Evan Cooch <evan.cooch at gmail.com> writes:
> Greetings --
>
> Consider the following matrix:
>
> a : matrix([-0.7-0.3*h1,0.6-0.6*h1],[0.5-0.5*h2,-0.35-0.65*h2]);
>
> I'm trying to solve for h1 and h2, such that dominant eigenvalue of a is 0.
>
> I know how to do this in a couple of other CAS systems, but am stumped
> with Maxima. I know the solution in thiscase is
>
> h2=(11-81*h1)/(151-21*h1)
>
> Any pointers would be much appreciated (to aid and abet my attempt to
> port a bunch of teaching material to Maxima).
>
> Thanks!
An almost complete answer:
Firstly, let's not use the icky decimals any longer than necessary:
(%i1) a : rat(matrix([-0.7-0.3*h1,0.6-0.6*h1],[0.5-0.5*h2,-0.35-0.65*h2]));
<snip: warnings from rat>
[ 3 h1 + 7 3 h1 - 3 ]
[ - -------- - -------- ]
[ 10 5 ]
(%o1)/R/ [ ]
[ h2 - 1 13 h2 + 7 ]
[ - ------ - --------- ]
[ 2 20 ]
Right, now use eivals to grab the eigenvalues:
(%i2) load("eigen")$
(%i3) evals: first(eivals(a));
2 2
(%o3) [- (sqrt(169 h2 + (324 h1 - 662) h2 + 36 h1 - 396 h1 + 529) + 13 h2
+ 6 h1 + 21)/40,
2 2
sqrt(169 h2 + (324 h1 - 662) h2 + 36 h1 - 396 h1 + 529) - 13 h2 - 6 h1 - 21
-----------------------------------------------------------------------------]
40
The solve() function doesn't seem to have much luck, but to_poly_solve
does much better:
(%i4) load("to_poly_solve")$
*** output flushed ***
(%i5) to_poly_solve(first(evals)=0, h2);
2
%pi 126 h1 + 588 h1 - 3314
(%o5) %union(%if((- --- < parg(- -----------------------))
2 21 h1 - 151
2
126 h1 + 588 h1 - 3314 %pi 81 h1 - 11
%and (parg(- -----------------------) <= ---), [h2 = -----------], %union()))
21 h1 - 151 2 21 h1 - 151
(%i6) to_poly_solve(second(evals)=0, h2);
2
%pi 126 h1 + 588 h1 - 3314
(%o6) %union(%if((- --- < parg(-----------------------))
2 21 h1 - 151
2
126 h1 + 588 h1 - 3314 %pi 81 h1 - 11
%and (parg(-----------------------) <= ---), [h2 = -----------], %union()))
21 h1 - 151 2 21 h1 - 151
Frustratingly, I can't see how to tell Maxima that h1 and h2 are real,
so all the parg values are zero. Also, the %union with only one element
is a bit rubbish. Anyway, I can do one more thing:
(%i7) eq: first(second(first(%o6)));
81 h1 - 11
(%o7) h2 = -----------
21 h1 - 151
Then:
(%i21) radcan(second(evals)), eq;
(%o21) 0
Looking at the picture given by
plot2d(first(evals), [h1, -10, 5]), eq;
it looks like something more complicated is going on with the other
eigenvalue. Maybe you know a bit more about the situation - I'd be
interested to know what the graph is saying.
Rupert
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 315 bytes
Desc: not available
URL: <http://www.math.utexas.edu/pipermail/maxima/attachments/20130102/1d03f8c4/attachment.pgp>;