+- HP Forums (https://www.hpmuseum.org/forum)
+-- Forum: HP Calculators (and very old HP Computers) (/forum-3.html)
+--- Forum: HP Prime (/forum-5.html)
+--- Thread: Eigenvector result is unusual (/thread-1698.html)
Eigenvector result is unusual - Rich - 06-24-2014 12:52 AM
Hi,
When i type in
EIGENVV([1,2],[4,3])
I get this result:
{[[0.4472135955,−0.7453559925],[0.894427191,0.7453559925]],[[5.,0],[0,-1.]]}
I am not sure if my math is wrong somewhere, but on paper i get the different eigenvalues, for 5, i get [1 2] and for -1, I get [1 -1] which is what wolfram alpha gets. My 50G gets a value of [.5 1] and [1 -1] for for 5 and -1 respectively. Which is also correct.
EDIT::: Actually looking a little past the abstract, the results are correct, but I am not sure why the HP Prime didn't come up with a number closer to a whole number or more readable output similar to the 50G.
RE: Eigenvector result is unusual - Helge Gabert - 06-24-2014 01:28 AM
I don't think the math is wrong.
The eigenvalues of lambda1 and lambda2 are correct (5, and -1), and this is also what Wolfram alpha shows.
The eigenvectors are [.44..., .89...], and multiplied by a suitable scaling factor equals [1,2], and ditto for the other eigenvector [-.74..., .74], which, again, multiplied with a suitable scaling factor equals [-1,1] - - again in agreement with Wolfram alpha. I believe you read the vectors the wrong way - - horizontally instead of vertically.
But we can ask - - why doesn't the Prime normalize the eigenvectors?
RE: Eigenvector result is unusual - Rich - 06-24-2014 01:39 AM
Thanks for the update, I was reading it wrong. I noticed that it when diagonalizing the matrix on the Prime, with those eigenvectors, it still gives me the correct result but instead of a zeros, it has -1.42109E-15, which is essentially zero... but maybe due to the original output.
RE: Eigenvector result is unusual - parisse - 06-24-2014 05:27 AM
EIGENVV is an approx instruction. Try an exact one, for example jordan.
RE: Eigenvector result is unusual - Rich - 06-24-2014 11:41 AM
Thank you, the Jordan function works perfectly.