Numerical solution of differential equations
Ordinary differential equations.
Simplest form of O.D.E.
Analytic
solution:
.
Digital solution presents the table of function:
x0 |
x1 |
x2 |
… |
xn |
… |
y0 |
y1 |
y2 |
… |
yn |
… |
Method of approximations.
Problem of Koshie:
-
O.D.E.
-
initial condition
.
Let
to find solution for
;
1-st
approximation:
,
;
2-nd approximation:
;
…
.
Numerical methods.
M
ethods
of Euler.
Put step: h; xi=x0 + ih, i= 0, 1, 2, …
The
value of
is tg of the corner of tangent at the point
,
and we consider the part of tangent instead of part of line
:
Modifications of method of Euler.
1-st mod.
C
onsider
point
.
O
…
We find the value of tangent of corner in the middle of distance, then
.
2-nd mod.
Equation of higher order may be reduced to system.
Define:
We obtain system:
Express
for last equation:
Apply, for example, method of Euler:
Method of Runghe-Kutta.
Method of Runghe-Kutta may be extrapolated for system:
Brink problem for differential equations.
Simplest case:
.
Boundary conditions
initial conditions: ending conditions:
-
problem of 1-st sort;
-
2-nd sort;
or or - mixed sort.
Consider linear variant:
Linear two-points brink problem for 2-nd order differential equation.
Reducing of linear 2-points brink problem for 2-nd order differential equation to problem of Koshie.
Let find solution in form:
where
Consider 1-st boundary condition:
Let guarantee equivalence without depending on coefficient c:
if
,
then
else
if
,
then
.
Consider c provided 2-nd boundary condition:
Methods, based on finite differences.
finite
value:
For boundary points:
,
but error is great;
take order of Tailor:
Analogously:
.
2-n derivative:
.
So,
we obtain system for
:
,
boundary
points -
This linear system may be resolved by any method, but it may be applied reducing to Koshie problem.
Let
.
Put digital model:
Exclude
:
for
we have recurrence:
Analogously
for
:
