=Paper=
{{Paper
|id=Vol-1638/Paper62
|storemode=property
|title=Reduction in initial boundary value problem for HIV evolution model
|pdfUrl=https://ceur-ws.org/Vol-1638/Paper62.pdf
|volume=Vol-1638
|authors=Alexey A. Archibasov
}}
==Reduction in initial boundary value problem for HIV evolution model ==
https://ceur-ws.org/Vol-1638/Paper62.pdf
Mathematical Modeling
REDUCTION IN INITIAL BOUNDARY VALUE
PROBLEM FOR HIV EVOLUTION MODEL
A. Archibasov
Samara National Research University, Samara, Russia
Abstract. HIV evolution model that describes the dynamics of concentrations
of uninfected and infected cells is considered in the paper. The introduction of
dimensionless variables and parameters leads to the initial boundary value prob-
lem for singularly perturbed system of partial integro-differential equations. By
Tikhonov-Vasil’eva boundary functions method asymptotic solution with
boundary layer is constructed. Numeric simulation of complete and reduced
systems are also given.
Keywords: singular perturbations, asymptotic expansions, boundary layer,
boundary functions.
Citation: Archibasov A. Reduction in initial boundary value problem for HIV
evolution model. CEUR Workshop Proceedings, 2016; 1638: 508-514. DOI:
10.18287/1613-0073-2016-1638-508-514
1 Introduction
Singularly perturbed equations arise in the mathematical modeling of processes in
chemical kinetics, biology, physiology and other areas of science. For problems of
this type the methods, which give an asymptotic representation of solution, are suc-
cessfully applied. The aforesaid is especially true for mathematical models of evolu-
tion biology, where an extremely slow biological evolution process proceeds against
the background of significantly faster interactions of different nature. In this paper,
the method of boundary functions is used for constructing asymptotic expansions of
the solution to a singularly perturbed system of integro-differential equations with
small parameter multiplying derivative.
2 Model
Let us consider the system of partial integro-differential equations
ut b u ( s )v(t , s )ds qu,
(1)
0
vt vss mv uv
with the initial and boundary conditions
Information Technology and Nanotechnology (ITNT-2016) 508
Mathematical Modeling Archibasov A. Reduction in initial…
u (0) u 0 , v(0, s) v 0 ( s), vs (t ,0) 0, v(t , ) 0. (2)
This system is a mathematical model of HIV evolution in a continuous phenotype
space (see [1]). It is a result of the development of the in vivo dynamic model AIDS-1
proposed in [2]. The system (1) describes interactions between uninfected target cells
of concentration u (t ) , 𝑐𝑒𝑙𝑙⁄𝑚𝑚3 , and infected cells with the density distribution
v (t , s ) , 𝑐𝑒𝑙𝑙⁄𝑚𝑚3 . Correspondingly, V (t ) v(t , s)ds , 𝑐𝑒𝑙𝑙⁄𝑚𝑚3 , is the total con-
0
centration of infected cells. In the framework of this model a phenotype space is as-
sumed to be continuous and one-dimensional ( s [0,) - a dimensionless quantity).
There is a continuous influx of the target cells (from the thymus, where they mature)
at a rate b , 𝑐𝑒𝑙𝑙⁄(𝑚𝑚3 ∙ 𝑑𝑎𝑦). They die of natural reasons unrelated to the virus
infection at a rate 𝑞𝑢, where 𝑞 > 0, 1⁄𝑑𝑎𝑦. Parameter ( s ) as , 𝑚𝑚3 ⁄(𝑐𝑒𝑙𝑙 ∙ 𝑑𝑎𝑦),
𝑎 > 0, can be interpreted as the efficiency of a single virus particle in infecting a
target cell. Infected cells die as a result of the infection or natural reasons at a rate
𝑚𝑣, where 𝑚 = 𝑚(𝑠) > 0, 1⁄𝑑𝑎𝑦. The average life spans of the uninfected and in-
fected cells are 1⁄𝑞, 1⁄𝑚 respectively. Random mutations are described by the dis-
persion 𝜇𝑣𝑠𝑠 , 𝜇 > 0, 1⁄𝑑𝑎𝑦.
Without loss of generality and for simplicity it is assumed that only one of the param-
eters, (s ) , depends on 𝑠, and 𝑚 = 𝑚∗ for all phenotypes. Besides although the mod-
el is formulated for s [0,) , usually 𝑠 is considered to be belong to a finite inter-
val [0, 𝑙], 𝑙 = 10, for a given normalization and the condition v(t , ) 0 is replaced
by vs (t , l ) 0 .
Introducing dimensionless variables and parameters in the same way as in [3], the
system (1) and conditions (2) can be written in the form
ut b u vds u ,
(3)
0
vt vss mv puv,
u(0) u 0 , v(0, s) v 0 (s), v s (t ,0) 0, v s (t , ) 0. (4)
In (3) 𝜀 = √𝜇𝑚∗ ⁄𝑞~10−3 for HIV, thereby this system is singularly perturbed sys-
tem.
3 Reduced system
Setting 𝜀 = 0 in (3), we obtain the degenerate (or reduced) system
v t mv pu v v ss ,
u 1 1 v ds ,
(5)
0
with initial and boundary conditions
Information Technology and Nanotechnology (ITNT-2016) 509
Mathematical Modeling Archibasov A. Reduction in initial…
v (0, s) v 0 ( s), vs (t ,0) 0, vs (t , ) 0. (6)
It should be noted that the solution 𝑢̅(𝑡) of this system in general does not satisfy the
initial condition in (4). But the solution of associated system (where 𝑣 is a parameter)
𝑙
𝑢̃𝜏 = − (1 + ∫0 𝛽𝑣𝑑𝑠) 𝑢̃ + 1, (7)
𝑙
is 𝑢̃(𝜏) = (𝑢0 − 1⁄𝑓 )𝑒𝑥𝑝(−𝑓𝜏) + 1⁄𝑓 → 1⁄𝑓 as 𝜏 → +∞, 𝑓 = 1 + ∫0 𝛽(𝑠)𝑣 0 (𝑠)𝑑𝑠,
𝑙
that is the isolated root 𝑢 = 1⁄(1 + ∫0 𝛽𝑣𝑑𝑠) is an asymptotically stable rest point of
system (7) and the initial value 𝑢0 belongs to the domain of attraction of this root. In
[4] it is shown that there is a passage to the limit
lim ut , u (t ), 0 t T ,
0 (8)
lim vt, s, v t, s , 0 t T , 0 s l.
0
Here 𝑢̅(𝑡), 𝑣̅ (𝑡, 𝑠) are the solutions of (5), (6) and 𝑢(𝑡, 𝜀), vt , s, are the solutions
of (3), (4).
Figure 1 shows the results of numerical simulations for the full and reduced systems
(the thin line represents the plots corresponding to the full system (3), and the bold
line represents the plots corresponding to the reduced system (5)). As we see, the
solutions for the reduced and full systems agree fairly everywhere except a relatively
short transition. For convenience of comparison results in this figure are shown for
physical dimensional variables.
Fig. 1. Solutions of the full system (3) and reduced system (5)
If the initial condition is taken on or near the slow manifold, then the results obtained
for both systems, full and reduced, coincide everywhere (see Fig. 2).
Using the approach developed in [5], we can prove the existence and uniqueness of
degenerated problem (5), (6).
Information Technology and Nanotechnology (ITNT-2016) 510
Mathematical Modeling Archibasov A. Reduction in initial…
Fig. 2. Solutions of the full (thin line) and reduced (markers) systems, when initial point is
taken on the slow manifold
4 Asymptotic expansions
Let us find the solution to problem (3), (4) in the form of asymptotic expansions in
powers of small parameter 𝜀. As we mentioned above, the solution of the degenerate
system 𝑢̅(𝑡) does not satisfy initial condition in (4) and, therefore, it may be assumed
the existence of a boundary layer structure in the solution as 𝜀 → 0. Following [6-9],
in accordance with the method of boundary functions [10,11], such a solution can be
found as a sum of a regular and boundary-layer series
u(t , ) u (t , ) u(, ),
(9)
v(t , s, ) v (t , s, ) v(, s, ),
where u (t , ) k 0 k uk (t ) , v (t , s, ) k 0 k vk (t , s) are the regular parts of the
asymptotic expansions, u(, ) k 0 k k u() , v(, s, ) k 0 k k v(, s) are
the boundary-layer parts, and t is the boundary-layer variable. Formally substi-
tuting series (9) into equations (3) and the initial and boundary conditions (4) and
equating the regular and boundary-layer parts (taking into account that
d dt d d ), we obtain the equations
u t 1 u u v ds,
0
u 1 v v ds u u vds,
(10)
0 0
vt mv pu v v ss ,
v mv pu v v u uv v ss ,
Information Technology and Nanotechnology (ITNT-2016) 511
Mathematical Modeling Archibasov A. Reduction in initial…
u (0, ) u (0, ) u 0 ,
v (0, s, ) v(0, s, ) v 0 ( s),
(11)
vs t ,0, 0, vs ,0, 0,
vs t , , 0, vs , , 0.
It should be noted that regular terms in the right-hand sides of equations for boundary
functions in (10) are calculated at t .
Regular terms u 0 t , v0 t , s are the solutions of degenerate problem (5), (6). 0u is
the solution of initial value problem
0 u 1 v 0 ds 0u ,
0
0u 0 u 0 u0 (0),
namely 0u u 0 u0 (0) exp 1 v 0 ds .
0
0 v, s 0 can be found from the corresponding equation and the fact that this
function is the boundary-layer function.
To construct more accurate approximations of the solution to the full system, it is
necessary to use higher order asymptotic expansions. Expanding u , , v , s,
in Taylor series around 0
u , k 0 k Ak , v , s, k 0 k Bk , s ,
k r v r
Ak r 0 k r u r( k r ) (0), Bk , s r 0 k r 0, s
k k
t k r
and substituting the resulting expansions into equations (3) and conditions (4), we
equate the coefficients multiplying equal powers of 𝜀 and find 𝑘-th, (𝑘 ≥ 1), terms of
the asymptotic expansions according to the following scheme.
Find k v ( , s ) from the equation
k v l 0 Al k l 1v Bk l 1 l u l u k l 1v
k 1
m k 1v k 1v ss
and the condition k v(, s ) 0, s [0, ).
u k (t ) , vk (t , s ) are the solution of the equations
Information Technology and Nanotechnology (ITNT-2016) 512
Mathematical Modeling Archibasov A. Reduction in initial…
k 1
uk 1t l 0 ul vk l ds
uk
0
,
1 v0 ds
0
k 1
vkt mv k pu0vk vkss pl 0 uk l vl ,
satisfying the conditions vk (0, s ) k v(0, s ) , vks t ,0 0 , vks (t , ) 0 .
k u () is the solution of initial value problem
k u 1 v 0 ds k u l 0 Al k l vds
k 1
0 0
l u Bk l k l v ds),
0
k u (0) u k (0).
The solution to this problem is always the boundary-layer function.
It should be noted that only zeroth regular terms of the asymptotic expansions are
obtained from nonlinear equations. Terms of higher order approximations can be
found from linear equations.
The asymptotic character of the expansions is justified as described [6-11].
5 Conclusion
Mathematical models in evolution biology should necessary combine the processes,
specific time scales of which differ by several orders of magnitude. Accordingly such
models are postulated in the form of the singularly perturbed systems of equations.
The results obtained for relatively simple model can be naturally extended to much
more complex models of evolution biology.
Acknowledgements
This work is supported by the Russian Foundation for Basic Research (grant № 14-
01-970-18-a) and the Ministry of education and science of the Russian Federation in
the framework of the implementation of Samara University for 2013-2020 years and
in the framework of the basic part of the state assignment (project № 214).
References
1. Korobeinikov A, Dempsey C. A continuous phenotype space model of RNA virus evolu-
tion within a host. Math. Biosci. Eng., 2014; 11(4): 919-927.
2. Nowak MA, May RM. Virus Dynamics: Mathematical Principles of Immunology and Vi-
rology. Oxford Univ. Press, 2000.
Information Technology and Nanotechnology (ITNT-2016) 513
Mathematical Modeling Archibasov A. Reduction in initial…
3. Archibasov AA, Korobeinikov A, Sobolev VA. Asymptotic expansions of solutions in a
singularly perturbed model of virus evolution. Comput. Math. Math. Phys., 2015; 55(2):
240-250.
4. Archibasov AA, Korobeinikov A, Sobolev VA. The passage to the limit in a singularly
perturbed system of partial integro-differential equations. Diff. Eq. [in print]
5. Dan Henry. Geometric Theory of Semilinear Parabolic Equations. Berlin, Springer-Verlag,
New York, Heidelberg, 1981.
6. Nefedov NN, Nikitin AG, Urazgil’dina TA. The Cauchy problem for a singularly per-
turbed Volterra integro-differential equation. Comput. Math. Math. Phys., 2006; 46(5):
768-775.
7. Nefedov NN, Nikitin AG. The Cauchy problem for a singularly perturbed integro-
differential Fredholm equation. Comput. Math. Math. Phys., 2007; 47(4): 629-637.
8. Nefedov NN, Nikitin AG. The initial boundary value problem for a nonlocal singularly
perturbed reaction-diffusion equation. Comput. Math. Math. Phys., 2012; 52(6): 926-931.
9. Nefedov NN, Omel’chenko OE. Boundary-layer solutions to quasilinear integro-
differential equations of the second order. Comput. Math. Math. Phys., 2002; 42(4), 470-
482.
10. Vasilieva AB, Butuzov VF, Kalachev LV. The Boundary Function Method for Singular
Perturbation Problems. Philadelphia, SIAM, 1995.
11. Vasil’eva AB, Butuzov VF. Asymptotic Expansions of Solutions to Singularly Perturbed
Equations. Moscow: Nauka, 1973. [in Russian]
Information Technology and Nanotechnology (ITNT-2016) 514