Namespaces
Variants
Actions

Difference between revisions of "Mixed and boundary value problems for parabolic equations and systems"

From Encyclopedia of Mathematics
Jump to: navigation, search
(Importing text file)
 
m (tex encoded by computer)
 
Line 1: Line 1:
 +
<!--
 +
m0641901.png
 +
$#A+1 = 190 n = 0
 +
$#C+1 = 190 : ~/encyclopedia/old_files/data/M064/M.0604190 Mixed and boundary value problems for parabolic equations and systems
 +
Automatically converted into TeX, above some diagnostics.
 +
Please remove this comment and the {{TEX|auto}} line below,
 +
if TeX found to be correct.
 +
-->
 +
 +
{{TEX|auto}}
 +
{{TEX|done}}
 +
 
Problems of finding solutions
 
Problems of finding solutions
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641901.png" /></td> </tr></table>
+
$$
 +
u ( x , t )  = ( u _ {1} ( x , t ) \dots u _ {m} ( x ,\
 +
t ) )
 +
$$
  
in a domain <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641902.png" /> of a Euclidean space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641903.png" /> (with points <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641904.png" />) of a parabolic system of equations or, when <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641905.png" />, of a parabolic equation satisfying additional conditions on some part of the boundary <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641906.png" /> of the domain <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641907.png" />.
+
in a domain $  D $
 +
of a Euclidean space $  \mathbf R  ^ {n+} 1 $(
 +
with points $  ( x , t ) = ( x _ {1} \dots x _ {n} , t ) $)  
 +
of a parabolic system of equations or, when $  m = 1 $,  
 +
of a parabolic equation satisfying additional conditions on some part of the boundary $  \partial  D $
 +
of the domain $  D $.
  
Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641908.png" /> be a domain in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m0641909.png" /> with sufficiently smooth boundary <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419010.png" /> and let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419011.png" /> be a cylinder <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419012.png" /> with lateral surface <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419013.png" />, lower base <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419014.png" /> and upper base <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419015.png" />. The mixed Petrovskii problem for a linear parabolic system
+
Let $  \Omega $
 +
be a domain in $  \mathbf R  ^ {n} $
 +
with sufficiently smooth boundary $  \partial  \Omega $
 +
and let $  D $
 +
be a cylinder $  \{ {x \in \Omega } : {0 < t < T } \} $
 +
with lateral surface $  \Gamma = \{ {x \in \partial  \Omega } : {0 < t < T } \} $,  
 +
lower base $  \Omega _ {0} = \{ {x \in \Omega } : {t = 0 } \} $
 +
and upper base $  \Omega _ {T} = \{ {x \in \Omega } : {t = T } \} $.  
 +
The mixed Petrovskii problem for a linear parabolic system
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419016.png" /></td> <td valign="top" style="width:5%;text-align:right;">(1)</td></tr></table>
+
$$ \tag{1 }
 +
u _ {t} + \sum _ {| \alpha | \leq  2 p }
 +
A _  \alpha  ( x , t ) D _ {x}  ^  \alpha  u  = f
 +
( x , t ) ,\  ( x , t ) \in D ,
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419017.png" /></td> </tr></table>
+
$$
 +
f ( x , t )  = ( f _ {1} ( x , t ) \dots f _ {m} ( x , t ) ) ,
 +
$$
  
in the cylinder <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419018.png" /> consists of finding solutions of this system satisfying the initial conditions
+
in the cylinder $  D $
 +
consists of finding solutions of this system satisfying the initial conditions
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419019.png" /></td> <td valign="top" style="width:5%;text-align:right;">(2)</td></tr></table>
+
$$ \tag{2 }
 +
u \mid  _ {\Omega _ {0}  }  = \phi ( x) ,
 +
$$
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419020.png" />, and the boundary condition
+
where $  \phi ( x) = ( \phi _ {1} ( x) \dots \phi _ {m} ( x) ) $,  
 +
and the boundary condition
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419021.png" /></td> <td valign="top" style="width:5%;text-align:right;">(3)</td></tr></table>
+
$$ \tag{3 }
 +
\left . B \left ( x ,t ,
 +
\frac \partial {\partial  x }
 +
\right )
 +
u \right | _  \Gamma  = \psi ( x , t ) ,
 +
$$
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419022.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419023.png" /> is a rectangular matrix with
+
where $  \psi ( x , t ) = ( \psi _ {1} ( x , t ) \dots \psi _ {p} ( x , t ) ) $
 +
and $  B ( x , t , \partial  / \partial  x ) $
 +
is a rectangular matrix with
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419024.png" /></td> </tr></table>
+
$$
 +
B _ {ij} \left ( x , t ,
 +
\frac \partial {\partial  x }
 +
\right )  = \
 +
\sum _ {| \alpha | \leq  q _ {i,j} }
 +
b _  \alpha  ^ {i,j} ( x , t ) D _ {x}  ^  \alpha  ,
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419025.png" /></td> </tr></table>
+
$$
 +
= 1 \dots m ; \  j  = 1 \dots p .
 +
$$
  
 
Suppose that the system is uniformly parabolic.
 
Suppose that the system is uniformly parabolic.
  
A classical solution of the mixed problem (1)–(3) is a vector function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419026.png" /> belonging to
+
A classical solution of the mixed problem (1)–(3) is a vector function $  U ( x , t ) $
 +
belonging to
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419027.png" /></td> </tr></table>
+
$$
 +
C _ {x,t}  ^ {2p,1} ( D)  \cap  C _ {x,1}  ^ {q,0}
 +
( D \cup \Gamma )  \cap  C ( D \cup \Gamma \cup \Omega
 +
bar _ {0} ) ,
 +
$$
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419028.png" /> for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419029.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419030.png" />, and satisfying (1) in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419031.png" /> and conditions (2) and (3) on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419032.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419033.png" />, respectively. Sometimes one considers more general solutions than this. In particular, one may drop the requirement that the solution be continuous at the points of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419034.png" /> and replace it by the condition that it is bounded in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419035.png" />.
+
where $  q = \max  q _ {i,j} $
 +
for $  1 \leq  i \leq  m $,  
 +
$  1 \leq  j \leq  p $,  
 +
and satisfying (1) in $  D $
 +
and conditions (2) and (3) on $  \Omega _ {0} $
 +
and $  \Gamma $,  
 +
respectively. Sometimes one considers more general solutions than this. In particular, one may drop the requirement that the solution be continuous at the points of $  \overline \Gamma \; \cap \overline \Omega \; _ {0} $
 +
and replace it by the condition that it is bounded in $  D $.
  
If the complementarity (or Lopatinskii) condition holds (and if, for the sake of simplicity, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419036.png" /> is assumed to be bounded), then for sufficiently smooth data (the coefficients in (1) and (3) and the vector functions <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419037.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419038.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419039.png" />) and under certain compatibility conditions, a classical solutions exists and is unique.
+
If the complementarity (or Lopatinskii) condition holds (and if, for the sake of simplicity, $  \Omega $
 +
is assumed to be bounded), then for sufficiently smooth data (the coefficients in (1) and (3) and the vector functions $  f $,  
 +
$  \phi $
 +
and $  \psi $)  
 +
and under certain compatibility conditions, a classical solutions exists and is unique.
  
 
The basic mixed problems for a general linear uniformly-parabolic second-order equation
 
The basic mixed problems for a general linear uniformly-parabolic second-order equation
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419040.png" /></td> <td valign="top" style="width:5%;text-align:right;">(1prm)</td></tr></table>
+
$$ \tag{1'}
 +
u _ {t} - L u  \equiv  u _ {t} -
 +
\sum _ {i , j = 1 } ^ { n }
 +
( a _ {ij} ( x , t ) u _ {x _ {i}  } ) _ {x _ {j}  } +
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419041.png" /></td> </tr></table>
+
$$
 +
- \sum _ {i = 1 } ^ { n }  b _ {i} ( x , t ) u _ {
 +
x _ {i}  } - c ( x , t ) u  = f ( x , t ) ,\  ( x , t ) \in D ,
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419042.png" /></td> </tr></table>
+
$$
 +
a _ {ij} ( x , t )  = a _ {ji} ( x , t ) ,\  i , j = 1 \dots n ,
 +
$$
  
are those of finding solutions of (1prm) that satisfy the initial condition
+
are those of finding solutions of (1'}) that satisfy the initial condition
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419043.png" /></td> <td valign="top" style="width:5%;text-align:right;">(2prm)</td></tr></table>
+
$$ \tag{2'}
 +
u \mid  _ {\Omega _ {0}  }  = \phi ( x)
 +
$$
  
 
and one of the boundary conditions
 
and one of the boundary conditions
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419044.png" /></td> <td valign="top" style="width:5%;text-align:right;">(4)</td></tr></table>
+
$$ \tag{4 }
 +
u \mid  _  \Gamma  = \psi ( x , t ) ,
 +
$$
  
 
the first mixed problem,
 
the first mixed problem,
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419045.png" /></td> <td valign="top" style="width:5%;text-align:right;">(5)</td></tr></table>
+
$$ \tag{5 }
 +
\left .  
 +
\frac{\partial  u }{\partial  \nu }
 +
\right | _  \Gamma  = \psi ( x , t ) ,
 +
$$
  
 
the second mixed problem, or
 
the second mixed problem, or
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419046.png" /></td> <td valign="top" style="width:5%;text-align:right;">(6)</td></tr></table>
+
$$ \tag{6 }
 +
\left . \left (
 +
 
 +
\frac{\partial  u }{\partial  N }
 +
+
 +
\sigma ( x , t ) u \right )
 +
\right | _  \Gamma  = \psi ( x , t ) ,
 +
$$
 +
 
 +
the third mixed problem, where  $  N $
 +
is a co-normal of the elliptic operator  $  L $.
  
the third mixed problem, where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419047.png" /> is a co-normal of the elliptic operator <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419048.png" />.
+
Each of these problems satisfies the complementarity condition and, consequently, when the data are sufficiently smooth and the compatibility conditions hold, each has a classical solution. This solution can be obtained by the method of potentials, the method of finite differences, the [[Galerkin method|Galerkin method]], or, in the case when the functions  $  a _ {i,j} $(
 +
$  i , j = 1 \dots n $),
 +
$  c $
 +
and  $  \sigma $
 +
do not depend on  $  t $
 +
and  $  b _ {i} \equiv 0 $,
 +
$  i = 1 \dots n $,
 +
by the [[Fourier method|Fourier method]]. For example, in order to solve the first mixed problem for equation (1) it is sufficient to require that the coefficients of the equation belong to the Hölder space  $  C  ^  \alpha  ( \overline{D}\; ) $
 +
for some  $  \alpha > 0 $,
 +
that the coefficients  $  a _ {i,j} ( x , t ) $
 +
have derivatives  $  \partial  a _ {i,j} / \partial  x _ {i} $
 +
in  $  C  ^  \alpha  ( \overline{D}\; ) $,
 +
$  i , j = 1 \dots n $,
 +
that  $  f ( x , t ) $
 +
belongs to  $  C  ^  \alpha  ( \overline{D}\; ) $,
 +
that  $  \phi $
 +
and  $  \psi $
 +
are continuous on  $  \overline \Omega \; _ {0} $
 +
and  $  \overline \Gamma \; $,
 +
respectively, and that  $  \phi \mid  _ {\partial  \Omega }  = \psi ( x , 0 ) $.
 +
For this it is sufficient that the boundary  $  \partial  \Omega $
 +
of  $  \Omega $
 +
satisfies the following condition: For any point  $  x  ^ {0} \in \partial  \Omega $
 +
there is a closed sphere  $  S $
 +
having a unique point in common with  $  \Omega $,
 +
namely, the point  $  x  ^ {0} $:
 +
$  S \cap \Omega = x  ^ {0} $.  
 +
Under certain conditions on the lateral surface (that it contains no characteristic points, i.e. points of contact with the planes  $  t = \textrm{ const } $),
 +
an analogous statement also holds in the case of a non-cylindrical domain  $  D $.
  
Each of these problems satisfies the complementarity condition and, consequently, when the data are sufficiently smooth and the compatibility conditions hold, each has a classical solution. This solution can be obtained by the method of potentials, the method of finite differences, the [[Galerkin method|Galerkin method]], or, in the case when the functions <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419049.png" /> (<img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419050.png" />), <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419051.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419052.png" /> do not depend on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419053.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419054.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419055.png" />, by the [[Fourier method|Fourier method]]. For example, in order to solve the first mixed problem for equation (1) it is sufficient to require that the coefficients of the equation belong to the Hölder space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419056.png" /> for some <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419057.png" />, that the coefficients <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419058.png" /> have derivatives <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419059.png" /> in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419060.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419061.png" />, that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419062.png" /> belongs to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419063.png" />, that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419064.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419065.png" /> are continuous on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419066.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419067.png" />, respectively, and that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419068.png" />. For this it is sufficient that the boundary <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419069.png" /> of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419070.png" /> satisfies the following condition: For any point <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419071.png" /> there is a closed sphere <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419072.png" /> having a unique point in common with <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419073.png" />, namely, the point <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419074.png" />: <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419075.png" />. Under certain conditions on the lateral surface (that it contains no characteristic points, i.e. points of contact with the planes <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419076.png" />), an analogous statement also holds in the case of a non-cylindrical domain <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419077.png" />.
+
Existence theorems for the basic mixed problems for equation (1'}) also hold under other conditions on the given functions and the domain  $  \Omega $.
 +
For example, in the case of the first mixed problem in a cylindrical domain  $  D $,  
 +
for the homogeneous [[Heat equation|heat equation]] with continuous functions  $  \phi $
 +
and  $  \psi $
 +
satisfying the compatibility condition  $  \phi \mid  _ {\partial  \Omega }  = \psi ( x , 0 ) $,  
 +
a solution exists provided that  $  \Omega $
 +
is such that the [[Dirichlet problem|Dirichlet problem]] for the [[Laplace equation|Laplace equation]] is solvable in $  \Omega $(
 +
there is a classical solution) for an arbitrary continuous boundary function.
  
Existence theorems for the basic mixed problems for equation (1prm) also hold under other conditions on the given functions and the domain <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419078.png" />. For example, in the case of the first mixed problem in a cylindrical domain <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419079.png" />, for the homogeneous [[Heat equation|heat equation]] with continuous functions <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419080.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419081.png" /> satisfying the compatibility condition <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419082.png" />, a solution exists provided that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419083.png" /> is such that the [[Dirichlet problem|Dirichlet problem]] for the [[Laplace equation|Laplace equation]] is solvable in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419084.png" /> (there is a classical solution) for an arbitrary continuous boundary function.
+
Let the coefficients  $  a _ {ij} $,
 +
$  b _ {i} $
 +
and  $  c $
 +
be measurable and bounded in  $  D $,
 +
and let  $  \sigma $
 +
be measurable and bounded on  $  \Gamma $.
 +
Further, let  $  f \in L _ {2} ( D) $,
 +
$  \phi \in L _ {0} ( \Omega ) $
 +
and, in the case of the first mixed problem, let  $  \psi $
 +
be the trace on  $  \Gamma $
 +
of some function from the [[Sobolev space|Sobolev space]] $  W _ {2}  ^ {1,0} ( D) $,
 +
while in the case of the third (or second) mixed problem, let  $  \psi $
 +
belong to  $  L _ {2} ( \Gamma ) $.
  
Let the coefficients <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419085.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419086.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419087.png" /> be measurable and bounded in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419088.png" />, and let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419089.png" /> be measurable and bounded on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419090.png" />. Further, let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419091.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419092.png" /> and, in the case of the first mixed problem, let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419093.png" /> be the trace on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419094.png" /> of some function from the [[Sobolev space|Sobolev space]] <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419095.png" />, while in the case of the third (or second) mixed problem, let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419096.png" /> belong to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419097.png" />.
+
A function  $  u ( x , t ) $
 +
belonging to  $  W _ {2}  ^ {1,0} ( D) $
 +
and with trace on $  \Gamma $
 +
equal to  $  \psi $:  
 +
$  u \mid  _  \Gamma  = \psi $,  
 +
is called a generalized solution of the first mixed problem (1'}), (2'}), (4) if it satisfies the integral identity
  
A function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419098.png" /> belonging to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m06419099.png" /> and with trace on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190100.png" /> equal to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190101.png" />: <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190102.png" />, is called a generalized solution of the first mixed problem (1prm), (2prm), (4) if it satisfies the integral identity
+
$$
 +
\int\limits _ { D } \left [ - u v _ {t} +
 +
\sum _ {i , j = 1 } ^ { n }
 +
a _ {i,j} u _ {x _ {i}  } v _ {x _ {j}  }
 +
- \left ( \sum _ { i= } 1 ^ { n }  b _ {i} u _ {x _ {i}  } +
 +
c u \right ) v \right ]  d x  d t =
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190103.png" /></td> </tr></table>
+
$$
 +
= \
 +
\int\limits _ { D } f v  d x  d t + \int\limits _ {\Omega _ {0} } \phi v  d x
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190104.png" /></td> </tr></table>
+
for all  $  v $
 +
in the Sobolev space  $  W _ {2}  ^ {1} ( D) $
 +
for which  $  v \mid  _  \Gamma  = 0 $,
 +
$  v \mid  _ {\Omega _ {T}  } = 0 $.
  
for all <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190105.png" /> in the Sobolev space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190106.png" /> for which <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190107.png" />, <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190108.png" />.
+
A function  $  u ( x , t ) $
 +
belonging to  $  W _ {2}  ^ {1,0} ( D) $
 +
is called a generalized solution of the third (second, if  $  \sigma \equiv 0 $)
 +
mixed problem (1), (2), (6) if it satisfies the integral identity
  
A function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190109.png" /> belonging to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190110.png" /> is called a generalized solution of the third (second, if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190111.png" />) mixed problem (1), (2), (6) if it satisfies the integral identity
+
$$
 +
\int\limits _ { D } \left [ - u v _ {t} +
 +
\sum _ {i , j = 1 } ^ { n }
 +
\alpha _ {i,j} u _ {x _ {i}  }
 +
v _ {x _ {j}  }
 +
- \left ( \sum _ { i= } 1 ^ { n }
 +
b _ {i} u _ {x _ {i}  } + c u \right )
 +
v \right ]  d x  d t +
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190112.png" /></td> </tr></table>
+
$$
 +
+
 +
\int\limits _  \Gamma  \sigma u v  d S =
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190113.png" /></td> </tr></table>
+
$$
 +
= \
 +
\int\limits _ { D } f v  d x  d t + \int\limits _ {\Omega _ {0} } \phi v  d x + \int\limits _  \Gamma  \psi v  d S
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190114.png" /></td> </tr></table>
+
for all  $  v $
 +
in  $  W _ {2}  ^ {1} $
 +
such that  $  v \mid  _ {\Omega _ {T}  } = 0 $.
  
for all <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190115.png" /> in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190116.png" /> such that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190117.png" />.
+
A generalized solution of each of these problems exists and is unique; moreover, if  $  f \in L _ {p} ( D) $
 +
is continuous in  $  D $,
 +
for sufficiently large  $  p $,
 +
then it even satisfies a Hölder condition for some exponent  $  \alpha > 0 $.  
 +
By increasing the smoothness of the given functions and the boundary of the domain subject to the compatibility conditions, the smoothness of the generalized solution increases. For example, consider the heat equation with  $  \phi \equiv 0 $
 +
and  $  \psi \equiv 0 $,
 +
and let  $  \partial  \Omega $
 +
be a sufficiently smooth surface. Then the generalized solution of the first mixed problem belongs to  $  W _ {2} ^ {2 ( s + 1 ) , s + 1 } ( D) $,
 +
provided that $  f \in W _ {2} ^ {2 s , s } ( D) $
 +
and the compatibility conditions
  
A generalized solution of each of these problems exists and is unique; moreover, if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190118.png" /> is continuous in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190119.png" />, for sufficiently large <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190120.png" />, then it even satisfies a Hölder condition for some exponent <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190121.png" />. By increasing the smoothness of the given functions and the boundary of the domain subject to the compatibility conditions, the smoothness of the generalized solution increases. For example, consider the heat equation with <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190122.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190123.png" />, and let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190124.png" /> be a sufficiently smooth surface. Then the generalized solution of the first mixed problem belongs to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190125.png" />, provided that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190126.png" /> and the compatibility conditions
+
$$ \tag{7 }
 +
f  \mid  _ {\partial  \Omega _ {0}  }  = \
 +
( \Delta f + f _ {t} ) \mid  _ {\partial  \Omega _ {0}  }  = \dots = \
 +
\sum _ {i = 0 } ^ { {s }  - 1 }
 +
\Delta  ^ {i}
 +
\left .  
 +
\frac{\partial  ^ {s - 1 - i } f }{\partial  t ^ {s - 1 - i } }
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190127.png" /></td> <td valign="top" style="width:5%;text-align:right;">(7)</td></tr></table>
+
\right | _ {\partial  \Omega _ {0}  }
 +
$$
  
 
hold.
 
hold.
  
In particular, if <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190128.png" />, then the solution belongs to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190129.png" /> when <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190130.png" />, it satisfies the heat equation and its trace on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190131.png" /> is equal to zero. If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190132.png" /> for sufficiently large <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190133.png" /> and the compatibility conditions (7) hold, then by virtue of the [[Imbedding theorems|imbedding theorems]], the generalized solution is classical. An analogous statement holds for generalized solutions of the basic mixed problems for equation (1prm) when the coefficients are sufficiently smooth.
+
In particular, if $  f \in L _ {2} ( D) $,  
 +
then the solution belongs to $  W _ {2}  ^ {2,1} ( D) $
 +
when $  ( x , t ) \in D $,  
 +
it satisfies the heat equation and its trace on $  \Omega _ {0} $
 +
is equal to zero. If $  f \in W _ {2}  ^ {2s,s} $
 +
for sufficiently large $  s $
 +
and the compatibility conditions (7) hold, then by virtue of the [[Imbedding theorems|imbedding theorems]], the generalized solution is classical. An analogous statement holds for generalized solutions of the basic mixed problems for equation (1'}) when the coefficients are sufficiently smooth.
 +
 
 +
Let  $  \Omega = \mathbf R  ^ {n} $.
 +
The problem of finding a solution in the strip  $  D = \mathbf R  ^ {n} \times ( 0 , T ) $
 +
for the parabolic system (1) satisfying the initial condition (2) on the characteristic  $  \Omega _ {0} = \{ x \in \mathbf R  ^ {n} ,  t = 0 \} $
 +
is called the [[Cauchy problem|Cauchy problem]] for (1). A classical solution of the Cauchy problem (1), (2) is a vector function  $  u ( x , t ) $
 +
belonging to  $  C  ^ {2p,1} ( D) \cap C ( D \cup \Omega _ {0} ) $
 +
and satisfying (1) in  $  D $
 +
and (2) on  $  \Omega $.
 +
If the right-hand side  $  f ( x , t ) $
 +
belongs to the Hölder space  $  C  ^  \alpha  ( \overline{D}\; ) $
 +
for some  $  \alpha > 0 $,
 +
and the coefficients are sufficiently smooth in  $  \overline{D}\; $(
 +
they and their derivatives are bounded), then for any bounded continuous initial vector function  $  \phi ( x) $
 +
on  $  \mathbf R  ^ {n} $
 +
there is a bounded solution of the Cauchy problem on  $  D $,
 +
and this bounded solution is unique.
 +
 
 +
The condition of boundedness can be replaced by the condition of  "not too rapid growth" . For example, the following is true for second-order equations. Let the coefficients of equation (1'}),
  
Let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190134.png" />. The problem of finding a solution in the strip <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190135.png" /> for the parabolic system (1) satisfying the initial condition (2) on the characteristic <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190136.png" /> is called the [[Cauchy problem|Cauchy problem]] for (1). A classical solution of the Cauchy problem (1), (2) is a vector function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190137.png" /> belonging to <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190138.png" /> and satisfying (1) in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190139.png" /> and (2) on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190140.png" />. If the right-hand side <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190141.png" /> belongs to the Hölder space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190142.png" /> for some <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190143.png" />, and the coefficients are sufficiently smooth in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190144.png" /> (they and their derivatives are bounded), then for any bounded continuous initial vector function <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190145.png" /> on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190146.png" /> there is a bounded solution of the Cauchy problem on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190147.png" />, and this bounded solution is unique.
+
$$
 +
a _ {i,j} ( x , t ) ,\
 +
b _ {i} ( x , t ) ,\
 +
c ( x , t ) \  \textrm{ and } \ \
  
The condition of boundedness can be replaced by the condition of "not too rapid growth" . For example, the following is true for second-order equations. Let the coefficients of equation (1prm),
+
\frac{\partial a _ {i,j} ( x , t ) }{\partial  x _ {i} }
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190148.png" /></td> </tr></table>
+
$$
  
belong to the Hölder space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190149.png" /> for some <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190150.png" />. Further, let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190151.png" /> be continuous in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190152.png" /> and let <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190153.png" /> be continuous in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190154.png" />, locally Hölder continuous in <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190155.png" /> uniformly for <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190156.png" /> (for some exponent <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190157.png" />), and such that
+
belong to the Hölder space $  C  ^  \alpha  ( \overline{D}\; ) $
 +
for some $  \alpha > 0 $.  
 +
Further, let $  \phi ( x) $
 +
be continuous in $  \mathbf R  ^ {n} $
 +
and let $  f ( x , t ) $
 +
be continuous in $  \overline{D}\; $,  
 +
locally Hölder continuous in $  x $
 +
uniformly for $  t \in [ 0 , T ] $(
 +
for some exponent $  \alpha > 0 $),  
 +
and such that
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190158.png" /></td> </tr></table>
+
$$
 +
| \phi ( x) |  \leq  C e ^ {h | x |  ^ {2} } ,\ \
 +
x \in \mathbf R  ^ {n} ,
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190159.png" /></td> </tr></table>
+
$$
 +
| f ( x , t ) |  \leq  C e ^ {h | x |  ^ {2} } ,\  ( x , t ) \in D .
 +
$$
  
Then for sufficiently small <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190160.png" /> (depending on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190161.png" />), there is a solution of the Cauchy problem (1prm), (2prm) in the strip <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190162.png" />. It can be written in the form
+
Then for sufficiently small $  T $(
 +
depending on $  h $),  
 +
there is a solution of the Cauchy problem (1'}), (2'}) in the strip $  D = \mathbf R  ^ {n} \times ( 0 , T ) $.  
 +
It can be written in the form
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190163.png" /></td> </tr></table>
+
$$
 +
u ( x , t )  = \int\limits _ {\mathbf R  ^ {n} }
 +
\Gamma ( x , t ; \xi , 0 ) \phi ( \xi )  d \xi +
 +
$$
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190164.png" /></td> </tr></table>
+
$$
 +
+
 +
\int\limits _ { 0 } ^ { t }  \int\limits _ {\mathbf R  ^ {n} } \Gamma ( x , t
 +
; \xi , \tau ) f ( \xi , \tau )  d \xi  d \tau ,
 +
$$
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190165.png" /> is a fundamental solution of (1prm), and satisfies the estimate
+
where $  \Gamma ( x , t ;  \xi , \tau ) $
 +
is a fundamental solution of (1'}), and satisfies the estimate
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190166.png" /></td> <td valign="top" style="width:5%;text-align:right;">(8)</td></tr></table>
+
$$ \tag{8 }
 +
| u ( x , t ) |  \leq  C _ {1} e ^ {h | x |  ^ {2} }
 +
$$
  
for some positive constants <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190167.png" /> and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190168.png" />. Condition (8) guarantees the uniqueness of the solution of the Cauchy problem.
+
for some positive constants $  C _ {1} $
 +
and $  k $.  
 +
Condition (8) guarantees the uniqueness of the solution of the Cauchy problem.
  
 
In the case of an equation with constant coefficients it is possible to find a condition of type (8) on the growth of the solution that is necessary and sufficient for its uniqueness. For example, for a solution of the Cauchy problem for the heat equation to be unique in the class of functions satisfying the inequality
 
In the case of an equation with constant coefficients it is possible to find a condition of type (8) on the growth of the solution that is necessary and sufficient for its uniqueness. For example, for a solution of the Cauchy problem for the heat equation to be unique in the class of functions satisfying the inequality
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190169.png" /></td> </tr></table>
+
$$
 +
| u ( x , t ) |  \leq  C e ^ {| x | h ( | x | ) } ,
 +
$$
  
where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190170.png" /> is a positive continuous function on <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190171.png" />, it is necessary and sufficient that the integral <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190172.png" /> diverges.
+
where $  h ( | x | ) $
 +
is a positive continuous function on $  [ 0 , \infty ) $,
 +
it is necessary and sufficient that the integral $  \int _ {0}  ^  \infty  dr / h( r) $
 +
diverges.
  
For parabolic equations one can also consider problems without initial conditions (the Fourier problem). For example, one can ask for the solution of the homogeneous heat equation in the cylinder <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190173.png" />, where <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190174.png" /> is a bounded domain with sufficiently smooth boundary <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190175.png" />, satisfying the boundary condition
+
For parabolic equations one can also consider problems without initial conditions (the Fourier problem). For example, one can ask for the solution of the homogeneous heat equation in the cylinder $  D = \{ x \in \Omega,  - \infty < t < + \infty \} $,  
 +
where $  \Omega $
 +
is a bounded domain with sufficiently smooth boundary $  \partial  \Omega $,  
 +
satisfying the boundary condition
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190176.png" /></td> </tr></table>
+
$$
 +
u ( x , t ) \mid  _ {x \in \partial  \Omega }  = \psi ( x , t ) .
 +
$$
  
If <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190177.png" /> is continuous and bounded, then there is a bounded solution of the Fourier problem, and this is the unique bounded solution.
+
If $  \psi $
 +
is continuous and bounded, then there is a bounded solution of the Fourier problem, and this is the unique bounded solution.
  
For parabolic equations and systems it is also possible to consider the first mixed problem in a non-cylindrical domain <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190178.png" /> in the case when the lateral surface contains characteristic points (points of contact with the planes <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190179.png" />). In particular, it is possible to consider the [[Dirichlet problem|Dirichlet problem]], where boundary conditions are given on the entire boundary <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190180.png" />. Under specific conditions on the set of characteristic points and on the order of contact of the characteristic points of <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190181.png" /> with a characteristic plane, the Dirichlet problem has a unique solution (in the space <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190182.png" />). For example, suppose (for the sake of simplicity) that <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190183.png" /> is a strictly-convex domain and that the equation of the boundary in a neighbourhood of the upper characteristic point <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190184.png" /> has the form <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190185.png" /> when <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190186.png" />, and <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190187.png" /> when <img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190188.png" /> (<img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190189.png" />). Then the divergence of both integrals
+
For parabolic equations and systems it is also possible to consider the first mixed problem in a non-cylindrical domain $  D $
 +
in the case when the lateral surface contains characteristic points (points of contact with the planes $  t = \textrm{ const } $).  
 +
In particular, it is possible to consider the [[Dirichlet problem|Dirichlet problem]], where boundary conditions are given on the entire boundary $  \partial  \Omega $.  
 +
Under specific conditions on the set of characteristic points and on the order of contact of the characteristic points of $  \partial  \Omega $
 +
with a characteristic plane, the Dirichlet problem has a unique solution (in the space $  W _ {2} ^ {B  ^ {0} } $).  
 +
For example, suppose (for the sake of simplicity) that $  D \subset  \mathbf R  ^ {2} $
 +
is a strictly-convex domain and that the equation of the boundary in a neighbourhood of the upper characteristic point $  ( x  ^ {0} , t  ^ {0} ) $
 +
has the form $  x - x  ^ {0} = \phi _ {1} ( t) $
 +
when $  x \leq  x  ^ {0} $,  
 +
and $  x - x  ^ {0} = \phi _ {2} ( t) $
 +
when $  x \geq  x  ^ {0} $(
 +
$  t  ^ {0} - \delta < t \leq  t  ^ {0} $).  
 +
Then the divergence of both integrals
  
<table class="eq" style="width:100%;"> <tr><td valign="top" style="width:94%;text-align:center;"><img align="absmiddle" border="0" src="https://www.encyclopediaofmath.org/legacyimages/m/m064/m064190/m064190190.png" /></td> </tr></table>
+
$$
 +
\int\limits _ {t  ^ {0} -
 +
\delta } ^ { {t  ^ {0}} } | \phi _ {i} ( t)  ^ {-} 2p |  dt ,\ \
 +
i = 1 , 2 ,
 +
$$
  
 
guarantees the existence and uniqueness of a solution of the Dirichlet problem for a second-order parabolic equation. This condition is also necessary in this class of equations.
 
guarantees the existence and uniqueness of a solution of the Dirichlet problem for a second-order parabolic equation. This condition is also necessary in this class of equations.
Line 137: Line 415:
 
====References====
 
====References====
 
<table><TR><TD valign="top">[1]</TD> <TD valign="top">  V.S. Vladimirov,  "Equations of mathematical physics" , MIR  (1984)  (Translated from Russian)</TD></TR><TR><TD valign="top">[2]</TD> <TD valign="top">  V.A. Il'in,  "The solvability of mixed problems for hyperbolic and parabolic equations"  ''Russian Math. Surveys'' , '''15''' :  2  (1960)  pp. 85–142  ''Uspekhi Mat. Nauk'' , '''15''' :  2  (1960)  pp. 97–154</TD></TR><TR><TD valign="top">[3]</TD> <TD valign="top">  A.M. Il'in,  A.S. Kalashnikov,  O.A. Oleinik,  "Linear equations of the second order of parabolic type"  ''Russian Math. Surveys'' , '''17''' :  3  (1962)  pp. 1–143  ''Uspekhi Mat. Nauk'' , '''17''' :  3  (1962)  pp. 3–146</TD></TR><TR><TD valign="top">[4]</TD> <TD valign="top">  S.N. Kruzhkov,  "A priori bounds and some properties of solutions of elliptic and parabolic equations"  ''Mat. Sb.'' , '''65''' :  4  (1964)  pp. 522–570  (In Russian)</TD></TR><TR><TD valign="top">[5]</TD> <TD valign="top">  O.A. Ladyzhenskaya,  V.A. Solonnikov,  N.N. Ural'tseva,  "Linear and quasi-linear equations of parabolic type" , Amer. Math. Soc.  (1968)  (Translated from Russian)</TD></TR><TR><TD valign="top">[6]</TD> <TD valign="top">  O.A. Ladyzhenskaya,  "The boundary value problems of mathematical physics" , Springer  (1985)  (Translated from Russian)</TD></TR><TR><TD valign="top">[7]</TD> <TD valign="top">  O.A. Ladyzhenskaya,  "On uniqueness of a solution of the Cauchy problem for a linear parabolic equation"  ''Mat. Sb.'' , '''27''' :  2  (1950)  pp. 175–184  (In Russian)</TD></TR><TR><TD valign="top">[8]</TD> <TD valign="top">  V.P. Mikhailov,  "The Dirichlet problem for a parabolic equation I"  ''Mat. Sb.'' , '''61''' :  1  (1963)  pp. 40–64  (In Russian)</TD></TR><TR><TD valign="top">[9]</TD> <TD valign="top">  V.P. Mikhailov,  "The Dirichlet problem for a parabolic equation II"  ''Mat. Sb.'' , '''62''' :  2  (1963)  pp. 140–159  (In Russian)</TD></TR><TR><TD valign="top">[10]</TD> <TD valign="top">  J. Nash,  "Continuity of solutions of parabolic and elliptic equations"  ''Amer. J. Math.'' , '''80'''  (1958)  pp. 931–954</TD></TR><TR><TD valign="top">[11]</TD> <TD valign="top">  I.G. Petrovskii,  "Partial differential equations" , Saunders  (1967)  (Translated from Russian)</TD></TR><TR><TD valign="top">[12]</TD> <TD valign="top">  I.G. Petrovskii,  "On Cauchy's problem for systems of linear partial differential equations in a nonanalytic function domain"  ''Bull. Moskov. Gos. Univ. Ser. A.'' , '''1''' :  7  (1938)  pp. 1–72  (In Russian)</TD></TR><TR><TD valign="top">[13]</TD> <TD valign="top">  I.G. Petrovskii,  "Zur ersten Randwertaufgabe der Wärmeleitungsgleichung"  ''Comp. Math.'' , '''1'''  (1934)  pp. 383–419</TD></TR><TR><TD valign="top">[14]</TD> <TD valign="top">  S.L. Sobolev,  "Partial differential equations of mathematical physics" , Pergamon  (1964)  (Translated from Russian)</TD></TR><TR><TD valign="top">[15]</TD> <TD valign="top">  V.A. Solonnikov,  "Boundary value problems of mathematical physics III"  ''Proc. Steklov Inst. Math.'' , '''83'''  (1965)  ''Trudy Mat. Inst. Steklov'' , '''83'''  (1965)</TD></TR><TR><TD valign="top">[16]</TD> <TD valign="top">  A.N. [A.N. Tikhonov] Tichonoff,  A.A. Samarskii,  "Differentialgleichungen der mathematischen Physik" , Deutsch. Verlag Wissenschaft.  (1959)  (Translated from Russian)</TD></TR><TR><TD valign="top">[17]</TD> <TD valign="top">  A.N. Tikhonov,  ''Byull. Moskov. Univ. (A)'' , '''1''' :  9  (1938)  pp. 1–43</TD></TR><TR><TD valign="top">[18]</TD> <TD valign="top">  A.N. Tikhonov,  ''Mat. Sb.'' , '''42''' :  2  (1935)  pp. 199–216</TD></TR><TR><TD valign="top">[19]</TD> <TD valign="top">  A. Friedman,  "Partial differential equations of parabolic type" , Prentice-Hall  (1964)</TD></TR><TR><TD valign="top">[20]</TD> <TD valign="top">  S.D. Eidel'man,  "Parabolic systems" , North-Holland  (1969)  (Translated from Russian)</TD></TR></table>
 
<table><TR><TD valign="top">[1]</TD> <TD valign="top">  V.S. Vladimirov,  "Equations of mathematical physics" , MIR  (1984)  (Translated from Russian)</TD></TR><TR><TD valign="top">[2]</TD> <TD valign="top">  V.A. Il'in,  "The solvability of mixed problems for hyperbolic and parabolic equations"  ''Russian Math. Surveys'' , '''15''' :  2  (1960)  pp. 85–142  ''Uspekhi Mat. Nauk'' , '''15''' :  2  (1960)  pp. 97–154</TD></TR><TR><TD valign="top">[3]</TD> <TD valign="top">  A.M. Il'in,  A.S. Kalashnikov,  O.A. Oleinik,  "Linear equations of the second order of parabolic type"  ''Russian Math. Surveys'' , '''17''' :  3  (1962)  pp. 1–143  ''Uspekhi Mat. Nauk'' , '''17''' :  3  (1962)  pp. 3–146</TD></TR><TR><TD valign="top">[4]</TD> <TD valign="top">  S.N. Kruzhkov,  "A priori bounds and some properties of solutions of elliptic and parabolic equations"  ''Mat. Sb.'' , '''65''' :  4  (1964)  pp. 522–570  (In Russian)</TD></TR><TR><TD valign="top">[5]</TD> <TD valign="top">  O.A. Ladyzhenskaya,  V.A. Solonnikov,  N.N. Ural'tseva,  "Linear and quasi-linear equations of parabolic type" , Amer. Math. Soc.  (1968)  (Translated from Russian)</TD></TR><TR><TD valign="top">[6]</TD> <TD valign="top">  O.A. Ladyzhenskaya,  "The boundary value problems of mathematical physics" , Springer  (1985)  (Translated from Russian)</TD></TR><TR><TD valign="top">[7]</TD> <TD valign="top">  O.A. Ladyzhenskaya,  "On uniqueness of a solution of the Cauchy problem for a linear parabolic equation"  ''Mat. Sb.'' , '''27''' :  2  (1950)  pp. 175–184  (In Russian)</TD></TR><TR><TD valign="top">[8]</TD> <TD valign="top">  V.P. Mikhailov,  "The Dirichlet problem for a parabolic equation I"  ''Mat. Sb.'' , '''61''' :  1  (1963)  pp. 40–64  (In Russian)</TD></TR><TR><TD valign="top">[9]</TD> <TD valign="top">  V.P. Mikhailov,  "The Dirichlet problem for a parabolic equation II"  ''Mat. Sb.'' , '''62''' :  2  (1963)  pp. 140–159  (In Russian)</TD></TR><TR><TD valign="top">[10]</TD> <TD valign="top">  J. Nash,  "Continuity of solutions of parabolic and elliptic equations"  ''Amer. J. Math.'' , '''80'''  (1958)  pp. 931–954</TD></TR><TR><TD valign="top">[11]</TD> <TD valign="top">  I.G. Petrovskii,  "Partial differential equations" , Saunders  (1967)  (Translated from Russian)</TD></TR><TR><TD valign="top">[12]</TD> <TD valign="top">  I.G. Petrovskii,  "On Cauchy's problem for systems of linear partial differential equations in a nonanalytic function domain"  ''Bull. Moskov. Gos. Univ. Ser. A.'' , '''1''' :  7  (1938)  pp. 1–72  (In Russian)</TD></TR><TR><TD valign="top">[13]</TD> <TD valign="top">  I.G. Petrovskii,  "Zur ersten Randwertaufgabe der Wärmeleitungsgleichung"  ''Comp. Math.'' , '''1'''  (1934)  pp. 383–419</TD></TR><TR><TD valign="top">[14]</TD> <TD valign="top">  S.L. Sobolev,  "Partial differential equations of mathematical physics" , Pergamon  (1964)  (Translated from Russian)</TD></TR><TR><TD valign="top">[15]</TD> <TD valign="top">  V.A. Solonnikov,  "Boundary value problems of mathematical physics III"  ''Proc. Steklov Inst. Math.'' , '''83'''  (1965)  ''Trudy Mat. Inst. Steklov'' , '''83'''  (1965)</TD></TR><TR><TD valign="top">[16]</TD> <TD valign="top">  A.N. [A.N. Tikhonov] Tichonoff,  A.A. Samarskii,  "Differentialgleichungen der mathematischen Physik" , Deutsch. Verlag Wissenschaft.  (1959)  (Translated from Russian)</TD></TR><TR><TD valign="top">[17]</TD> <TD valign="top">  A.N. Tikhonov,  ''Byull. Moskov. Univ. (A)'' , '''1''' :  9  (1938)  pp. 1–43</TD></TR><TR><TD valign="top">[18]</TD> <TD valign="top">  A.N. Tikhonov,  ''Mat. Sb.'' , '''42''' :  2  (1935)  pp. 199–216</TD></TR><TR><TD valign="top">[19]</TD> <TD valign="top">  A. Friedman,  "Partial differential equations of parabolic type" , Prentice-Hall  (1964)</TD></TR><TR><TD valign="top">[20]</TD> <TD valign="top">  S.D. Eidel'man,  "Parabolic systems" , North-Holland  (1969)  (Translated from Russian)</TD></TR></table>
 
 
  
 
====Comments====
 
====Comments====
In the current literature the initial boundary value problems (1prm), (2prm), (4)–(6) are not referred to as  "mixed" . Sometimes expressions like Cauchy–Dirichlet or Cauchy–Neumann are used. Quite often, by a problem with Dirichlet data for a parabolic equation is meant a problem in which such data are prescribed on the parabolic boundary. Besides the first, second and third kind of boundary data, higher-order problems are also considered. For further comments and more references see [[Linear parabolic partial differential equation and system|Linear parabolic partial differential equation and system]].
+
In the current literature the initial boundary value problems (1'}), (2'}), (4)–(6) are not referred to as  "mixed" . Sometimes expressions like Cauchy–Dirichlet or Cauchy–Neumann are used. Quite often, by a problem with Dirichlet data for a parabolic equation is meant a problem in which such data are prescribed on the parabolic boundary. Besides the first, second and third kind of boundary data, higher-order problems are also considered. For further comments and more references see [[Linear parabolic partial differential equation and system|Linear parabolic partial differential equation and system]].

Latest revision as of 08:01, 6 June 2020


Problems of finding solutions

$$ u ( x , t ) = ( u _ {1} ( x , t ) \dots u _ {m} ( x ,\ t ) ) $$

in a domain $ D $ of a Euclidean space $ \mathbf R ^ {n+} 1 $( with points $ ( x , t ) = ( x _ {1} \dots x _ {n} , t ) $) of a parabolic system of equations or, when $ m = 1 $, of a parabolic equation satisfying additional conditions on some part of the boundary $ \partial D $ of the domain $ D $.

Let $ \Omega $ be a domain in $ \mathbf R ^ {n} $ with sufficiently smooth boundary $ \partial \Omega $ and let $ D $ be a cylinder $ \{ {x \in \Omega } : {0 < t < T } \} $ with lateral surface $ \Gamma = \{ {x \in \partial \Omega } : {0 < t < T } \} $, lower base $ \Omega _ {0} = \{ {x \in \Omega } : {t = 0 } \} $ and upper base $ \Omega _ {T} = \{ {x \in \Omega } : {t = T } \} $. The mixed Petrovskii problem for a linear parabolic system

$$ \tag{1 } u _ {t} + \sum _ {| \alpha | \leq 2 p } A _ \alpha ( x , t ) D _ {x} ^ \alpha u = f ( x , t ) ,\ ( x , t ) \in D , $$

$$ f ( x , t ) = ( f _ {1} ( x , t ) \dots f _ {m} ( x , t ) ) , $$

in the cylinder $ D $ consists of finding solutions of this system satisfying the initial conditions

$$ \tag{2 } u \mid _ {\Omega _ {0} } = \phi ( x) , $$

where $ \phi ( x) = ( \phi _ {1} ( x) \dots \phi _ {m} ( x) ) $, and the boundary condition

$$ \tag{3 } \left . B \left ( x ,t , \frac \partial {\partial x } \right ) u \right | _ \Gamma = \psi ( x , t ) , $$

where $ \psi ( x , t ) = ( \psi _ {1} ( x , t ) \dots \psi _ {p} ( x , t ) ) $ and $ B ( x , t , \partial / \partial x ) $ is a rectangular matrix with

$$ B _ {ij} \left ( x , t , \frac \partial {\partial x } \right ) = \ \sum _ {| \alpha | \leq q _ {i,j} } b _ \alpha ^ {i,j} ( x , t ) D _ {x} ^ \alpha , $$

$$ i = 1 \dots m ; \ j = 1 \dots p . $$

Suppose that the system is uniformly parabolic.

A classical solution of the mixed problem (1)–(3) is a vector function $ U ( x , t ) $ belonging to

$$ C _ {x,t} ^ {2p,1} ( D) \cap C _ {x,1} ^ {q,0} ( D \cup \Gamma ) \cap C ( D \cup \Gamma \cup \Omega bar _ {0} ) , $$

where $ q = \max q _ {i,j} $ for $ 1 \leq i \leq m $, $ 1 \leq j \leq p $, and satisfying (1) in $ D $ and conditions (2) and (3) on $ \Omega _ {0} $ and $ \Gamma $, respectively. Sometimes one considers more general solutions than this. In particular, one may drop the requirement that the solution be continuous at the points of $ \overline \Gamma \; \cap \overline \Omega \; _ {0} $ and replace it by the condition that it is bounded in $ D $.

If the complementarity (or Lopatinskii) condition holds (and if, for the sake of simplicity, $ \Omega $ is assumed to be bounded), then for sufficiently smooth data (the coefficients in (1) and (3) and the vector functions $ f $, $ \phi $ and $ \psi $) and under certain compatibility conditions, a classical solutions exists and is unique.

The basic mixed problems for a general linear uniformly-parabolic second-order equation

$$ \tag{1'} u _ {t} - L u \equiv u _ {t} - \sum _ {i , j = 1 } ^ { n } ( a _ {ij} ( x , t ) u _ {x _ {i} } ) _ {x _ {j} } + $$

$$ - \sum _ {i = 1 } ^ { n } b _ {i} ( x , t ) u _ { x _ {i} } - c ( x , t ) u = f ( x , t ) ,\ ( x , t ) \in D , $$

$$ a _ {ij} ( x , t ) = a _ {ji} ( x , t ) ,\ i , j = 1 \dots n , $$

are those of finding solutions of (1'}) that satisfy the initial condition

$$ \tag{2'} u \mid _ {\Omega _ {0} } = \phi ( x) $$

and one of the boundary conditions

$$ \tag{4 } u \mid _ \Gamma = \psi ( x , t ) , $$

the first mixed problem,

$$ \tag{5 } \left . \frac{\partial u }{\partial \nu } \right | _ \Gamma = \psi ( x , t ) , $$

the second mixed problem, or

$$ \tag{6 } \left . \left ( \frac{\partial u }{\partial N } + \sigma ( x , t ) u \right ) \right | _ \Gamma = \psi ( x , t ) , $$

the third mixed problem, where $ N $ is a co-normal of the elliptic operator $ L $.

Each of these problems satisfies the complementarity condition and, consequently, when the data are sufficiently smooth and the compatibility conditions hold, each has a classical solution. This solution can be obtained by the method of potentials, the method of finite differences, the Galerkin method, or, in the case when the functions $ a _ {i,j} $( $ i , j = 1 \dots n $), $ c $ and $ \sigma $ do not depend on $ t $ and $ b _ {i} \equiv 0 $, $ i = 1 \dots n $, by the Fourier method. For example, in order to solve the first mixed problem for equation (1) it is sufficient to require that the coefficients of the equation belong to the Hölder space $ C ^ \alpha ( \overline{D}\; ) $ for some $ \alpha > 0 $, that the coefficients $ a _ {i,j} ( x , t ) $ have derivatives $ \partial a _ {i,j} / \partial x _ {i} $ in $ C ^ \alpha ( \overline{D}\; ) $, $ i , j = 1 \dots n $, that $ f ( x , t ) $ belongs to $ C ^ \alpha ( \overline{D}\; ) $, that $ \phi $ and $ \psi $ are continuous on $ \overline \Omega \; _ {0} $ and $ \overline \Gamma \; $, respectively, and that $ \phi \mid _ {\partial \Omega } = \psi ( x , 0 ) $. For this it is sufficient that the boundary $ \partial \Omega $ of $ \Omega $ satisfies the following condition: For any point $ x ^ {0} \in \partial \Omega $ there is a closed sphere $ S $ having a unique point in common with $ \Omega $, namely, the point $ x ^ {0} $: $ S \cap \Omega = x ^ {0} $. Under certain conditions on the lateral surface (that it contains no characteristic points, i.e. points of contact with the planes $ t = \textrm{ const } $), an analogous statement also holds in the case of a non-cylindrical domain $ D $.

Existence theorems for the basic mixed problems for equation (1'}) also hold under other conditions on the given functions and the domain $ \Omega $. For example, in the case of the first mixed problem in a cylindrical domain $ D $, for the homogeneous heat equation with continuous functions $ \phi $ and $ \psi $ satisfying the compatibility condition $ \phi \mid _ {\partial \Omega } = \psi ( x , 0 ) $, a solution exists provided that $ \Omega $ is such that the Dirichlet problem for the Laplace equation is solvable in $ \Omega $( there is a classical solution) for an arbitrary continuous boundary function.

Let the coefficients $ a _ {ij} $, $ b _ {i} $ and $ c $ be measurable and bounded in $ D $, and let $ \sigma $ be measurable and bounded on $ \Gamma $. Further, let $ f \in L _ {2} ( D) $, $ \phi \in L _ {0} ( \Omega ) $ and, in the case of the first mixed problem, let $ \psi $ be the trace on $ \Gamma $ of some function from the Sobolev space $ W _ {2} ^ {1,0} ( D) $, while in the case of the third (or second) mixed problem, let $ \psi $ belong to $ L _ {2} ( \Gamma ) $.

A function $ u ( x , t ) $ belonging to $ W _ {2} ^ {1,0} ( D) $ and with trace on $ \Gamma $ equal to $ \psi $: $ u \mid _ \Gamma = \psi $, is called a generalized solution of the first mixed problem (1'}), (2'}), (4) if it satisfies the integral identity

$$ \int\limits _ { D } \left [ - u v _ {t} + \sum _ {i , j = 1 } ^ { n } a _ {i,j} u _ {x _ {i} } v _ {x _ {j} } - \left ( \sum _ { i= } 1 ^ { n } b _ {i} u _ {x _ {i} } + c u \right ) v \right ] d x d t = $$

$$ = \ \int\limits _ { D } f v d x d t + \int\limits _ {\Omega _ {0} } \phi v d x $$

for all $ v $ in the Sobolev space $ W _ {2} ^ {1} ( D) $ for which $ v \mid _ \Gamma = 0 $, $ v \mid _ {\Omega _ {T} } = 0 $.

A function $ u ( x , t ) $ belonging to $ W _ {2} ^ {1,0} ( D) $ is called a generalized solution of the third (second, if $ \sigma \equiv 0 $) mixed problem (1), (2), (6) if it satisfies the integral identity

$$ \int\limits _ { D } \left [ - u v _ {t} + \sum _ {i , j = 1 } ^ { n } \alpha _ {i,j} u _ {x _ {i} } v _ {x _ {j} } - \left ( \sum _ { i= } 1 ^ { n } b _ {i} u _ {x _ {i} } + c u \right ) v \right ] d x d t + $$

$$ + \int\limits _ \Gamma \sigma u v d S = $$

$$ = \ \int\limits _ { D } f v d x d t + \int\limits _ {\Omega _ {0} } \phi v d x + \int\limits _ \Gamma \psi v d S $$

for all $ v $ in $ W _ {2} ^ {1} $ such that $ v \mid _ {\Omega _ {T} } = 0 $.

A generalized solution of each of these problems exists and is unique; moreover, if $ f \in L _ {p} ( D) $ is continuous in $ D $, for sufficiently large $ p $, then it even satisfies a Hölder condition for some exponent $ \alpha > 0 $. By increasing the smoothness of the given functions and the boundary of the domain subject to the compatibility conditions, the smoothness of the generalized solution increases. For example, consider the heat equation with $ \phi \equiv 0 $ and $ \psi \equiv 0 $, and let $ \partial \Omega $ be a sufficiently smooth surface. Then the generalized solution of the first mixed problem belongs to $ W _ {2} ^ {2 ( s + 1 ) , s + 1 } ( D) $, provided that $ f \in W _ {2} ^ {2 s , s } ( D) $ and the compatibility conditions

$$ \tag{7 } f \mid _ {\partial \Omega _ {0} } = \ ( \Delta f + f _ {t} ) \mid _ {\partial \Omega _ {0} } = \dots = \ \sum _ {i = 0 } ^ { {s } - 1 } \Delta ^ {i} \left . \frac{\partial ^ {s - 1 - i } f }{\partial t ^ {s - 1 - i } } \right | _ {\partial \Omega _ {0} } $$

hold.

In particular, if $ f \in L _ {2} ( D) $, then the solution belongs to $ W _ {2} ^ {2,1} ( D) $ when $ ( x , t ) \in D $, it satisfies the heat equation and its trace on $ \Omega _ {0} $ is equal to zero. If $ f \in W _ {2} ^ {2s,s} $ for sufficiently large $ s $ and the compatibility conditions (7) hold, then by virtue of the imbedding theorems, the generalized solution is classical. An analogous statement holds for generalized solutions of the basic mixed problems for equation (1'}) when the coefficients are sufficiently smooth.

Let $ \Omega = \mathbf R ^ {n} $. The problem of finding a solution in the strip $ D = \mathbf R ^ {n} \times ( 0 , T ) $ for the parabolic system (1) satisfying the initial condition (2) on the characteristic $ \Omega _ {0} = \{ x \in \mathbf R ^ {n} , t = 0 \} $ is called the Cauchy problem for (1). A classical solution of the Cauchy problem (1), (2) is a vector function $ u ( x , t ) $ belonging to $ C ^ {2p,1} ( D) \cap C ( D \cup \Omega _ {0} ) $ and satisfying (1) in $ D $ and (2) on $ \Omega $. If the right-hand side $ f ( x , t ) $ belongs to the Hölder space $ C ^ \alpha ( \overline{D}\; ) $ for some $ \alpha > 0 $, and the coefficients are sufficiently smooth in $ \overline{D}\; $( they and their derivatives are bounded), then for any bounded continuous initial vector function $ \phi ( x) $ on $ \mathbf R ^ {n} $ there is a bounded solution of the Cauchy problem on $ D $, and this bounded solution is unique.

The condition of boundedness can be replaced by the condition of "not too rapid growth" . For example, the following is true for second-order equations. Let the coefficients of equation (1'}),

$$ a _ {i,j} ( x , t ) ,\ b _ {i} ( x , t ) ,\ c ( x , t ) \ \textrm{ and } \ \ \frac{\partial a _ {i,j} ( x , t ) }{\partial x _ {i} } $$

belong to the Hölder space $ C ^ \alpha ( \overline{D}\; ) $ for some $ \alpha > 0 $. Further, let $ \phi ( x) $ be continuous in $ \mathbf R ^ {n} $ and let $ f ( x , t ) $ be continuous in $ \overline{D}\; $, locally Hölder continuous in $ x $ uniformly for $ t \in [ 0 , T ] $( for some exponent $ \alpha > 0 $), and such that

$$ | \phi ( x) | \leq C e ^ {h | x | ^ {2} } ,\ \ x \in \mathbf R ^ {n} , $$

$$ | f ( x , t ) | \leq C e ^ {h | x | ^ {2} } ,\ ( x , t ) \in D . $$

Then for sufficiently small $ T $( depending on $ h $), there is a solution of the Cauchy problem (1'}), (2'}) in the strip $ D = \mathbf R ^ {n} \times ( 0 , T ) $. It can be written in the form

$$ u ( x , t ) = \int\limits _ {\mathbf R ^ {n} } \Gamma ( x , t ; \xi , 0 ) \phi ( \xi ) d \xi + $$

$$ + \int\limits _ { 0 } ^ { t } \int\limits _ {\mathbf R ^ {n} } \Gamma ( x , t ; \xi , \tau ) f ( \xi , \tau ) d \xi d \tau , $$

where $ \Gamma ( x , t ; \xi , \tau ) $ is a fundamental solution of (1'}), and satisfies the estimate

$$ \tag{8 } | u ( x , t ) | \leq C _ {1} e ^ {h | x | ^ {2} } $$

for some positive constants $ C _ {1} $ and $ k $. Condition (8) guarantees the uniqueness of the solution of the Cauchy problem.

In the case of an equation with constant coefficients it is possible to find a condition of type (8) on the growth of the solution that is necessary and sufficient for its uniqueness. For example, for a solution of the Cauchy problem for the heat equation to be unique in the class of functions satisfying the inequality

$$ | u ( x , t ) | \leq C e ^ {| x | h ( | x | ) } , $$

where $ h ( | x | ) $ is a positive continuous function on $ [ 0 , \infty ) $, it is necessary and sufficient that the integral $ \int _ {0} ^ \infty dr / h( r) $ diverges.

For parabolic equations one can also consider problems without initial conditions (the Fourier problem). For example, one can ask for the solution of the homogeneous heat equation in the cylinder $ D = \{ x \in \Omega, - \infty < t < + \infty \} $, where $ \Omega $ is a bounded domain with sufficiently smooth boundary $ \partial \Omega $, satisfying the boundary condition

$$ u ( x , t ) \mid _ {x \in \partial \Omega } = \psi ( x , t ) . $$

If $ \psi $ is continuous and bounded, then there is a bounded solution of the Fourier problem, and this is the unique bounded solution.

For parabolic equations and systems it is also possible to consider the first mixed problem in a non-cylindrical domain $ D $ in the case when the lateral surface contains characteristic points (points of contact with the planes $ t = \textrm{ const } $). In particular, it is possible to consider the Dirichlet problem, where boundary conditions are given on the entire boundary $ \partial \Omega $. Under specific conditions on the set of characteristic points and on the order of contact of the characteristic points of $ \partial \Omega $ with a characteristic plane, the Dirichlet problem has a unique solution (in the space $ W _ {2} ^ {B ^ {0} } $). For example, suppose (for the sake of simplicity) that $ D \subset \mathbf R ^ {2} $ is a strictly-convex domain and that the equation of the boundary in a neighbourhood of the upper characteristic point $ ( x ^ {0} , t ^ {0} ) $ has the form $ x - x ^ {0} = \phi _ {1} ( t) $ when $ x \leq x ^ {0} $, and $ x - x ^ {0} = \phi _ {2} ( t) $ when $ x \geq x ^ {0} $( $ t ^ {0} - \delta < t \leq t ^ {0} $). Then the divergence of both integrals

$$ \int\limits _ {t ^ {0} - \delta } ^ { {t ^ {0}} } | \phi _ {i} ( t) ^ {-} 2p | dt ,\ \ i = 1 , 2 , $$

guarantees the existence and uniqueness of a solution of the Dirichlet problem for a second-order parabolic equation. This condition is also necessary in this class of equations.

References

[1] V.S. Vladimirov, "Equations of mathematical physics" , MIR (1984) (Translated from Russian)
[2] V.A. Il'in, "The solvability of mixed problems for hyperbolic and parabolic equations" Russian Math. Surveys , 15 : 2 (1960) pp. 85–142 Uspekhi Mat. Nauk , 15 : 2 (1960) pp. 97–154
[3] A.M. Il'in, A.S. Kalashnikov, O.A. Oleinik, "Linear equations of the second order of parabolic type" Russian Math. Surveys , 17 : 3 (1962) pp. 1–143 Uspekhi Mat. Nauk , 17 : 3 (1962) pp. 3–146
[4] S.N. Kruzhkov, "A priori bounds and some properties of solutions of elliptic and parabolic equations" Mat. Sb. , 65 : 4 (1964) pp. 522–570 (In Russian)
[5] O.A. Ladyzhenskaya, V.A. Solonnikov, N.N. Ural'tseva, "Linear and quasi-linear equations of parabolic type" , Amer. Math. Soc. (1968) (Translated from Russian)
[6] O.A. Ladyzhenskaya, "The boundary value problems of mathematical physics" , Springer (1985) (Translated from Russian)
[7] O.A. Ladyzhenskaya, "On uniqueness of a solution of the Cauchy problem for a linear parabolic equation" Mat. Sb. , 27 : 2 (1950) pp. 175–184 (In Russian)
[8] V.P. Mikhailov, "The Dirichlet problem for a parabolic equation I" Mat. Sb. , 61 : 1 (1963) pp. 40–64 (In Russian)
[9] V.P. Mikhailov, "The Dirichlet problem for a parabolic equation II" Mat. Sb. , 62 : 2 (1963) pp. 140–159 (In Russian)
[10] J. Nash, "Continuity of solutions of parabolic and elliptic equations" Amer. J. Math. , 80 (1958) pp. 931–954
[11] I.G. Petrovskii, "Partial differential equations" , Saunders (1967) (Translated from Russian)
[12] I.G. Petrovskii, "On Cauchy's problem for systems of linear partial differential equations in a nonanalytic function domain" Bull. Moskov. Gos. Univ. Ser. A. , 1 : 7 (1938) pp. 1–72 (In Russian)
[13] I.G. Petrovskii, "Zur ersten Randwertaufgabe der Wärmeleitungsgleichung" Comp. Math. , 1 (1934) pp. 383–419
[14] S.L. Sobolev, "Partial differential equations of mathematical physics" , Pergamon (1964) (Translated from Russian)
[15] V.A. Solonnikov, "Boundary value problems of mathematical physics III" Proc. Steklov Inst. Math. , 83 (1965) Trudy Mat. Inst. Steklov , 83 (1965)
[16] A.N. [A.N. Tikhonov] Tichonoff, A.A. Samarskii, "Differentialgleichungen der mathematischen Physik" , Deutsch. Verlag Wissenschaft. (1959) (Translated from Russian)
[17] A.N. Tikhonov, Byull. Moskov. Univ. (A) , 1 : 9 (1938) pp. 1–43
[18] A.N. Tikhonov, Mat. Sb. , 42 : 2 (1935) pp. 199–216
[19] A. Friedman, "Partial differential equations of parabolic type" , Prentice-Hall (1964)
[20] S.D. Eidel'man, "Parabolic systems" , North-Holland (1969) (Translated from Russian)

Comments

In the current literature the initial boundary value problems (1'}), (2'}), (4)–(6) are not referred to as "mixed" . Sometimes expressions like Cauchy–Dirichlet or Cauchy–Neumann are used. Quite often, by a problem with Dirichlet data for a parabolic equation is meant a problem in which such data are prescribed on the parabolic boundary. Besides the first, second and third kind of boundary data, higher-order problems are also considered. For further comments and more references see Linear parabolic partial differential equation and system.

How to Cite This Entry:
Mixed and boundary value problems for parabolic equations and systems. Encyclopedia of Mathematics. URL: http://encyclopediaofmath.org/index.php?title=Mixed_and_boundary_value_problems_for_parabolic_equations_and_systems&oldid=15472
This article was adapted from an original article by V.P. Mikhailov (originator), which appeared in Encyclopedia of Mathematics - ISBN 1402006098. See original article