Derivation of the midpoint method Midpoint method
illustration of numerical integration equation
y
′
=
y
,
y
(
0
)
=
1.
{\displaystyle y =y,y(0)=1.}
blue: euler method, green: midpoint method, red: exact solution,
y
=
e
t
.
{\displaystyle y=e^{t}.}
step size
h
=
1.0.
{\displaystyle h=1.0.}
the same illustration
h
=
0.25.
{\displaystyle h=0.25.}
seen midpoint method converges faster euler method.
the midpoint method refinement of euler s method
y
n
+
1
=
y
n
+
h
f
(
t
n
,
y
n
)
,
{\displaystyle y_{n+1}=y_{n}+hf(t_{n},y_{n}),\,}
and derived in similar manner. key deriving euler s method approximate equality
y
(
t
+
h
)
≈
y
(
t
)
+
h
f
(
t
,
y
(
t
)
)
(
2
)
{\displaystyle y(t+h)\approx y(t)+hf(t,y(t))\qquad \qquad (2)}
which obtained slope formula
y
′
(
t
)
≈
y
(
t
+
h
)
−
y
(
t
)
h
(
3
)
{\displaystyle y (t)\approx {\frac {y(t+h)-y(t)}{h}}\qquad \qquad (3)}
and keeping in mind
y
′
=
f
(
t
,
y
)
.
{\displaystyle y =f(t,y).}
for midpoint methods, 1 replaces (3) more accurate
y
′
(
t
+
h
2
)
≈
y
(
t
+
h
)
−
y
(
t
)
h
{\displaystyle y \left(t+{\frac {h}{2}}\right)\approx {\frac {y(t+h)-y(t)}{h}}}
when instead of (2) find
y
(
t
+
h
)
≈
y
(
t
)
+
h
f
(
t
+
h
2
,
y
(
t
+
h
2
)
)
.
(
4
)
{\displaystyle y(t+h)\approx y(t)+hf\left(t+{\frac {h}{2}},y\left(t+{\frac {h}{2}}\right)\right).\qquad \qquad (4)}
one cannot use equation find
y
(
t
+
h
)
{\displaystyle y(t+h)}
1 not know
y
{\displaystyle y}
@
t
+
h
/
2
{\displaystyle t+h/2}
. solution use taylor series expansion if using euler method solve
y
(
t
+
h
/
2
)
{\displaystyle y(t+h/2)}
:
y
(
t
+
h
2
)
≈
y
(
t
)
+
h
2
y
′
(
t
)
=
y
(
t
)
+
h
2
f
(
t
,
y
(
t
)
)
,
{\displaystyle y\left(t+{\frac {h}{2}}\right)\approx y(t)+{\frac {h}{2}}y (t)=y(t)+{\frac {h}{2}}f(t,y(t)),}
which, when plugged in (4), gives us
y
(
t
+
h
)
≈
y
(
t
)
+
h
f
(
t
+
h
2
,
y
(
t
)
+
h
2
f
(
t
,
y
(
t
)
)
)
{\displaystyle y(t+h)\approx y(t)+hf\left(t+{\frac {h}{2}},y(t)+{\frac {h}{2}}f(t,y(t))\right)}
and explicit midpoint method (1e).
the implicit method (1i) obtained approximating value @ half step
t
+
h
/
2
{\displaystyle t+h/2}
midpoint of line segment
y
(
t
)
{\displaystyle y(t)}
y
(
t
+
h
)
{\displaystyle y(t+h)}
y
(
t
+
h
2
)
≈
1
2
(
y
(
t
)
+
y
(
t
+
h
)
)
{\displaystyle y\left(t+{\frac {h}{2}}\right)\approx {\frac {1}{2}}{\bigl (}y(t)+y(t+h){\bigr )}}
and thus
y
(
t
+
h
)
−
y
(
t
)
h
≈
y
′
(
t
+
h
2
)
≈
k
=
f
(
t
+
h
2
,
1
2
(
y
(
t
)
+
y
(
t
+
h
)
)
)
{\displaystyle {\frac {y(t+h)-y(t)}{h}}\approx y \left(t+{\frac {h}{2}}\right)\approx k=f\left(t+{\frac {h}{2}},{\frac {1}{2}}{\bigl (}y(t)+y(t+h){\bigr )}\right)}
inserting approximation
y
n
+
h
k
{\displaystyle y_{n}+h\,k}
y
(
t
n
+
h
)
{\displaystyle y(t_{n}+h)}
results in implicit runge-kutta method
k
=
f
(
t
n
+
h
2
,
y
n
+
h
2
k
)
y
n
+
1
=
y
n
+
h
k
{\displaystyle {\begin{aligned}k&=f\left(t_{n}+{\frac {h}{2}},y_{n}+{\frac {h}{2}}k\right)\\y_{n+1}&=y_{n}+h\,k\end{aligned}}}
which contains implicit euler method step size
h
/
2
{\displaystyle h/2}
first part.
because of time symmetry of implicit method, terms of degree in
h
{\displaystyle h}
of local error cancel, local error automatically of order
o
(
h
3
)
{\displaystyle {\mathcal {o}}(h^{3})}
. replacing implicit explicit euler method in determination of
k
{\displaystyle k}
results again in explicit midpoint method.
Comments
Post a Comment