution. In the continuous or regularized case a mapping usually called
Definition 3.5. Let the function
1.
2. For
Note that if
Remark 3.6. We note that the mapping
Now, we present a Riemann solver for two types of junctions. First, we consider a junction with
Theorem 3.7. Let
$ uini∈{{uini,0}∪(η(uini,0),umax],ifuini,0∈[0,f−1+],{uini,0}∪[u∗,umax],ifuini,0∈(f−1+,u∗),[u∗,umax],ifuini,0∈[u∗,umax], $
|
(3.4) |
and
$ uoutj∈{[0,u∗],ifuoutj,0∈[0,u∗),[0,u∗],ifuoutj,0=u∗andfreeflowing,{u∗}∪[0,f−1+),ifuoutj,0=u∗andcongested,{uoutj,0}∪[0,η(uoutj,0)),ifuoutj,0∈(u∗,umax]. $
|
(3.5) |
Proof. Using the definition of the supply and demand functions in definition 3.3 and the results from [12,section 5.2.3] we can follow the proof of [12,theorem 5.1.2] and uniquely determine the inflows which maximize the flux through junctions subject to the distribution parameters. It remains to show that by the choice of the density values the correct waves are induced.
We start with considering the outgoing roads. If
For
On the contrary, considering the incoming roads and
Now, let
$ s=fini−f(uini,0)u∗−uini,0<0. $
|
Further, if
Now, let us turn to the remaining case of
$ s=fini−f(uini,0)u∗−uini,0≤0, $
|
as
Hence, the choices of the densities induce the correct waves.
Now, we consider the case of more incoming than outgoing roads. Exemplary, we study the 2–to–1 situation, even though the results can be easily extended to the
Theorem 3.8. Let
$ uini∈{{uini,0}∪(η(uini,0),umax],ifuini,0∈[0,f−1+],{uini,0}∪[u∗,umax],ifuini,0∈(f−1+,u∗),[u∗,umax],ifuini,0∈[u∗,umax], $
|
(3.6) |
and
$ uout1∈{[0,u∗],ifuout1,0∈[0,u∗),[0,u∗],ifuout1,0=u∗andfreeflowing,{u∗}∪[0,f−1+),ifuout1,0=u∗andcongested,{uout1,0}∪[0,η(uout1,0)),ifuout1,0∈(u∗,umax]. $
|
(3.7) |
Proof. Following [11,Section 3.2.2], the flux values at the junction can be calculated with the following steps:
1. Calculate the maximal possible flux
2. Consider the right of way parameter and the flux maximization and calculate the intersection
3. If
4. The flux values are given by
Completely analogous to theorem 3.7 we can show that the choice of the densities admits the correct wave speeds.
The Riemann solutions proposed in theorem 3.7 and theorem 3.8 are the key ingredients for the splitting algorithm on networks in the next section.
Different problems might occur when designing a numerical scheme for a conservation law with discontinuous flux. However, the main difficulties are induced by the zero waves. Since these waves have infinite speed, the regular CFL condition is scaled by the regularization parameter
We consider a flux function
$ g(u)=−αH(u−u∗), $
|
where
This case has been already treated in [22] and will be the basis for the splitting algorithm on networks. When solving the scalar conservation law (2.1) on a single road, the boundary value in the case
$ ˜g(u)={0,u∈[0,u∗),[−α,0],u=u∗,−α,u∈(u∗,umax]. $
|
(4.1) |
Furthermore, we define
$ F(t)=f(u∗−)⇔G(t)=0,F(t)=f(u∗+)⇔G(t)=−α. $
|
(4.2) |
Additionally to the assumptions 2.2, we assume:
Assumption 4.1. [22,Assumption 1.1] The initial data satisfies
Remark 4.2. We emphasize that the original splitting algorithm for a single road [22] is not limited to piecewise linear discontinuous flux functions. Another prominent example might be concave piecewise quadratic flux functions with discontinuity again at
Then, we are able to handle the flux function
$ \lambda = \frac{\Delta t}{\Delta x}. $ |
For an integer
As the algorithm splits the function
We denote the backward spatial difference by
$ rn=rn+12=r(tn),Un+120=Un0=rnsn=sn+12=s(tn)Un+12K+1=UnK+1=sn. $
|
The function
$ gn+12K+1=gnK+1=G(tn)={0,ifs(tn)<u∗,0,ifs(tn)=u∗,trafficaheadofx=bisfree−flowing,−α,ifs(tn)=u∗,trafficaheadofx=biscongested,−α,ifs(tn)>u∗. $
|
(4.3) |
That means, we can describe the boundary value
Definition 4.3 [22,Eq (3.7)] Let
$ ˜G(u)={u,u∈[0,u∗),[u∗,u∗+λα],u=u∗,u+λα,u∈(u∗,umax], ˜G−1(u)={u,u∈[0,u∗),u∗,u∈[u∗,u∗+λα),u−λα,u∈[u∗+λα,umax+λα]. $
|
The splitting algorithm [22] can be then expressed as
$ {{Un+1/2k=˜G−1(Unk−λgn+1/2k+1),k=K,K−1,…,1,gn+1/2k=(Un+1/2k−Unk+λgn+1/2k+1)/λ,k=K,K−1,…,1,Un+1k=Un+1/2k−λΔ−pg(Un+1/2k+1,Un+1/2k), k∈K. $
|
(4.4) |
Note that the first half step, which includes the first two equations, is implicit. Nevertheless, instead of solving a nonlinear system of equations, the equation can be solved backwards in space starting with
We note that for the implicit equation a CFL condition is not needed, but it is required for the third step. As
As shown in [22,Theorem 5.1] the splitting algorithm (4.4) converges to a weak solution of Eq (2.1). However, obtaining a similar statement about weak entropy solutions is still an open problem.
The key idea to numerically solve such discontinuous conservation laws on networks is to use the splitting algorithm only on the roads and determine the correct in- and outflows at the boundaries by the help of the Riemann solver established in, e.g., theorem 3.7. As the splitting algorithm works with flux values, there is no need to compute the exact densities at the junction. Instead we need to know how the solution at the junction influences the flux values. The algorithm for a single junction is depicted in algorithm 1. The general description of the algorithm allows for either junctions with given distribution or right of way parameters. For simplicity, we assume that each road is represented by the same interval
Remark 4.4. Note that this simplification enables the use of the same grid points on each road which spares further sub- or superindices. However, the algorithm can be easily adapted to different road lengths.
We assume in the following that the space and time grid is the same as in the previous subsection. The approximate solutions are denoted by
The overall strategy of the splitting algorithm on a network consists of three important steps:
1. Solve the optimization problem induced by definition 3.2 (in particular item 4) at the junction to calculate the flux values
Here, it is crucial to use the discontinuous flux function
These flux values bring us now to:
Require: number of incoming roads Ensure: approximate solutions 1: Initilization: 2: 3: 4: 5: for 6: Solve the by definition 3.2 induced optimization problem at the junction based on the flux 7: Compute the densities at the junction with an appropriate Riemann solver 8: Compute the adjusted flux values for the incoming roads 9: for 10: 11: for 12: 13: 14: end for 15: 16: for 17: 18: end for 19: end for 20: for 21: Compute 22: for 23: 24: 25: end for 26: 27: for 28: 29: end for 30: end for 31: end for |
Using the calculated (unadjusted) flux values from step one, we can determine the densities at the junction with the help of the appropriate Riemann solver (theorem 3.7 and Eqs (3.4)–(3.5) or theorem 3.8 and Eqs (3.6)–(3.7)) at the junction (line 7). Then, these density values can be used to calculate the corresponding flux value of
In addition, the first steps give us all the ingredients for the final step:
The adjusted flux values from the previous step are important for the second half step (line 17 or 28) of the splitting algorithm which uses a Godunov type scheme based on
Further boundary data is needed in the first half step of the algorithm, lines 12–13 and 23–24. Here, we start with
$ gn+1/2,ini,K+1=gn,ini,K+1=fini−fini,adj. $
|
(4.5) |
Note that the definition of
Furthermore, we can decrease the computational costs of the algorithm: In the second step (or in line 7 of algorithm 1) the density at the junction is computed. This can be very expensive and hence we aim to avoid this. In the third step we have seen that for the missing boundary data only the adjusted flux values are necessary and not the densities at the junction themselves. Therefore, studying first each junction type in detail allows to determine the corresponding flux values based on the density values and the supply and demand functions and the intermediate expensive step for the computation of the exact densities can be skipped. Hence, we combine the first and second step of the strategy in one single step. In the following, we will study a 1–to–1 junction in detail and present the tailored algorithm. As the strategy is completely analogous for the 1–to–2 junction and 2–to–1 junction, we will only present the algorithms and discuss important properties. The algorithms can then be used to replace the lines 6 to 8 in algorithm 1.
Remark 4.5. The extension to
Remark 4.6. Further note that the presented strategy and also algorithm 1 can be used for arbitrary junctions and nonlinear discontinuous flux functions once an appropriate Riemann solver similar to theorem 3.7 and 3.8 is established.
First, we consider a junction with one incoming and one outgoing road in detail. Let
Case A: demand and supply are equal
There are two different situations depending on the density value of the incoming road where demand and supply can be equal.
1.
2.
Case B: supply is restrictive
If the supply is restrictive, i.e.
1.
2.
Case C: demand is restrictive
If the demand is restrictive, the outgoing road is able to take the whole amount of vehicles the demand wants to send. Here, we only have one possible situation for the initial condition on the incoming road:
1.
Note that we have the same flux function on each road. Hence, in the case
The whole procedure is summarized in algorithm 2.
Require: Demand Ensure: Flux values 1: 2: 3: if 4: 5: else if 6: 7: 8: end if |
Remark 4.7. Theoretically, the Riemann solver in theorem 3.7 coincides for a 1–to–1 junction with the Riemann solver on a single road. Hence, the procedure described in algorithm 2 leads to the same solution. In contrast to that on the numerical level, the splitting algorithm for a 1–to–1 junction does not exactly coincide with the splitting algorithm used for a single road [22]. The reason for the computational difference is that the splitting algorithm for the 1–to–1 junction considers the exact solution of the Riemann problem at the junction point and hence uses exact values for
The difference to the 1–to–1 junction is now that we have to consider two supplies. So the case distinctions to determine the flux values are a bit more complex. However, the procedure itself does not change. The details can be seen in algorithm 3.
We remark that if the inflow on the first road equals the demand and the restriction given by at least one of the supplies, the latter road needs to be congested. The Riemann solver states congestion such that the flux needs to be adjusted. Then, the incoming road either stays free flowing or the solution is given by
Require: Demand Ensure: Flux values 1: 2: 3: if 4: if 5: 6: else if 7: 8: end if 9: else if 10: if 11: 12: else 13: 14: end if 15: if 16: 17: 18: else if 19: 20: else if 21: 22: end if 23: end if |
If at least one of the supply restrictions is active, we need to adjust the corresponding flux values as in the 1–to–1 case and the incoming road. Nevertheless, here an interesting case (which is not possible in the 1–to–1 situation) can occur. The solution of the Riemann problem at the incoming road can be given by
Recall that for two incoming and one outgoing roads, the maximal possible flux on the outgoing road is given by
Require: Demand Ensure: Flux values 1: 2: 3: if 4: 6: 7: end if 8: 9: if 10: 11:else if 12: if 13: 14: end if 15: if 16: if 17: 18: else 19: 20: end if 21: end if 22: if 23: if 24: 25: else 26: 27: end if 28: end if 29: end if |
As before, no adjustment is needed when the demand on both roads is smaller than the supply. If the supply is active we might need to adjust the outgoing road and in most cases at least one incoming road. As in the 1–to–2 case,
In this section, we present some numerical examples to compare the splitting algorithm on networks with the Riemann solver. Further, we compute the solution using a regularized flux and the Godunov scheme. We consider the following flux function:
$ f(u)={u,u∈[0,0.5),0.5(1−u),u∈[0.5,1]. $
|
(5.1) |
The corresponding regularization is given by Eq (2.7).
We consider in particular the 1–to–2 and 2–to–1 situations. For our comparison, we choose constant initial data on each road. The junction is always located at
In both scenarios the supply of the first outgoing road is restrictive. The parameter settings are as follows:
The exact solution is given by
$ uin1(x,t)={0.4,x≥−32t,0.5,−32t<x<−12t,1315,−12t<x<0,uout1(x,t)=0.9,uout2(x,t)={115β20≤x≤841t,0.7,841t<x. $
|
Apparently, the solution induces two waves on the incoming road.
The exact solution is given by
$ uin1(x,t)={0.4,x≥−t,0.5,−t<x<0,uout1(x,t)=0.7,1≤x≤3,uout2(x,t)={0.3β21≤x≤t,0.2,t<x. $
|
This example generates
In Table 1, the
Example 1 | |||||
Splitt. | Reg. | ||||
33.44e-03 | 46.77e-03 | 82.26e-03 | 51.82e-03 | ||
24.17e-03 | 29.05e-03 | 65.59e-03 | 30.69e-03 | ||
14.16e-03 | 20.12e-03 | 60.86e-03 | 24.45e-03 | ||
8.97e-03 | 12.49e-03 | 58.37e-03 | 20.44e-03 | ||
CR | 0.64695 | 0.62453 | 0.1593 | 0.4353 | |
Example 2 | |||||
Splitt. | Reg. | ||||
4.58e-03 | 7.41e-03 | 44.70e-03 | 14.57e-03 | ||
2.97e-03 | 4.24e-03 | 43.47e-03 | 11.28e-03 | ||
2.03e-03 | 2.89e-03 | 42.50e-03 | 10.06e-03 | ||
1.24e-03 | 1.99e-03 | 41.80e-03 | 9.21e-03 | ||
CR | 0.61911 | 0.62327 | 0.0322 | 0.2150 |
We can see that the error terms obtained by the splitting algorithm are the lowest and so the computational costs. Obviously, the error terms increase with a lower CFL due to numerical diffusion. For a direct comparison with the regularized approach, the CFL condition should be the same. Meaning that the second column in Table 1 for the splitting algorithm should be compared with the first one of the regularized approach. In this case, we can see that the splitting algorithm also performs better in both examples. By choosing a smaller regularization parameter the performance of the regularized approach increases, but also the computational costs. To obtain similar error terms as for the splitting algorithm the regularization parameter needs to be further reduced at very high computational costs.
For
Here, we consider two scenarios, where in the first scenario the demand is restrictive while in the second one the supply. The parameter settings are as follows:
The exact solution is given by
$ uin1(x,t)=0.2,uout1(x,t)={0.450≤x≤t,0.3,t<xuin2(x,t)=0.25. $
|
The exact solution is given by
$ uin1(x,t)={0.5x≤−2t,0.6,−2t<x<0,uout1(x,t)={0.50≤x≤t,0.4,t<x,uin2(x,t)={0.8x≤−0.5t,0.7,−0.5t<x<0. $
|
In particular, the flux value for the first incoming road needs to be adjusted from
Considering the
Example 1 | |||||
Splitt. | Reg. | ||||
9.25e-03 | 16.22e-03 | 16.22e-03 | 17.01e-03 | ||
5.90e-03 | 11.63e-03 | 11.63e-03 | 12.19e-03 | ||
2.98e-03 | 8.13e-03 | 8.13e-03 | 8.52e-03 | ||
8.97e-03 | 5.71e-03 | 5.71e-03 | 5.99e-03 | ||
CR | 0.53838 | 0.50353 | 0.50353 | 0.50347 | |
Example 2 | |||||
Splitt. | Reg. | ||||
14.12e-03 | 20.10e-03 | 85.69e-03 | 27.55e-03 | ||
9.65e-03 | 13.86e-03 | 79.98e-03 | 21.65e-03 | ||
6.41e-03 | 9.57e-03 | 75.96e-03 | 17.49e-03 | ||
4.51e-03 | 6.69e-03 | 73.22e-03 | 14.65e-03 | ||
CR | 0.55295 | 0.52959 | 0.07551 | 0.30432 |
For
We have presented a Riemann solver at a junction for conservation laws with discontinuous flux. We have adapted the splitting algorithm of [22] to networks and demonstrated its validity in comparison with the exact solution. We have also pointed out that the splitting algorithm on networks is faster and more accurate than the approach with a regularized flux. Future research includes the investigation of other network models, where the flux is discontinuous.
J.F. was supported by the German Research Foundation (DFG) under grant HE 5386/18-1 and S.G. under grant GO 1920/10-1.
The authors declare there is no conflict of interest.
[1] | American Cancer Society. Cancer Facts & Figures 2015. Atlanta: American Cancer Society; 2015. |
[2] | Rousseau A, Badoual C (2011) Squamous cell carcinoma: an overvie Atlas Genet Cytogenet Oncol Haematol. Head and Neck, in press. |
[3] | Schantz SP, Harrison LB, Forastiere A (2001) Tumors of the nasal cavity and paranasal sinuses, nasopharynx, oral cavity, and oropharynx. In: DeVita VT, Hellman SA, Rosenberg SA, eds. Cancer: principles and practice of oncology. 6th ed. Philadelphia: Lippincott Williams & Wilkins: 797-860. |
[4] | Rubin Grandis J, Melhem MF, Gooding WE, et al. (1988) Levels of TGF-alpha and EGFR protein in head and neck squamous cell carcinoma and patient survival. J Natl Cancer Inst 90: 824-832. |
[5] | Chung CH, Ely K, McGavran L, et al. (2006) Increased epidermal growth factor receptor gene copy number is associated with poor prognosis in head and neck squamous cell carcinomas. J Clin Oncol 24: 4170-4176. |
[6] |
Temam S, Kawaguchi H, El-Naggar AK, et al. (2007) Epidermal growth factor receptor copy number alterations correlate with poor clinical outcome in patients with head and neck squamous cancer. J Clin Oncol 25: 2164-2170. doi: 10.1200/JCO.2006.06.6605
![]() |
[7] |
Bonner JA, Harari PM, Giralt J (2006) Radiotherapy plus cetuximab for squamous-cell carcinoma of the head and neck. N Engl J Med 354: 567-578. doi: 10.1056/NEJMoa053422
![]() |
[8] | Goldstein NI, Prewett M, Zuklys K, et al. (1995) Biological efficacy of a chimeric antibody to the epidermal growth factor receptor in a human tumor xenograft model. Clin Cancer Res 1: 1311-1318. |
[9] |
Li S, Schmitz KR, Jeffrey PD, et al. (2005) Structural basis for inhibition of the epidermal growth factor receptor by cetuximab. Cancer Cell 7: 301-311. doi: 10.1016/j.ccr.2005.03.003
![]() |
[10] | Sato JD, Kawamoto T, Le AD, et al. (1983) Biological effects in vitro of monoclonal antibodies to human epidermal growth factor receptors. Mol Biol Med 1: 511-529. |
[11] | Kang X, Patel D, Ng S, et al. (2007) High affinity Fc receptor binding and potent induction of antibody-dependent cellular cytotoxicity (ADCC) in vitro by anti-epidermal growth factor receptor antibody cetuximab. J Clin Oncol 25: 128s. |
[12] | Kimura H, Sakai K, Arao T, et al. (2007) Antibody-dependent cellular cytotoxicity of cetuximab against tumor cells with wild-type or mutant epidermal growth factor receptor. Cancer Sci 98: 1275-1280. |
[13] |
Zhang W, Gordon M, Schultheis AM, et al. (2007) FCGR2A and FCGR3A polymorphisms associated with clinical outcome of epidermal growth factor receptor expressing metastatic colorectal cancer patients treated with single-agent cetuximab. J Clin Oncol 25: 3712-3718. doi: 10.1200/JCO.2006.08.8021
![]() |
[14] | Fan Z, Baselga J, Masui H, et al. (1993) Antitumor effect of anti-epidermal growth factor receptor monoclonal antibodies plus cis-diamminedichloroplatinum on well established A431 cell xenografts. Cancer Res 53: 4637-4642. |
[15] |
Burtness B, Goldwasser MA, Flood W, et al. (2005) Phase III randomized trial of cisplatin plus placebo compared with cisplatin plus cetuximab in metastatic/recurrent head and neck cancer: an Eastern Cooperative Oncology Group study. J Clin Oncol 23: 8646-8654 doi: 10.1200/JCO.2005.02.4646
![]() |
[16] | Thomas KH, Patrick JS (2013) Antigen-specific immunotherapy in head and neck cancer. Adv Cell Mol Otolaryngol 1. |
[17] | Ira Mellman, George Coukos, Glenn Dranoff (2011) Cancer immunotherapy comes of age. Nature 480: 480-489. |
[18] | Andrew MS, James PA, Jedd DW (2012) Monoclonal antibodies in cancer therapy. Cancer Immun 12: 14. |
[19] |
Brahmer JR, Drake CG, Wollner I, et al. (2010) Phase I study of single-agent anti-programmed death-1 (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics and immunologic correlates. J Clin Oncol 28: 3167-3175. doi: 10.1200/JCO.2009.26.7609
![]() |
[20] |
Hodi FS, O'Day SJ, McDermott DF, et al. (2010) Improved survival with ipilimumab in patients with metastatic melanoma. N Engl J Med 363: 711-723. doi: 10.1056/NEJMoa1003466
![]() |
[21] |
Suntharalingam, G, Perry MR, Ward S, et al. (2006) Cytokine storm in a phase 1 trial of the anti-CD28 monoclonal antibody TGN1412. N Engl J Med 355: 1018-1028. doi: 10.1056/NEJMoa063842
![]() |
[22] |
Brahmer JR, Drake CG, Wollner I, et al. (2010). Phase I study of single-agent anti-programmed death-1 (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics and immunologic correlates. J Clin Oncol 28: 3167-3175. doi: 10.1200/JCO.2009.26.7609
![]() |
[23] |
Keir ME, Butte MJ, Freeman GJ, et al. (2008) PD-1 and its ligands in tolerance and immunity. Annu Rev Immunol 26: 677-704 doi: 10.1146/annurev.immunol.26.021607.090331
![]() |
[24] | Parsa AT, Waldron JS, Panner A, et al. (2006) Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in glioma. Nature Med 13: 84-88. |
[25] |
Gadiot J, Hooijkaas AI, Kaiser AD, et al. (2011) Overall survival and PD-L1 expression in metastasized malignant melanoma. Cancer 117: 2192-2201. doi: 10.1002/cncr.25747
![]() |
[26] | Gao Q, Wang XY, Qiu SJ, et al. (2009) Overexpression of PD-L1 significantly associates with tumor aggressiveness and postoperative recurrence in human hepatocellular carcinoma. Clin Cancer Res 15: 971-979. |
[27] |
Leach DR, Krummel MF, Allison JP (1996) Enhancement of antitumor immunity by CTLA-4. Science 271: 1734-1736. doi: 10.1126/science.271.5256.1734
![]() |
[28] | Van Cutsem E, Köhne CH, Hitre E, et al. (2009) Cetuximab and chemotherapy as initial treatment for metastatic colorectal cancer. N Engl J Med 360: 1408-1417. |
[29] | Schliemann C, Neri D (2010) Antibody-based vascular tumor targeting. Cancer Res 180: 201-216. |
[30] |
Pillay V, Gan HK, Scott AM (2011) Antibodies in oncology. N Biotechnol 28: 518-529. doi: 10.1016/j.nbt.2011.03.021
![]() |
[31] |
Divgi CR, Welt S, Kris M, et al. (1991) Phase I and imaging trial of indium 111-labeled anti-epidermal growth factor receptor monoclonal antibody 225 in patients with squamous cell lung carcinoma. J Natl Cancer Inst 83: 97-104. doi: 10.1093/jnci/83.2.97
![]() |
[32] |
Ellis LM, Hicklin DJ (2008) VEGF-targeted therapy: mechanisms of anti-tumor activity. Nat Rev Cancer 8: 579-591. doi: 10.1038/nrc2403
![]() |
[33] |
Friedman HS, Prados MD, Wen PY, et al. (2009) Bevacizumab alone and in combination with irinotecan in recurrent glioblastoma. J Clin Oncol 27: 4733-4740. doi: 10.1200/JCO.2008.19.8721
![]() |
[34] |
Heiss MM, Murawa P, Koralewski P, et al. (2010) The trifunctional antibody catumaxomab for the treatment of malignant ascites due to epithelial cancer: results of a prospective randomized phase II/III trial. Int J Cancer 127: 2209-2221. doi: 10.1002/ijc.25423
![]() |
[35] | Boland WK, Bebb G. (2009) Nimotuzumab: a novel anti-EGFR monoclonal antibody that retains anti-EGFR activity while minimizing skin toxicity. Expert Opin Biol Ther 9:1199-1206. |
[36] |
Curran D, Giralt J, Harari PM, et al. (2007) Quality of life in head and neck cancer patients after treatment with high-dose radiotherapy alone or in combination with cetuximab. J Clin Oncol 25: 2191-2197. doi: 10.1200/JCO.2006.08.8005
![]() |
[37] |
Bonner JA, Harari PM, Giralt J, et al. (2010) Radiotherapy plus cetuximab for locoregionally advanced head and neck cancer: 5-year survival data from a phase 3 randomized trial, and relation between cetuximab-induced rash and survival. Lancet Oncol 11: 21-28. doi: 10.1016/S1470-2045(09)70311-0
![]() |
[38] |
Koutcher L, Sherman E, Fury M, et al. (2011) Concurrent cisplatin and radiation versus cetuximab and radiation for locally advanced head-and-neck cancer. Int J Radiat Oncol Biol Phys 81: 915-922. doi: 10.1016/j.ijrobp.2010.07.008
![]() |
[39] | Chew A, Hay J, Laskin JJ, et al. (2011) Toxicity in combined modality treatment of HNSCC: Cisplatin versus cetuximab. J Clin Oncol 29: 5526. |
[40] | Shapiro LQ, Sherman EJ, Koutcher L, et al. (2012) Efficacy of concurrent cetuximab (C225) versus 5-fluorouracil/carboplatin (5FU/CBDCA) or cisplatin (CDDP) with intensity-modulated radiation therapy (IMRT) for locally advanced head and neck cancer (LAHNSCC). J Clin Oncol 30: 5537. |
[41] |
Pryor DI, Porceddu SV, Burmeister BH, et al. (2009) Enhanced toxicity with concurrent cetuximab and radiotherapy in head and neck cancer. Radiother Oncol 90: 172-176. doi: 10.1016/j.radonc.2008.09.018
![]() |
[42] |
Vermorken JB, Mesia R, Rivera F, et al. (2008) Platinum-based chemotherapy plus cetuximab in head and neck cancer. N Engl J Med 359: 1116-1127. doi: 10.1056/NEJMoa0802656
![]() |
[43] |
Hitt R, Irigoyen A, Cortes-Funes H, et al. (2012). Spanish Head and Neck Cancer Cooperative Group (TTCC) Phase II study of the combination of cetuximab and weekly paclitaxel in the first-line treatment of patients with recurrent and/or metastatic squamous cell carcinoma of head and neck. Ann Oncol 23: 1016-1022. doi: 10.1093/annonc/mdr367
![]() |
[44] | Ang KK, Zhang QE, Rosenthal DI, et al. (2011) A randomized phase III trial (RTOG 0522) of concurrent accelerated radiation plus cisplatin with or without cetuximab for stage III-IV head and neck squamous cell carcinomas (HNC). J Clin Oncol 30: 360. |
[45] | Ley J, Mehan P, Wildes TM, et al. (2012) Concurrent cisplatin vs. cetuximab with definitive radiation therapy (RT) for head and neck squamous cell carcinoma (HNSCC): A retrospective comparison. Multidisciplinary Head and Neck Cancer Symposium (Phoenix, AZ) |
[46] | Balz V, Scheckenbach K, Gotte K, et al. (2003) Is the p53 inactivation frequency in squamous cell carcinomas of the head and neck underestimated? Analysis of p53 exons 2-11 and human papillomavirus 16/18 E6 transcripts in 123 unselected tumor specimens. Cancer Res 63: 1188-1191. |
[47] |
Deleo AB (1998) p53-based immunotherapy of cancer. Crit Rev Immunol 18: 29-35. doi: 10.1615/CritRevImmunol.v18.i1-2.40
![]() |
[48] |
Hoffmann TK, Nakano K, Elder EM, et al. (2000) Generation of T cells specific for the wild-type sequence p53(264-272) peptide in cancer patients: Implications for immunoselection of epitope loss variants. J Immunol 165: 5938-5944. doi: 10.4049/jimmunol.165.10.5938
![]() |
[49] |
Andrade FPA, Ito D, Deleo AB, et al. (2010) CD8+ T cell recognition of polymorphic wild-type sequence p53(65-73) peptides in squamous cell carcinoma of the head and neck. Cancer Immunol Immunother 59: 1561-1568. doi: 10.1007/s00262-010-0886-1
![]() |
[50] | Chikamatsu K, Albers A, Stanson J, et al. (2003) P53(110-124)-specific human CD4+ T-helper cells enhance in vitro generation and antitumor function of tumor-reactive CD8+ T cells. Cancer Res 63: 3675-3681. |
[51] |
Albers AE, Ferris RL, Kim GG, et al. (2005) Immune responses to p53 in patients with cancer: Enrichment in tetramer+p53 peptide-specific T cells and regulatory T cells at tumor sites. Cancer Immunol Immunother 54: 1072-1081. doi: 10.1007/s00262-005-0670-9
![]() |
[52] |
Zhang Y, Sturgis EM, Huang Z, et al. (2012) Genetic variants of the p53 and p73 genes jointly increase risk of second primary malignancies in patients after index squamous cell carcinoma of the head and neck. Cancer 118: 485-492. doi: 10.1002/cncr.26222
![]() |
[53] | Clayman GL, El-Naggar AK, Lippman SM, et al. (1998) Adenovirus-mediated p53 gene transfer in patients with advanced recurrent head and neck squamous cell carcinoma. J Clin Oncol 16: 2221-2232. |
[54] |
Heusinkveld M, Goedemans R, Briet RJ, et al. (2012) Systemic and local human papillomavirus 16-specific T-cell immunity in patients with head and neck cancer. Int J Cancer 131: E74-85. doi: 10.1002/ijc.26497
![]() |
[55] |
Wansom D, Light E, Worden F, et al. (2010) Correlation of cellular immunity with human papillomavirus 16 status and outcome in patients with advanced oropharyngeal cancer. Arch Otolaryngol Head Neck Surg 136: 1267-1273. doi: 10.1001/archoto.2010.211
![]() |
[56] | The Cancer Genome Atlas Network, 2015. Comprehensive Genomic Characterization of Head and Neck Squamous Cell Carcinomas. Nature 517: 576-582. |
[57] | Wei G, John ZHL, Jimmy YWC, et al. (2012) mTOR pathway and mTOR inhibitors in head and neck cancer department of surgery. Otolaryngology. |
[58] |
Hay N, Sonenberg N (2004) Upstream and downstream of mTOR. Genes Dev 18: 1926-1945. doi: 10.1101/gad.1212704
![]() |
[59] |
Lionello SM, Loreggian BL (2012) High mTOR expression is associated with a worse oncological outcome in laryngeal carcinoma treated with postoperative radiotherapy: a pilot study. J Oral Pathol Med 41: 136-140. doi: 10.1111/j.1600-0714.2011.01083.x
![]() |
Require: number of incoming roads Ensure: approximate solutions 1: Initilization: 2: 3: 4: 5: for 6: Solve the by definition 3.2 induced optimization problem at the junction based on the flux 7: Compute the densities at the junction with an appropriate Riemann solver 8: Compute the adjusted flux values for the incoming roads 9: for 10: 11: for 12: 13: 14: end for 15: 16: for 17: 18: end for 19: end for 20: for 21: Compute 22: for 23: 24: 25: end for 26: 27: for 28: 29: end for 30: end for 31: end for |
Require: Demand Ensure: Flux values 1: 2: 3: if 4: 5: else if 6: 7: 8: end if |
Require: Demand Ensure: Flux values 1: 2: 3: if 4: if 5: 6: else if 7: 8: end if 9: else if 10: if 11: 12: else 13: 14: end if 15: if 16: 17: 18: else if 19: 20: else if 21: 22: end if 23: end if |
Require: Demand Ensure: Flux values 1: 2: 3: if 4: 6: 7: end if 8: 9: if 10: 11:else if 12: if 13: 14: end if 15: if 16: if 17: 18: else 19: 20: end if 21: end if 22: if 23: if 24: 25: else 26: 27: end if 28: end if 29: end if |
Example 1 | |||||
Splitt. | Reg. | ||||
33.44e-03 | 46.77e-03 | 82.26e-03 | 51.82e-03 | ||
24.17e-03 | 29.05e-03 | 65.59e-03 | 30.69e-03 | ||
14.16e-03 | 20.12e-03 | 60.86e-03 | 24.45e-03 | ||
8.97e-03 | 12.49e-03 | 58.37e-03 | 20.44e-03 | ||
CR | 0.64695 | 0.62453 | 0.1593 | 0.4353 | |
Example 2 | |||||
Splitt. | Reg. | ||||
4.58e-03 | 7.41e-03 | 44.70e-03 | 14.57e-03 | ||
2.97e-03 | 4.24e-03 | 43.47e-03 | 11.28e-03 | ||
2.03e-03 | 2.89e-03 | 42.50e-03 | 10.06e-03 | ||
1.24e-03 | 1.99e-03 | 41.80e-03 | 9.21e-03 | ||
CR | 0.61911 | 0.62327 | 0.0322 | 0.2150 |
Example 1 | |||||
Splitt. | Reg. | ||||
9.25e-03 | 16.22e-03 | 16.22e-03 | 17.01e-03 | ||
5.90e-03 | 11.63e-03 | 11.63e-03 | 12.19e-03 | ||
2.98e-03 | 8.13e-03 | 8.13e-03 | 8.52e-03 | ||
8.97e-03 | 5.71e-03 | 5.71e-03 | 5.99e-03 | ||
CR | 0.53838 | 0.50353 | 0.50353 | 0.50347 | |
Example 2 | |||||
Splitt. | Reg. | ||||
14.12e-03 | 20.10e-03 | 85.69e-03 | 27.55e-03 | ||
9.65e-03 | 13.86e-03 | 79.98e-03 | 21.65e-03 | ||
6.41e-03 | 9.57e-03 | 75.96e-03 | 17.49e-03 | ||
4.51e-03 | 6.69e-03 | 73.22e-03 | 14.65e-03 | ||
CR | 0.55295 | 0.52959 | 0.07551 | 0.30432 |
Require: number of incoming roads Ensure: approximate solutions 1: Initilization: 2: 3: 4: 5: for 6: Solve the by definition 3.2 induced optimization problem at the junction based on the flux 7: Compute the densities at the junction with an appropriate Riemann solver 8: Compute the adjusted flux values for the incoming roads 9: for 10: 11: for 12: 13: 14: end for 15: 16: for 17: 18: end for 19: end for 20: for 21: Compute 22: for 23: 24: 25: end for 26: 27: for 28: 29: end for 30: end for 31: end for |
Require: Demand Ensure: Flux values 1: 2: 3: if 4: 5: else if 6: 7: 8: end if |
Require: Demand Ensure: Flux values 1: 2: 3: if 4: if 5: 6: else if 7: 8: end if 9: else if 10: if 11: 12: else 13: 14: end if 15: if 16: 17: 18: else if 19: 20: else if 21: 22: end if 23: end if |
Require: Demand Ensure: Flux values 1: 2: 3: if 4: 6: 7: end if 8: 9: if 10: 11:else if 12: if 13: 14: end if 15: if 16: if 17: 18: else 19: 20: end if 21: end if 22: if 23: if 24: 25: else 26: 27: end if 28: end if 29: end if |
Example 1 | |||||
Splitt. | Reg. | ||||
33.44e-03 | 46.77e-03 | 82.26e-03 | 51.82e-03 | ||
24.17e-03 | 29.05e-03 | 65.59e-03 | 30.69e-03 | ||
14.16e-03 | 20.12e-03 | 60.86e-03 | 24.45e-03 | ||
8.97e-03 | 12.49e-03 | 58.37e-03 | 20.44e-03 | ||
CR | 0.64695 | 0.62453 | 0.1593 | 0.4353 | |
Example 2 | |||||
Splitt. | Reg. | ||||
4.58e-03 | 7.41e-03 | 44.70e-03 | 14.57e-03 | ||
2.97e-03 | 4.24e-03 | 43.47e-03 | 11.28e-03 | ||
2.03e-03 | 2.89e-03 | 42.50e-03 | 10.06e-03 | ||
1.24e-03 | 1.99e-03 | 41.80e-03 | 9.21e-03 | ||
CR | 0.61911 | 0.62327 | 0.0322 | 0.2150 |
Example 1 | |||||
Splitt. | Reg. | ||||
9.25e-03 | 16.22e-03 | 16.22e-03 | 17.01e-03 | ||
5.90e-03 | 11.63e-03 | 11.63e-03 | 12.19e-03 | ||
2.98e-03 | 8.13e-03 | 8.13e-03 | 8.52e-03 | ||
8.97e-03 | 5.71e-03 | 5.71e-03 | 5.99e-03 | ||
CR | 0.53838 | 0.50353 | 0.50353 | 0.50347 | |
Example 2 | |||||
Splitt. | Reg. | ||||
14.12e-03 | 20.10e-03 | 85.69e-03 | 27.55e-03 | ||
9.65e-03 | 13.86e-03 | 79.98e-03 | 21.65e-03 | ||
6.41e-03 | 9.57e-03 | 75.96e-03 | 17.49e-03 | ||
4.51e-03 | 6.69e-03 | 73.22e-03 | 14.65e-03 | ||
CR | 0.55295 | 0.52959 | 0.07551 | 0.30432 |