196 3 Real Analysis
z(t)=(a+ib)exp
(
− 1 +
√
5
2
it
)
+(c+id)exp
(
− 1 −
√
5
2
it
)
,
for arbitrary real numbersa, b, c, d. Sincexandyare, respectively, the real and complex
parts of the solution, they have the general form
x(t)=acos
− 1 +
√
5
2
t−bsin
− 1 +
√
5
2
t+ccos
− 1 −
√
5
2
t−dsin
− 1 −
√
5
2
t,
y(t)=asin
− 1 +
√
5
2
t+bcos
− 1 +
√
5
2
t+csin
− 1 −
√
5
2
t+dcos
− 1 −
√
5
2
t.
The problem is solved.
Our second example is an equation published by M. Ghermanescu in the ̆ Mathematics
Gazette, Bucharest. Its solution combines several useful techniques.
Example.Solve the differential equation
2 (y′)^3 −yy′y′′−y^2 y′′′= 0.
Solution.In a situation like this, where the variablexdoes not appear explicitly, one can
reduce the order of the equation by takingyas the variable andp=y′as the function.
The higher-order derivatives ofy′′are
y′′=
d
dx
y′=
d
dy
p
dy
dx
=p′p,
y′′′=
d
dx
y′′=
(
d
dy
pp′
)
dy
dx
=
(
(p′)^2 +pp′′
)
p.
We end up with a second-order differential equation
2 p^3 −yp^2 p′−y^2 pp′′−y^2 p(p′)^2 = 0.
A family of solutions isp=0, that is,y′=0. This family consists of the constant
functionsy=C. Dividing the equation by−p, we obtain
y^2 p′′+y^2 (p′)^2 +ypp′− 2 p^2 = 0.
The distribution of the powers ofyreminds us of the Euler–Cauchy equation, while the
last terms suggests the substitutionu=p^2. And indeed, we obtain the Euler–Cauchy
equation
y^2 u′′+yu′− 4 u= 0 ,