[University Differential Equations] Calculating the exponential of a Matrix

Maybe you meant to say that (i&plusmn;1,&plusmn;1) is an eigenvector for the eigenvalue &plusmn;i; the + case works, but the &minus; case for your solution does not work, because

* i&minus;1&minus;2=i&minus;3&ne;&minus;i\*(i&minus;1) and i&minus;1&minus;1=i&minus;2&ne;&minus;i\*1, while
* i&minus;1+2=i+1=&minus;i\*(i&minus;1) and i&minus;1+1=i=&minus;i\*(&minus;1).

With that said, Symbolab's solution looks like yours, so it looks as if your work following the explicit eigenvectors went as if you had written them out correctly.

---
An alternative way to find the matrix exponential, the one that I learned, is to first diagonalize the matrix (which you can do in this case because the matrix has a full eigenbasis, because its eigenvalues are distinct), then exponentiate the diagonals, then multiply out; this works because the power series for exponentiation is also valid for matrices, although *why* it's valid is a bit beyond the scope of your course, and then for the diagonalizable case, the P^(&minus;1)P factors cancel out in each term, leaving terms of the form PD^(n)P^(&minus;1) for each whole number n, and then the P and P^(&minus;1) terms can be factored out, each power of a diagonal matrix is the matrix with its diagonal entries raised to that power, and so it reduces to ordinary exponentiation.

Okay, in case you haven't diagonalized a matrix before, the P here is a matrix with each eigenvector as a column, and then the D is a diagonal matrix with the corresponding eigenvalue as the corresponding diagonal entry:

P=⌈i+1 i−1⌉ D=⌈i  0⌉
⌊ 1  −1 ⌋   ⌊0 −i⌋

The idea is that, say, (i+1,1) will be transformed to (1,0), then to (i,0), then to i\*(i+1,1).

Then det(P)=&minus;2i, so

P^−1=⌈−½i ½i+½⌉
⌊−½i ½i−½⌋

and P^(&minus;1) does work as expected.

Then the diagonal elements of e^(Dt) are e^(it) and e^(&minus;it), and

Pe^(Dt)=⌈(i+1)e^(it) (i−1)e^(−it)⌉
⌊   e^(it)     −e^(−it)  ⌋

and finally,

e^(At)=Pe^(Dt)P^(−1)=⌈−½(i−1)e^(it)+½(i+1)e^(−it)      ie^(it)−ie^(−it)     ⌉
⌊    −½ie^(it)+½ie^(−it)     ½(i+1)e^(it)−½(i−1)e^(−it)⌋

This also agrees with Symbolab.

0 like 0 dislike