2.4 HW4

  2.4.1 Section 2.6 problem 19
  2.4.2 Section 2.6 problem 20
  2.4.3 Section 2.6 problem 21
  2.4.4 Section 2.6 problem 22
  2.4.5 Section 2.6 problem 23
  2.4.6 Section 2.6 problem 24
  2.4.7 Section 2.7 problem 20
  2.4.8 Section 3.1 problem 1
  2.4.9 Section 3.1 problem 2
  2.4.10 Section 3.1 problem 3
  2.4.11 Section 3.1 problem 4
  2.4.12 Section 3.1 problem 5
  2.4.13 Section 3.1 problem 6
  2.4.14 Section 3.1 problem 7
  2.4.15 Section 3.1 problem 8
PDF (letter size)
PDF (legal size)

2.4.1 Section 2.6 problem 19

Question Show that  \(x^{2}y^{3}+x\left ( 1+y^{2}\right ) y^{\prime }=0\) is not exact, and then becomes exact when multiplied by \(\mu \left ( x,y\right ) =\frac{1}{xy^{3}}\) and then solve.

Solution The first step is to apply theorem two and also check where the ODE is singular. Writing it as\[ \frac{dy}{dx}=f\left ( x,y\right ) =\frac{-x^{2}y^{3}}{x\left ( 1+y^{2}\right ) }\] This is non-linear first order ODE. There is a pole at \(x=0\,\). From theorem two, this says that unique solution is not guaranteed to exist since the first condition which says that \(f\left ( x,y\right ) \) must be continuous, was not satisfied. Now the ODE is solved.\[ \overset{M}{\overbrace{x^{2}y^{3}}}+\overset{N}{\overbrace{x\left ( 1+y^{2}\right ) }}y^{\prime }=0 \] Hence\begin{align*} M\left ( x,y\right ) & =x^{2}y^{3}\\ N\left ( x,y\right ) & =x\left ( 1+y^{2}\right ) \end{align*}

An ODE is exact when \(\frac{\partial M}{\partial y}=\frac{\partial N}{\partial x}\). These are now calculated to see if the ODE is exact or not\begin{align*} \frac{\partial M}{\partial y} & =3x^{2}y^{2}\\ \frac{\partial N}{\partial x} & =1+y^{2} \end{align*}

The above shows that that \(\frac{\partial M}{\partial y}\neq \frac{\partial N}{\partial x}\) therefore the ODE is not exact. Multiplying the original ODE by given integrating factor it becomes \begin{align*} \left ( \mu x^{2}y^{3}\right ) +\mu x\left ( 1+y^{2}\right ) y^{\prime } & =0\\ \frac{x^{2}y^{3}}{xy^{3}}+\frac{1}{xy^{3}}x\left ( 1+y^{2}\right ) y^{\prime } & =0\\ x+\frac{1}{y^{3}}\left ( 1+y^{2}\right ) y^{\prime } & =0 \end{align*}

Now \(\bar{M}=x\) and \(\bar{N}=\frac{1}{y^{3}}\left ( 1+y^{2}\right ) \). Checking that the new \(\bar{M},\bar{N}\,\ \)are indeed exact. \begin{align*} \frac{\partial \bar{M}}{\partial y} & =0\\ \frac{\partial \bar{N}}{\partial x} & =0 \end{align*}

The new ODE is exact. Now the ODE is solved using the standard method.\begin{align} \frac{\partial \Psi \left ( x,y\right ) }{\partial x} & =\bar{M}=x\tag{1}\\ \frac{\partial \Psi \left ( x,y\right ) }{\partial y} & =\bar{N}=\frac{1}{y^{3}}\left ( 1+y^{2}\right ) \tag{2} \end{align}

Integrating (1) w.r.t \(x\) gives\begin{align} \Psi & =\frac{1}{2}x^{2}+f\left ( y\right ) \tag{3}\\ \frac{\partial \Psi }{\partial y} & =f^{\prime }\left ( y\right ) \nonumber \end{align}

Comparing the above to (2) in order to solve for \(f^{\prime }\left ( y\right ) \) gives\begin{align} f^{\prime }\left ( y\right ) & =\frac{1+y^{2}}{y^{3}}\nonumber \\ f\left ( y\right ) & =\int \frac{1+y^{2}}{y^{3}}dy+c \tag{4} \end{align}

We need now to solve \(\int \frac{1+y^{2}}{y^{3}}dy\)\begin{align*} \int \frac{1+y^{2}}{y^{3}}dy & =\int \frac{1}{y^{3}}dy+\int \frac{y^{2}}{y^{3}}dy\\ & =-\frac{1}{2y^{2}}+\int \frac{1}{y}dy\\ & =-\frac{1}{2y^{2}}+\ln \left \vert y\right \vert \end{align*}

Using the above solution in (4) gives\[ f\left ( y\right ) =-\frac{1}{2y^{2}}+\ln \left \vert y\right \vert +c \] Using the above in (3) gives\[ \Psi =\frac{1}{2}x^{2}-\frac{1}{2y^{2}}+\ln \left \vert y\right \vert +c \] But \(\frac{d\Psi }{dx}=c_{0}\), therefore the above simplifies to, after collecting all constants to one\[ \frac{1}{2}x^{2}-\frac{1}{2y^{2}}+\ln \left \vert y\right \vert =C\qquad x\neq 0 \] Checking \(y=0\) as solution, shows that putting \(y=0\) in \(f\left ( x,y\right ) =\frac{-x^{2}y^{3}}{x\left ( 1+y^{2}\right ) }=0\). Hence \(y=0\) is also a solution.

Summary The solutions are\begin{align*} \frac{1}{2}x^{2}-\frac{1}{2y^{2}}+\ln \left \vert y\right \vert & =C\qquad x\neq 0,y\neq 0\\ y & =0\qquad x\neq 0 \end{align*}

2.4.2 Section 2.6 problem 20

Question Show that  \(\left ( \frac{\sin y}{y}-2e^{-x}\sin x\right ) +\left ( \frac{\cos y+2e^{-x}\cos x}{y}\right ) y^{\prime }=0\) is not exact, and then becomes exact when multiplied by \(\mu \left ( x,y\right ) =ye^{x}\) and then solve.

Solution First we will check where the ODE is singular. Writing it as\[ \frac{dy}{dx}=f\left ( x,y\right ) =\frac{\frac{\sin y}{y}-2e^{-x}\sin x}{\frac{\cos y+2e^{-x}\cos x}{y}}\] This is non-linear first order ODE. We see a pole at \(y=0\,\). Hence \(y\neq 0\). From theorem two, this says that that unique solution is not guaranteed since first condition which says that \(f\left ( x,y\right ) \) must be continuous, was not satisfied. \begin{align*} M\left ( x,y\right ) & =\frac{\sin y}{y}-2e^{-x}\sin x\\ N\left ( x,y\right ) & =\frac{\cos y+2e^{-x}\cos x}{y} \end{align*}

An ODE is exact when \(\frac{\partial M}{\partial y}=\frac{\partial N}{\partial x}\). These are now calculated to see if the ODE is exact or not\begin{align*} \frac{\partial M}{\partial y} & =\ln y\sin y+\frac{1}{y}\cos y\\ \frac{\partial N}{\partial x} & =\frac{\partial }{\partial x}\left ( \frac{1}{y}\cos y+\frac{1}{y}2e^{-x}\cos x\right ) =\frac{-1}{y}2e^{-x}\cos x-\frac{1}{y}2e^{-x}\sin x=\frac{-2e^{-x}}{y}\left ( \cos x+\sin x\right ) \end{align*}

From above we see that \(\frac{\partial M}{\partial y}\neq \frac{\partial N}{\partial x}\) therefore the ODE is not exact. Multiplying the original ODE by given integrating factor it becomes \begin{align*} \mu \left ( \frac{\sin y}{y}-2e^{-x}\sin x\right ) +\mu \left ( \frac{\cos y+2e^{-x}\cos x}{y}\right ) y^{\prime } & =0\\ ye^{x}\left ( \frac{\sin y}{y}-2e^{-x}\sin x\right ) +ye^{x}\left ( \frac{\cos y+2e^{-x}\cos x}{y}\right ) y^{\prime } & =0\\ \left ( e^{x}\sin y-2y\sin x\right ) +\left ( e^{x}\cos y+2\cos x\right ) y^{\prime } & =0 \end{align*}

Now \begin{align*} \bar{M} & =e^{x}\sin y-2y\sin x\\ \bar{N} & =e^{x}\cos y+2\cos x \end{align*}

Checking now the new \(\bar{M},\bar{N}\,\ \)are indeed exact.\begin{align*} \frac{\partial \bar{M}}{\partial y} & =e^{x}\cos y-2\sin x\\ \frac{\partial \bar{N}}{\partial x} & =e^{x}\cos y-2\sin x \end{align*}

The new ODE is exact. Now the ODE is solved using the standard method.\begin{align} \frac{\partial \Psi \left ( x,y\right ) }{\partial x} & =\bar{M}=e^{x}\sin y-2y\sin x\tag{1}\\ \frac{\partial \Psi \left ( x,y\right ) }{\partial y} & =\bar{N}=e^{x}\cos y+2\cos x \tag{2} \end{align}

Integrating (1) w.r.t \(x\) gives\begin{align} \Psi & =e^{x}\sin y+2y\cos x+f\left ( y\right ) \tag{3}\\ \frac{\partial \Psi }{\partial y} & =e^{x}\cos y+2\cos x+f^{\prime }\left ( y\right ) \nonumber \end{align}

Comparing the above to (2) in order to solve for \(f^{\prime }\left ( y\right ) \) gives\begin{align} e^{x}\cos y+2\cos x+f^{\prime }\left ( y\right ) & =e^{x}\cos y+2\cos x\nonumber \\ f^{\prime }\left ( y\right ) & =0\nonumber \\ f\left ( y\right ) & =c \tag{4} \end{align}

Substituting the above into (3) gives\[ \Psi =e^{x}\sin y+2y\cos x+c \] But \(\frac{d\Psi }{dx}=c_{0}\), therefore the above simplifies to, after collecting all constants to one\[ e^{x}\sin y+2y\cos x=C\qquad y\neq 0 \]

2.4.3 Section 2.6 problem 21

Question Show that  \(y+\left ( 2x-ye^{y}\right ) y^{\prime }=0\) is not exact, and then becomes exact when multiplied by \(\mu \left ( x,y\right ) =y\) and then solve.

Solution \begin{align*} M\left ( x,y\right ) & =y\\ N\left ( x,y\right ) & =2x-ye^{y} \end{align*}

An ODE is exact when \(\frac{\partial M}{\partial y}=\frac{\partial N}{\partial x}\). These are now calculated to see if the ODE is exact or not\begin{align*} \frac{\partial M}{\partial y} & =1\\ \frac{\partial N}{\partial x} & =2 \end{align*}

From above we see that \(\frac{\partial M}{\partial y}\neq \frac{\partial N}{\partial x}\) therefore the ODE is not exact. Multiplying the original ODE by given integrating factor it becomes \begin{align*} \mu y+\mu \left ( 2x-ye^{y}\right ) y^{\prime } & =0\\ y^{2}+\left ( 2xy-y^{2}e^{y}\right ) y^{\prime } & =0 \end{align*}

Now \begin{align*} \bar{M} & =y^{2}\\ \bar{N} & =2xy-y^{2}e^{y} \end{align*}

Checking now the new \(\bar{M},\bar{N}\,\ \)are indeed exact.\begin{align*} \frac{\partial \bar{M}}{\partial y} & =2y\\ \frac{\partial \bar{N}}{\partial x} & =2y \end{align*}

The new ODE is exact. Now the ODE is solved using the standard method.\begin{align} \frac{\partial \Psi \left ( x,y\right ) }{\partial x} & =\bar{M}=y^{2}\tag{1}\\ \frac{\partial \Psi \left ( x,y\right ) }{\partial y} & =\bar{N}=2xy-y^{2}e^{y} \tag{2} \end{align}

Integrating (1) w.r.t \(x\) gives\begin{align} \Psi & =y^{2}x+f\left ( y\right ) \tag{3}\\ \frac{\partial \Psi }{\partial y} & =2yx+f^{\prime }\left ( y\right ) \nonumber \end{align}

Comparing the above to (2) in order to solve for \(f^{\prime }\left ( y\right ) \) gives\begin{align} 2yx+f^{\prime }\left ( y\right ) & =2xy-y^{2}e^{y}\nonumber \\ f^{\prime }\left ( y\right ) & =-y^{2}e^{y}\nonumber \\ f\left ( y\right ) & =-\int y^{2}e^{y}dy+c \tag{4} \end{align}

The integral \(\int y^{2}e^{y}dy\) can be found using integration by parts. Let \(u=y^{2},dv=e^{y}\rightarrow du=2y,v=e^{y}\), therefore\begin{align*} \int y^{2}e^{y}dy & =\int udv\\ & =uv-\int vdu\\ & =y^{2}e^{y}-2\int ye^{y}dy \end{align*}

Applying integration by parts again to \(\int ye^{y}dy\), where now \(u=y,dv=e^{y}\rightarrow du=1,v=e^{y}\), the above becomes\begin{align*} \int y^{2}e^{y}dy & =y^{2}e^{y}-2\left ( ye^{y}-\int e^{y}dy\right ) \\ & =y^{2}e^{y}-2\left ( ye^{y}-e^{y}\right ) \\ & =y^{2}e^{y}-2ye^{y}+2e^{y}\\ & =e^{y}\left ( y^{2}-2y+2\right ) \end{align*}

Therefore from (4)\[ f\left ( y\right ) =-e^{y}\left ( y^{2}-2y+2\right ) +c \] Substituting the above into (3) gives\[ \Psi =y^{2}x-e^{y}\left ( y^{2}-2y+2\right ) +c \] But \(\frac{d\Psi }{dx}=c_{0}\), therefore the above simplifies to, after collecting all constants to one\[ y^{2}x-e^{y}\left ( y^{2}-2y+2\right ) =C \]

2.4.4 Section 2.6 problem 22

Question Show that  \(\left ( x+2\right ) \sin y+\left ( x\cos y\right ) y^{\prime }=0\) is not exact, and then becomes exact when multiplied by \(\mu \left ( x,y\right ) =xe^{x}\) and then solve.

Solution \begin{align*} M\left ( x,y\right ) & =\left ( x+2\right ) \sin y\\ N\left ( x,y\right ) & =x\cos y \end{align*}

An ODE is exact when \(\frac{\partial M}{\partial y}=\frac{\partial N}{\partial x}\). These are now calculated to see if the ODE is exact or not\begin{align*} \frac{\partial M}{\partial y} & =\left ( x+2\right ) \cos y\\ \frac{\partial N}{\partial x} & =\cos y \end{align*}

From above we see that \(\frac{\partial M}{\partial y}\neq \frac{\partial N}{\partial x}\) therefore the ODE is not exact. Multiplying the original ODE by given integrating factor it becomes \begin{align*} \mu \left ( x+2\right ) \sin y+\mu \left ( x\cos y\right ) y^{\prime } & =0\\ xe^{x}\left ( x+2\right ) \sin y+xe^{x}\left ( x\cos y\right ) y^{\prime } & =0 \end{align*}

Now \begin{align*} \bar{M} & =\left ( x^{2}e^{x}+2xe^{x}\right ) \sin y\\ \bar{N} & =x^{2}e^{x}\cos y \end{align*}

Checking now the new \(\bar{M},\bar{N}\,\ \)are indeed exact.\begin{align*} \frac{\partial \bar{M}}{\partial y} & =\left ( x^{2}e^{x}+2xe^{x}\right ) \cos y\\ \frac{\partial \bar{N}}{\partial x} & =2xe^{x}\cos y+x^{2}e^{x}\cos y=\left ( x^{2}e^{x}+2xe^{x}\right ) \cos y \end{align*}

The new ODE is exact. Now the ODE is solved using the standard method.\begin{align} \frac{\partial \Psi \left ( x,y\right ) }{\partial x} & =\bar{M}=\left ( x^{2}e^{x}+2xe^{x}\right ) \sin y\tag{1}\\ \frac{\partial \Psi \left ( x,y\right ) }{\partial y} & =\bar{N}=x^{2}e^{x}\cos y \tag{2} \end{align}

Integrating (2) w.r.t \(y\) as it is simpler than integrating (1) w.r.t. \(x\), gives\begin{align} \Psi & =\int x^{2}e^{x}\cos ydy=x^{2}e^{x}\sin y+f\left ( x\right ) \tag{3}\\ \frac{\partial \Psi }{\partial x} & =2xe^{x}\sin y+x^{2}e^{x}\sin y+f^{\prime }\left ( x\right ) \nonumber \end{align}

Comparing the above to (1) in order to solve for \(f^{\prime }\left ( x\right ) \) gives\begin{align} 2xe^{x}\sin y+x^{2}e^{x}\sin y+f^{\prime }\left ( x\right ) & =\left ( x^{2}e^{x}+2xe^{x}\right ) \sin y\nonumber \\ f^{\prime }\left ( x\right ) & =0\nonumber \\ f\left ( x\right ) & =c \tag{4} \end{align}

Substituting the above into (3) gives\[ \Psi =x^{2}e^{x}\sin y+c \] But \(\frac{d\Psi }{dx}=c_{0}\), therefore \(\Psi =c_{1}\) and the above simplifies to, after collecting all constants to one\[ x^{2}e^{x}\sin y=C \]

2.4.5 Section 2.6 problem 23

Question Show that if \(\frac{N_{x}-M_{y}}{M}=Q\) where \(Q\) is function of \(y\) only, then \(M+Ny^{\prime }=0\) has integrating factor of form \(\mu \left ( y\right ) =e^{\int Q\left ( y\right ) dt}\)

Solution Given the differential equation\[ M\left ( x,y\right ) +N\left ( x,y\right ) \frac{dy\left ( x\right ) }{dx}=0 \] Multiplying by \(\mu \left ( y\right ) \) results in\[ \mu M+\mu Ny^{\prime }=0 \] The above is exact if \[ \frac{\partial \left ( \mu M\right ) }{\partial y}=\frac{\partial \left ( \mu N\right ) }{\partial x}\] Performing the above, taking into account that \(\mu \) depends on \(y\) only, results in\[ \frac{d\mu }{dy}M+\mu \frac{\partial M}{\partial y}=\mu \frac{\partial N}{\partial x}\] The above is first order ODE in \(\mu \,\)\begin{align*} \frac{d\mu }{dy}M & =\mu \frac{\partial N}{\partial x}-\mu \frac{\partial M}{\partial y}\\ \frac{d\mu }{dy} & =\mu \left ( \frac{\frac{\partial N}{\partial x}-\frac{\partial M}{\partial y}}{M}\right ) \end{align*}

Let \(Q=\frac{\frac{\partial N}{\partial x}-\frac{\partial M}{\partial y}}{M}\). If \(Q\) depends on \(y\) only, then the above ODE is separable. Hence\begin{align*} \frac{d\mu }{dy} & =\mu Q\left ( y\right ) \\ \frac{d\mu }{\mu } & =Q\left ( y\right ) dy \end{align*}

Integrating both sides gives\begin{align*} \ln \left \vert \mu \right \vert & =\int Q\left ( y\right ) dy+C\\ \left \vert \mu \right \vert & =e^{\int Q\left ( y\right ) dy+C}\\ \mu \left ( y\right ) & =Ae^{\int Q\left ( y\right ) dy} \end{align*}

Where \(A\) is some constant, which can be taken to be \(1\) leading to the result required to show. The above procedure works only when \(Q=\frac{\frac{\partial N}{\partial x}-\frac{\partial M}{\partial y}}{M}\) happened to be function of \(y\) only. This complete the proof.

2.4.6 Section 2.6 problem 24

Question Show that if \(\frac{N_{x}-M_{y}}{xM-yN}=R\) where \(R\) is function of \(xy\) only, then \(M+Ny^{\prime }=0\) has integrating factor of form \(\mu \left ( x,y\right ) \). Find the general formula for \(\mu \).

Solution Given the differential equation\[ M\left ( x,y\right ) +N\left ( x,y\right ) \frac{dy\left ( x\right ) }{dx}=0 \] Let \(\mu \left ( t\right ) \) where \(t=xy\). Multiplying the above with \(\mu \left ( t\right ) \) gives\[ \mu \left ( t\right ) M\left ( x,y\right ) +\mu \left ( t\right ) N\left ( x,y\right ) \frac{dy\left ( x\right ) }{dx}=0 \] The above is exact when \[ \frac{\partial \mu M}{\partial y}=\frac{\partial \mu N}{\partial x}\] Hence\begin{equation} \frac{\partial \mu }{\partial y}M+\mu \frac{\partial M}{\partial y}=\frac{\partial \mu }{\partial x}N+\mu \frac{\partial N}{\partial x}\tag{1} \end{equation} However,\begin{equation} \frac{\partial \mu }{\partial y}=\frac{d\mu }{dt}\frac{\partial t}{dy}=\frac{d\mu }{dt}x\tag{2} \end{equation} And \begin{equation} \frac{\partial \mu }{\partial x}=\frac{d\mu }{dt}\frac{\partial t}{dx}=\frac{d\mu }{dt}y\tag{3} \end{equation} Substituting (2,3) into (1) gives\begin{align*} \frac{d\mu }{dt}xM+\mu \frac{\partial M}{\partial y} & =\frac{d\mu }{dt}yN+\mu \frac{\partial N}{\partial x}\\ \frac{d\mu }{dt}\left ( xM-yN\right ) & =\mu \frac{\partial N}{\partial x}-\mu \frac{\partial M}{\partial y}\\ \frac{d\mu \left ( t\right ) }{dt} & =\mu \frac{\left ( \frac{\partial N}{\partial x}-\frac{\partial M}{\partial y}\right ) }{\left ( xM-yN\right ) } \end{align*}

In the above, \(\mu \) depends on \(t\) only, where \(t\) is function of \(xy\) only. If \(\frac{\left ( \frac{\partial N}{\partial x}-\frac{\partial M}{\partial y}\right ) }{\left ( xM-yN\right ) }\) depends on \(t\) only, then the above can be considered a separable first order ODE in \(\mu \). Let \(R\left ( t\right ) =\frac{\left ( \frac{\partial N}{\partial x}-\frac{\partial M}{\partial y}\right ) }{\left ( xM-yN\right ) }\) and the above can be written as\[ \frac{d\mu \left ( t\right ) }{dt}=\mu R\left ( t\right ) \] Since separable, then\begin{align*} \frac{d\mu \left ( t\right ) }{\mu } & =R\left ( t\right ) dt\\ \int \frac{d\mu }{\mu } & =\int Rdt\\ \ln \left \vert \mu \right \vert & =\int Rdt+C\\ \left \vert \mu \right \vert & =e^{\int Rdt+C}\\ \mu & =Ae^{\int Rdt} \end{align*}

Where \(A\) is constant of integration which can be taken to be \(1\). Hence \(\mu =e^{\int Rdt}\). This works only if \(R\) is function of \(t\) only.

2.4.7 Section 2.7 problem 20

   2.4.7.1 part a
   2.4.7.2 Part b
   2.4.7.3 Part c
   2.4.7.4 Part d

pict

2.4.7.1 part a

\begin{align*} y^{\prime } & =1-t+y\\ y\left ( t_{0}\right ) & =y_{0} \end{align*}

This is linear first order ODE. Writing it as \(y^{\prime }-y=1-t\), then the integrating factor is \(\mu =e^{-\int dt}=e^{-t}\) and the ODE becomes\[ \frac{d}{dt}\left ( ye^{-t}\right ) =e^{-t}\left ( 1-t\right ) \] Integrating both sides\begin{align} ye^{-t} & =\int e^{-t}\left ( 1-t\right ) dt+c\nonumber \\ & =\int e^{-t}dt-\int te^{-t}dt+c\tag{1} \end{align}

But \(\int te^{-t}dt=\int udv\) where \(u=t,dv=e^{-t}\rightarrow du=1,v=-e^{-t}\), hence\begin{align*} \int te^{-t}dt & =uv-\int vdu\\ & =-te^{-t}+\int e^{-t}dt\\ & =-te^{-t}-e^{-t} \end{align*}

Putting this result in (1) gives\begin{align*} ye^{-t} & =-e^{-t}-\left ( -te^{-t}-e^{-t}\right ) +c\\ & =-e^{-t}+te^{-t}+e^{-t}+c\\ & =te^{-t}+c \end{align*}

Therefore solving for \(y\) gives\begin{equation} y=t+ce^{t}\tag{2} \end{equation} The constant \(c\) is now found from initial conditions.\begin{align*} y_{0} & =t_{0}+ce^{t_{0}}\\ c & =\left ( y_{0}-t_{0}\right ) e^{-t_{0}} \end{align*}

Substituting \(c\) found back into (2) gives the final solution\begin{align} y & =t+\left ( y_{0}-t_{0}\right ) e^{-t_{0}}e^{t}\nonumber \\ & =\left ( y_{0}-t_{0}\right ) e^{t-t_{0}}+t\tag{3} \end{align}

2.4.7.2 Part b

Euler formula is \begin{equation} y_{k}=hf\left ( t_{k-1},y_{k-1}\right ) +y_{k-1}\qquad k=1,2,3,\cdots \tag{1} \end{equation} Where in this problem \(f\left ( t_{k-1},y_{k-1}\right ) \) is the RHS\(\ \)of \(y^{\prime }=1-t+y\) but evaluated at \(t_{k-1}\). Hence\[ f\left ( t_{k-1},y_{k-1}\right ) =1-t_{k-1}+y_{k-1}\] Substituting this into (1) gives\begin{align*} y_{k} & =h\left ( 1-t_{k-1}+y_{k-1}\right ) +y_{k-1}\qquad \\ & =h-ht_{k-1}+hy_{k-1}+y_{k-1}\\ & =\left ( 1+h\right ) y_{k-1}+h-ht_{k-1}\qquad k=1,2,3,\cdots \end{align*}

Which is the required formula asked to derive.

2.4.7.3 Part c

The formula given \(y_{1}=\left ( 1+h\right ) \left ( y_{0}-t_{0}\right ) +t_{1}\) can be found as follows. Since\begin{align*} y_{1} & =y_{0}+hf\left ( t_{0},y_{0}\right ) \\ & =y_{0}+h\left ( 1-t_{0}+y_{0}\right ) \\ & =y_{0}+h-ht_{1}+hy_{0} \end{align*}

Adding \(t_{0}-t_{0}\) to the above will not changed anything, hence\[ y_{1}=y_{0}+h-ht_{1}+hy_{0}+t_{0}-t_{0}\] But \(t_{1}=t_{0}+h\) by definition, hence the above becomes, by replacing \(t_{0}+h\) above with \(t_{1}\)\[ y_{1}=y_{0}+t_{1}-ht_{1}+hy_{0}-t_{0}\] Simplifying\begin{align*} y_{1} & =\left ( y_{0}-t_{0}\right ) +h\left ( y_{0}-t_{0}\right ) +t_{1}\\ & =\left ( 1+h\right ) \left ( y_{0}-t_{0}\right ) +t_{1} \end{align*}

Now the question will be answered. Need to show that \(y_{n}=\left ( 1+h\right ) ^{n}\left ( y_{0}-t_{0}\right ) +t_{n}\) is true, using induction.  This is true for \(k=1\) as shown above. Now assuming it is true for \(k\), we then need to show it is true for \(k+1\).

By assumption, it is true for \(k\), hence\begin{equation} y_{k}=\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k}\tag{1} \end{equation} But using Euler formula\begin{align} y_{k+1} & =y_{k}+hf\left ( t_{k},y_{k}\right ) \nonumber \\ & =y_{k}+h\left ( 1-t_{k}+y_{k}\right ) \tag{2} \end{align}

Substituting (1) into RHS of (2)\begin{align*} y_{k+1} & =\left ( \left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k}\right ) +h\left ( 1-t_{k}+\left ( \left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k}\right ) \right ) \\ & =\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k}+h-ht_{k}+h\left ( \left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k}\right ) \\ & =\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k}+h-ht_{k}+h\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +ht_{k}\\ & =\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k}+h+h\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) \end{align*}

But \(t_{k}+h=t_{k+1}\) by definition, hence\begin{align*} y_{k+1} & =\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) +t_{k+1}+h\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) \\ & =\left ( 1+h\right ) ^{k}\left ( y_{0}-t_{0}\right ) \left ( 1+h\right ) +t_{k+1}\\ & =\left ( 1+h\right ) ^{k+1}\left ( y_{0}-t_{0}\right ) +t_{k+1} \end{align*}

The above shows it is true for \(k+1\) given it is true for \(k\). Therefore, it is true for any positive integer \(n\).

2.4.7.4 Part d

Using

\[ y_{n}=\left ( 1+h\right ) ^{n}\left ( y_{0}-t_{0}\right ) +t_{n}\]

Replacing \(h=\frac{t_{n}-t_{0}}{n}\) in the above gives\[ y_{n}=\left ( 1+\left ( \frac{t_{n}-t_{0}}{n}\right ) \right ) ^{n}\left ( y_{0}-t_{0}\right ) +t_{n}\] Taking the limit\[ \lim _{n\rightarrow \infty }y_{n}=\lim _{n\rightarrow \infty }\left ( 1+\left ( \frac{t_{n}-t_{0}}{n}\right ) \right ) ^{n}\left ( y_{0}-t_{0}\right ) +\lim _{n\rightarrow \infty }t_{n}\] But \(\lim _{n\rightarrow \infty }t_{n}=t\), hence replacing all \(t_{n}\) with \(t\) in the above gives\[ \lim _{n\rightarrow \infty }y_{n}=\lim _{n\rightarrow \infty }\left ( 1+\left ( \frac{t-t_{0}}{n}\right ) \right ) ^{n}\left ( y_{0}-t_{0}\right ) +t \] Using hint that \(\lim _{n\rightarrow \infty }\left ( 1+\frac{a}{n}\right ) ^{n}=e^{a}\) the above simplifies to\begin{align*} y\left ( t\right ) & =\lim _{n\rightarrow \infty }y_{n}\\ & =e^{\left ( t-t_{0}\right ) }\left ( y_{0}-t_{0}\right ) +t \end{align*}

Which is the analytical solution found in part (a).

2.4.8 Section 3.1 problem 1

Find the general solution to \(y^{\prime \prime }+2y^{\prime }-3y=0\).

This is second order, linear, constant coefficient ODE. Letting \(y=e^{rt}\) and replacing this into the ODE gives\[ e^{rt}\left ( r^{2}+2r-3\right ) =0 \] Since \(e^{rt}\neq 0\), the above reduces to what is called the characteristic equation of the ODE\[ \fbox{$r^2+2r-3=0$}\] Which can be written as \(\left ( r-1\right ) \left ( r+3\right ) =0\). Hence \(r_{1}=1,r_{2}=-3\). Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}e^{t}+c_{2}e^{-3t}\]

2.4.9 Section 3.1 problem 2

Find the general solution to \(y^{\prime \prime }+3y^{\prime }+2y=0\).

This is second order, linear, constant coefficient ODE. The characteristic equation of the ODE\[ \fbox{$r^2+3r+2=0$}\] Which can be written as \(\left ( r+1\right ) \left ( r+2\right ) =0\). Hence \(r_{1}=-1,r_{2}=-2\). Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}e^{-t}+c_{2}e^{-2t}\]

2.4.10 Section 3.1 problem 3

Find the general solution to \(6y^{\prime \prime }-y^{\prime }-y=0\).

This is second order, linear, constant coefficient ODE. The characteristic equation of the ODE\[ \fbox{$6r^2-r-1=0$}\] Hence \(r=\frac{-b\pm \sqrt{b^{2}-4ac}}{2a}=\frac{1\pm \sqrt{1-4\left ( 6\right ) \left ( -1\right ) }}{12}=\frac{1\pm \sqrt{1+24}}{12}=\frac{1\pm 5}{12}\). Hence \(r_{1}=\frac{1}{2},r_{2}=\frac{-1}{3}\). Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}e^{\frac{1}{2}t}+c_{2}e^{\frac{-1}{3}t}\]

2.4.11 Section 3.1 problem 4

Find the general solution to \(2y^{\prime \prime }-3y^{\prime }+y=0\).

This is second order, linear, constant coefficient ODE. The characteristic equation of the ODE\[ \fbox{$2r^2-3r+1=0$}\] Hence \(r=\frac{-b\pm \sqrt{b^{2}-4ac}}{2a}=\frac{3\pm \sqrt{9-4\left ( 2\right ) \left ( 1\right ) }}{4}=\frac{3\pm 1}{4}\). Hence \(r_{1}=1,r_{2}=\frac{1}{2}\). Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}e^{t}+c_{2}e^{\frac{1}{2}t}\]

2.4.12 Section 3.1 problem 5

Find the general solution to \(y^{\prime \prime }+5y^{\prime }=0\).

This is second order, linear, constant coefficient ODE. The characteristic equation of the ODE\[ \fbox{$r^2+5r=0$}\] Which can be written as \(r\left ( r+5\right ) =0\), hence \(r_{1}=0,r_{2}=-5.\)Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}+c_{2}e^{-5t}\]

2.4.13 Section 3.1 problem 6

Find the general solution to \(4y^{\prime \prime }-9y=0\).

This is second order, linear, constant coefficient ODE. The characteristic equation of the ODE\[ \fbox{$4r^2-9=0$}\] Therefore \(r^{2}=\frac{9}{4}\) or \(r=\pm \sqrt{\frac{9}{4}}=\pm \frac{3}{2}\). Hence \(r_{1}=\frac{3}{2},r_{2}=-\frac{3}{2}.\)Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}e^{\frac{3}{2}t}+c_{2}e^{-\frac{3}{2}t}\]

2.4.14 Section 3.1 problem 7

Find the general solution to \(y^{\prime \prime }-9y^{\prime }+9y=0\).

This is second order, linear, constant coefficient ODE. The characteristic equation of the ODE is\[ \fbox{$r^2-9r+9=0$}\] Hence \(r=\frac{-b\pm \sqrt{b^{2}-4ac}}{2a}=\frac{9\pm \sqrt{81-4\left ( 1\right ) \left ( 9\right ) }}{2}=\frac{9\pm \sqrt{81-36}}{2}=\frac{9\pm \sqrt{45}}{2}=\frac{9\pm 3\sqrt{5}}{2}\). Hence \(r_{1}=\frac{9+3\sqrt{5}}{2},r_{2}=\frac{9-3\sqrt{5}}{2}\). Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}e^{\frac{9+3\sqrt{5}}{2}t}+c_{2}e^{\frac{9-3\sqrt{5}}{2}t}\]

2.4.15 Section 3.1 problem 8

Find the general solution to \(y^{\prime \prime }-2y^{\prime }-2y=0\).

This is second order, linear, constant coefficient ODE. The characteristic equation of the ODE is\[ \fbox{$r^2-2r-2=0$}\] Hence \(r=\frac{-b\pm \sqrt{b^{2}-4ac}}{2a}=\frac{2\pm \sqrt{4-4\left ( 1\right ) \left ( -2\right ) }}{2}=\frac{2\pm \sqrt{4+8}}{2}=\frac{2\pm \sqrt{12}}{2}=\frac{2\pm 2\sqrt{3}}{2}=1\pm \sqrt{3}\). Hence \(r_{1}=1+\sqrt{3},r_{2}=1-\sqrt{3}\). Therefore the solution is\[ y\left ( t\right ) =c_{1}e^{r_{1}t}+c_{2}e^{r_{2}t}\] Where \(c_{1},c_{2}\) are constants which can be found from initial conditions. Hence the general solution is\[ y\left ( t\right ) =c_{1}e^{\left ( 1+\sqrt{3}\right ) t}+c_{2}e^{\left ( 1-\sqrt{3}\right ) t}\]