Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Find the general solution of this system of equations:

$$x' = \pmatrix{-1&0&0\\1&0&-1\\1&1&0}x$$

I got the eigenvalues to be: $\lambda = -1,\pm i$

The eigenvectors corresponding to each eigenvalue:

$v_1 = \pmatrix{1\\-1\\0}$, $v_2 = \pmatrix{0\\i\\1}$, $v_3 = \pmatrix{0\\-i\\1}$

But then how can I write the general solution?

share|cite|improve this question
Take $v_1 e^t+\mathrm{Re}\,[v_2 e^{ix}]+\mathrm{Im}\,[v_2 e^{ix}]$ – Artem Dec 7 '12 at 5:45
@Artem I know how $v_1$ will look, and $v_2$, $v_3$ exponential function will equal 1 so I dont have to worry about $e^{ix}$. But how can I plug in the weird symbols you have into my eigenvectors? – Q.matin Dec 7 '12 at 5:49
You should read about real and imaginary parts of a complex number. (and of course instead of $x$ should be $t$ in my comment above) – Artem Dec 7 '12 at 5:51
@Artem I see you changed your symbols, what does Re and Im mean? – Q.matin Dec 7 '12 at 5:57
up vote 1 down vote accepted

I left a comment yesterday which had some typos, and still contains errors, therefore, I will post a full answer to the original question.

If you have a linear system of three equations this means that you need to find three linearly independent solutions which would form a basis of the set of solutions. If you have three distinct eigenvalues $\lambda_1,\lambda_2,\lambda_3$ and three corresponding eigenvectors $v_1,v_2,v_3$ then the general solution is $$ y=c_1v_1e^{\lambda_1t}+c_2v_2e^{\lambda_2t}+c_3v_3e^{\lambda_3t}, $$ where $c_1,c_2,c_3$ are arbitrary constants. This is a correct solution, however in many cases you are asked to find a real-valued solution if the original system has real coefficients. In this case you can use the real and imaginary parts of any of the solutions, e.g., $$ y=c_1v_1e^{\lambda_1t}+c_2\mathrm{Re}\,[v_2e^{\lambda_2t}]+c_3\mathrm{Im}\,[v_2e^{\lambda_2t}]. $$ In your case you have $$ v_2e^{\lambda_2 t}= \begin{bmatrix} 0\\ i\\ 1 \end{bmatrix}e^{it}= \begin{bmatrix} 0\\ i\\ 1 \end{bmatrix}(\cos t+i\sin t)= \begin{bmatrix} 0\\ -\sin t+i\cos t\\ \cos t+i\sin t \end{bmatrix}= \begin{bmatrix} 0\\ -\sin t\\ \cos t \end{bmatrix}+i\begin{bmatrix} 0\\ \cos t\\ \sin t \end{bmatrix}. $$ Therefore $$ \mathrm{Re}\,[v_2e^{\lambda_2 t}]=\begin{bmatrix} 0\\ -\sin t\\ \cos t \end{bmatrix},\quad \mathrm{Im}\,[v_2e^{\lambda_2 t}]=\begin{bmatrix} 0\\ \cos t\\ \sin t \end{bmatrix}. $$ Putting everything together $$ y(t)=c_1\begin{bmatrix} 1\\ -1\\ 0 \end{bmatrix}e^{-t}+c_2 \begin{bmatrix} 0\\ -\sin t\\ \cos t \end{bmatrix}+c_3\begin{bmatrix} 0\\ \cos t\\ \sin t \end{bmatrix}, $$ which is exactly the answer you have in your comments.

share|cite|improve this answer
Thanks a lot Artem! This is what I was looking for but I am confused on one last thing, is $c_3$ the set of imaginary numbers right and $c_2$ isn't the set of imaginary numbers? And why would the solution set confuse me like that by splitting it up into two parts in $c_2$ and $c_3$? What was the point of splitting it up? – Q.matin Dec 7 '12 at 21:51
Ok, $c_1,c_2,c_3$ from the first line have nothing to do with $c_1,c_2,c_3$ in the last line. In the last line these are just arbitrary real constants. – Artem Dec 7 '12 at 23:38
Thats not what I was trying to say, let me clarify. Why did the solution set have this as the solution:$$x(t) = c_1e^{-t}\pmatrix{1\\-1\\0} = c_2[cos(t)\pmatrix{0\\0\\1} -sin(t)\pmatrix{0\\1\\0}] +c_3[sin(t)\pmatrix{0\\0\\1}+cos(t)\pmatrix{0\\1\\0}]$$ The way they written it is very confusing but the way you wrote it makes more sense to me. $$y(t)=c_1\begin{bmatrix} 1\\ -1\\ 0 \end{bmatrix}e^{-t}+c_2 \begin{bmatrix} 0\\ -\sin t\\ \cos t \end{bmatrix}+c_3\begin{bmatrix} 0\\ \cos t\\ \sin t \end{bmatrix}$$. Why did they write it like that instead of how you written it? What is the difference? – Q.matin Dec 8 '12 at 1:01
There is no difference. Except for the fact that where you have your second $=$ should be $+$. – Artem Dec 8 '12 at 1:27

Write the general solution as

$$ y(t)=c_1 v_1 e^{\lambda_1 t}+c_2 v_1 e^{\lambda_2 t}+c_3 v_3 e^{\lambda_3 t}, $$

where $c_1,c_2,c_3$ are arbitrary constants and simplify the expression. Here is the final solution computed by maple, so you can compare your answer

$$ \left\{ x \left( t \right) ={\it C_3}\,{{\rm e}^{-t}},y \left( t \right) ={\it C_1}\,\cos \left( t \right) -{\it C_2}\,\sin \left( t \right) -{\it C_3}\,{{\rm e}^{-t}},z \left( t \right) ={\it C_1}\, \sin \left( t \right) +{\it C_2}\,\cos \left( t \right) \right\} $$

share|cite|improve this answer
Thanks! The solution is $$x(t) = c_1e^{-t}\pmatrix{1\\-1\\0} = c_2[cos(t)\pmatrix{0\\0\\1} -sin(t)\pmatrix{0\\1\\0}] +c_3[sin(t)\pmatrix{0\\0\\1}+cos(t)\pmatrix{0\\1\\0}]$$ how did they get that? – Q.matin Dec 7 '12 at 6:59
@Q.matin: Write the expression as one vector and simplify it. Note that, you need to introduce new constants in terms of the old ones. – Mhenni Benghorbal Dec 7 '12 at 7:23

Let $V=\frac{1}{\sqrt{2}}\begin{bmatrix} 0 & 0 & 1 \\ 1 & 1 & -1 \\ -i & i & 0 \end{bmatrix}$. Some tedious work shows that $V^{-1}=\frac{1}{\sqrt{2}}\begin{bmatrix} 1 & 1 & i \\ 1 & 1 & -i \\ 2 & 0 & 0 \end{bmatrix}$.

Then $\Lambda = V^{-1} A V = \begin{bmatrix} i & 0 & 0 \\ 0 & -i & 0 \\ 0 & 0 & -1 \end{bmatrix}$. Solving $\dot{y} = \Lambda y$ gives $y(t) = e^{\Lambda (t-t_0)} y(t_0) =\begin{bmatrix} e^{i(t-t_0)} & 0 & 0 \\ 0 & e^{-i(t-t_0)} & 0 \\ 0 & 0 & e^{-(t-t_0)} \end{bmatrix} y(t_0)$.

Since $\dot{y} = \Lambda y$, we have $\dot{Vy} = V\Lambda y$, and letting $x = Vy$, we have $\dot{(V V^{-1} x)} = \dot{x} = V\Lambda V^{-1} x = A x$ (and $x(t_0) = V y(t_0)$, of course). Hence the solution is

$$x(t) = V e^{\Lambda (t-t_0)} V^{-1} x(t_0)$$

Note that $e^{\Lambda (t-t_0)} = e^{i(t-t_0)} e_1 e_1^T + e^{-i(t-t_0)} e_2 e_2^T +e^{-(t-t_0)}e_3 e_3^T$, we can compute $M_i = V e_1 e_1^T V^{-1}$, $M_{-i} = V e_2 e_2^T V^{-1}$ and $M_1 = V e_3 e_3^T V^{-1}$ to get

$$x(t) = (M_i e^{i(t-t_0)} + M_{-i} e^{-i(t-t_0)} + M_1 e^{-(t-t_0)}) x(t_0)$$

Grinding through the details gives $$M_i=\frac{1}{2}\begin{bmatrix} 0 & 0 & 0 \\ 1 & 1 & i \\ -i & -i & 1 \end{bmatrix},\ \ \ M_{-i} = \overline{M_i},\ \ \ M_1=\begin{bmatrix} 1 & 0 & 0 \\ -1 & 0 & 0 \\ 0 & 0 & 0 \end{bmatrix}$$

Combining all of this together gives

$$x(t) = \begin{bmatrix} e^{-(t-t_0)} & 0 & 0 \\ \cos(t-t_0)-e^{-(t-t_0)} & \cos(t-t_0) & -\sin(t-t_0) \\ \sin(t-t_0) & \sin(t-t_0) & \cos(t-t_0) \end{bmatrix} x(t_0)$$

share|cite|improve this answer
Thanks a lot!!! I haven't seen this type of general solution. In the solution set it gives: $$x(t) = c_1e^{-t}\pmatrix{1\\-1\\0} = c_2[cos(t)\pmatrix{0\\0\\1} -sin(t)\pmatrix{0\\1\\0}] +c_3[sin(t)\pmatrix{0\\0\\1}+cos(t)\pmatrix{0\\1\\0}]$$ is that and yours the samething? If not, how did they get that solution I just gave? – Q.matin Dec 7 '12 at 6:58
This would be the same with $x(t_0) = (c_1, c_3-c_1, c_2)^T$. – copper.hat Dec 7 '12 at 7:14
To see this, note that $x(t_0) = \sum_i [x(t_0)]_i e_i$ Then compute $x(t)$ with initial conditions $e_i$ and sum them together. Same answer. – copper.hat Dec 7 '12 at 7:17
Thanks a lot for explaining that to me! – Q.matin Dec 7 '12 at 21:52
I find the matrix approach simpler albeit computationally intensive. – copper.hat Dec 7 '12 at 22:19

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.