Captured On
[2019-10-29 Tue 13:41]
Source
Linear Operators, Linear Time Invariance | Unit II: Second Order Constant Coefficient Linear Equations | Differential Equations | Mathematics | MIT OpenCourseWare

1 What is the name of this symbol \(p(D)\)?

1.1 Front

What is the name of this symbol $p(D)$?

Where \(D\) is a differential operator

1.2 Back

\(p(D)\) is a polynomial operator

2 What is a polynomial differential operator with constant coefficients?

2.1 Front

What is a polynomial differential operator with constant coefficients?

2.2 Back

\(p(D)\) that you can apply to any function (sufficiently differentible)

\(p(D) = a_0 D^n + a_1 D^{n-1} + \dots + a_n\)

3 What is the sum rule for differential operator \(D\)?

3.1 Front

What is the sum rule for differential operator $D$?

3.2 Back

Suppose that functions involved are sufficiently differentiable

Where \(p(D)\) and \(q(D)\) are polynomial operators

\([p(D) + q(D)]u = p(D)u + q(D)u\)

4 What is the general linear ODE of order \(n\) written with operators?

4.1 Front

What is the general linear ODE of order $n$ written with operators?

4.2 Back

It must have constant coefficients.

\(y^{(n)} + a_1 y^{(n-1)} + \dots + a_n y = q(t)\)

where \(p(D)\) is a polynomial differential operator with constant coefficients, \(p(D) = D^n + a_1 D^{n-1} + \dots + a_n\)

\(p(D) y = q(t)\)

5 What is the linearity rule for differential operator?

5.1 Front

What is the linearity rule for differential operator?

5.2 Back

If \(f\) and \(g\) are functions and \(c_1\) and \(c_2\) are constants,

\({\displaystyle p(D)(c_1 f + c_2 g) = c_1 p(D)f + c_2 p(D)g}\)

6 Can we apply the linearity rule of differential operator of any order?

6.1 Front

Can we apply the linearity rule of differential operator of any order?

\(D^n(c_1 f + c_2 g)\)

6.2 Back

Yes, \({\displaystyle D^n(c_1 f + c_2 g) = c_1 D^n f + c_2 D^n g}\)

7 What does mean this equation?

7.1 Front

What does mean this equation?

If \(f\) and \(g\) are functions and \(c_1\) and \(c_2\) are constants,

\(aD^n(c_1 f + c_2 g)\), \(a\) could be a constant or a function of \(t\)

7.2 Back

\(a \dv[n]{t} (c_1 f + c_2 g) = c_1 a f^{(n)} + c_2 a g^{(n)} = c_1 a D^n f + c_2 a D^n g\)

8 What does means this expression?

8.1 Front

What does means this expression?

\(p(D)u\), where \(p(D) = g(D)h(D)\)

8.2 Back

\(D^m(a D^k u) = a D^{m+k} u\), which could be extends to general polynomial operators \(h(D)\) by linearity.

In this case \(a\) must be a constant

9 Is this expression correct?

9.1 Front

Is this expression correct?

\(g(D)(h(D) u) = h(D)(g(D) u)\)

9.2 Back

It’s true if \(g(D)\) and \(h(D)\) are polynomial operators with constant coefficients for every function \(u(t)\)

10 What is the substitution rule of polynomial differential operators?

10.1 Front

What is the substitution rule of polynomial differential operators?

10.2 Back

\(p(D) e^{at} = p(a) e^{at}\)

11 Expand this expression?

11.1 Front

Expand this expression?

\(p(D)e^{at} u(t)\)

11.2 Back

\(e^{at} p(D+a)u(t)\), by the exponential-shift rule

12 Expand this expression?

12.1 Front

Expand this expression?

\(D^3 e^{-t} \sin(t)\)

Using several methods

12.2 Back

Solution using the exponential-shift rule (and binomial theorem)

\begin{align*} D^3 e^{-t} \sin(t) &= e^{-t}(D-1)^3 \sin(t)\\\ &= e^{-t}(D^3 -3D^2 + 3D -1) \sin(t) \\\ &= e^{-t}(2 \cos(t) + 2 \sin(t)) \end{align*}

Solution using the substitution rule

We can write \(e^{-t} \sin(t) = \operatorname{Im}(e^{(-1 + i)t})\)

\begin{align*} D^3e^{(-1 + i)t} &= (-1 + i)^3 e^{(-1 + i)t} \\\ &= (2+2i)e^t(\cos(t) + i \sin(t))\\\ &= e^t(2 \cos(t) + 2 \sin(t)) \text{ Getting the imaginary part} \end{align*}

13 Can we use this expansion if \(c\) and \(a\) are complex numbers?

13.1 Front

Can we use this expansion if $c$ and $a$ are complex numbers?

\(D(c e^{at}) = c a e^{at}\)

And the others rules of differential operator

13.2 Back

Yes, you can. Also, you can apply the rules and arguments for differential operators

14 What is a linear time invariant “LTI”?

14.1 Front

What is a linear time invariant “LTI”?

14.2 Back

On linear ODE \(p(D)x = q(t)\) where \(p(D)\) is a constant coefficient operator, you can apply the time invariance of \(p(D)\) and you can say for another ODE \(p(D)y = q(t-c)\) that its particular solution is \(y(t) = x(t-c)\)

15 In which cases can I say that \(y(t) = x(t - c)\) for …

15.1 Front

In which cases can I say that $y(t) = x(t - c)$ for …

the ODE \(p(D) y = q(t-c)\) and \(y(t)\) is a solution of \(p(D)y = q(t)\)

15.2 Back

In case that \(p(D)\) is a constant coefficient operator

It’s called Linear Time Invariant operator

16 Which guess solution could be used for getting the particular solution?

16.1 Front

Which guess solution could be used for getting the particular solution?

\(y’’ + y’ + y = 2xe^x\)

16.2 Back

You can try \(y_p = a_1 xe^x + a_2 e^x\)

  • \(y_p’ = a_1(x+1) e^x + a_2 e^x\)
  • \(y_p’’ = a_1(x+2) e^x + a_2e^x\)
  • \(2xe^x = (a_1(x+2) e^x + a_2e^x) + (a_1(x+1) e^x + a_2 e^x) + (a_1 xe^x + a_2 e^x) = 3a_1xe^x + (3a_1 + 3a_2)e^x\)
    • \(3a_1 = 2 \implies a_1 = 2/3\)
    • \((3a_1 + 3a_2) = 0 \implies 2 + 3 a_2 \implies a_2 = -2/3\)

17 Which guess solution could be used for getting the particular solution?

17.1 Front

Which guess solution could be used for getting the particular solution?

\(y^{(4)} - 2 y’’ + y = xe^x\)

17.2 Back

  • \(p(s) = s^4 -2 s^2 + 1 = (s^2 - 1)^2\)
  • so \(p(1)=0\) repeated root (order 2)
  • guess: \(y_p = x^2(Ax + B)e^{x}\)

18 Graph the amplitude in terms of \(\omega\) of this ODE

18.1 Front

Graph the amplitude in terms of $\omega$ of this ODE

\(x’’ + \omega_0^2 x = F_0 \cos(\omega t)\)

Graph also the response when \(\omega = \omega_0\)

18.2 Back

Solving the ODE with complex replacement, characteristic polynomial and Exponential Response Formula (ERF)

  • \({\displaystyle x_p = \frac{F_0 \cos(\omega t)}{\omega_0^2 - \omega^2}}\) when \(\omega_0 \neq \omega\)
  • \({\displaystyle x_p = \frac{F_0 t \sin(\omega_0 t)}{2 \omega_0}}\) when \(\omega_0 = \omega\)

\({\displaystyle A(\omega) = \abs{\frac{F_0}{\omega_0^2 - \omega_2}}}\)

19 What happens when \(\omega\) are really close to \(\omega_{0}\) in this system?

19.1 Front

What happens when $\omega$ are really close to $\omega_{0}$ in this system?

\((D^2 + \omega_0^2)x = F_0 \cos(\omega t)\)

ERF: \({\displaystyle x_p(t) = \frac{F_0 \cos(\omega t)}{\omega_0^2 - \omega^2}}\)

Study how is the response when \(\omega \to \omega_0\)

19.2 Back

In this case the amplitude of the output is bigger than the amplitude of the input, but the period is the same.

When \(\omega = \omega_0\), it the case on pure resonance, and the RRF is completely different \({\displaystyle x_p(t) = \frac{F_0 t \sin(\omega t)}{2 \omega_0}}\)

We can analyse what happens with another particular solution (getting a part from the complementary solution associated to homogeneous model), and study it what happens when \(\omega \to \omega_0\)

\({\displaystyle x_p(t) = \frac{F_0 \cos(\omega t)}{\omega_0^2 - \omega^2} - \frac{F_0 \cos(\omega_0 t)}{\omega_0^2 - \omega^2}}\)

\({\displaystyle \lim_{\omega \to \omega_0} x_p = \lim_{\omega \to \omega_0} \frac{- F_0 t \sin(\omega t)}{- 2 \omega} = \frac{F_0 t \sin(\omega_0 t)}{2 \omega_0}}\)

This is the response to the resonance.

20 How is the graph of this solution when \(\omega \to \omega_{d}\)

20.1 Front

How is the graph of this solution when $\omega \to \omega_{d}$

\({\displaystyle x_p(t) = \frac{F_0 \cos(\omega t)}{\omega_0^2 - \omega^2} - \frac{F_0 \cos(\omega_0 t)}{\omega_0^2 - \omega^2}}\)

20.2 Back

You need to use the trigonometric identify \(\cos(B) - \cos(A) = 2 \sin(\frac{A-B}{2}) \sin(\frac{A+B}{2})\)

For this case \({\displaystyle \cos(\omega t) - \cos(\omega_0 t) = 2 \sin(\frac{t(\omega - \omega_0)}{2}) \sin(\frac{t(\omega + \omega_0)}{2})}\)

As \(\omega \to \omega_{0}\),

\({\displaystyle \frac{F_0 \sin(\frac{t(\omega - \omega_0)}{2})}{\omega_0^2 - \omega^2}}\), describe a very low frequency function with really big amplitude.

And inside, there is the function \(\sin(t \omega_0)\)

21 How can we proof that this particular solution is valid for this system?

21.1 Front

How can we proof that this particular solution is valid for this system?

\(x_p(t) = \frac{t e^{at}}{p’(a)}\), is a solution of \(p(D)x = e^{at}\) and \(p(a) = 0\)

21.2 Back

We can write \(p(D) = q(D)(D - a)\), where \(q(a) \neq 0\) because of \(p(D)\) could be factor by \((D - a)\)

So, \(p’(D) = q’(D)(D-a) + q(D)\), so \(p’(a) = q’(a)(a - a) + q(a) = q(a)\)

\begin{align*} p(D)x_p(t) &= p(D) \frac{t e^{at}}{p’(a)} \\\ &= \frac{e^{at}}{p’(a)} p(D + a)t \\\ &= \frac{e^{at}}{p’(a)} q(D + a) D t \\\ &= \frac{e^{at}}{q(a)} q(D + a) 1 \\\ &= \frac{e^{at}}{q(a)} q(a) = e^{at} \end{align*}

22 Which is the value of this expression?

22.1 Front

Which is the value of this expression?

\(p(D + a)1\), where \(a\) is a constant

22.2 Back

\(p(a)\)