Conversation started Jan 30, 2015 at 23:36.
Jan 30, 2015 23:36
The typical way to diagonalize a matrix is to first tridiagonalize it (or approximate it by a tridiagonal matrix), and then compute the eigenvalues of the tridiagonal matrix by some method.
@ThomasKlimpel the thing is the resources I am using don't explain it well..Is the characteristic poylnomial the way to do this?
The critical part is normally the tridiagonalization, because computing the eigenvalues of a tridiagonal matrix is not too expensive anyway.
@ThomasKlimpel examples?
@ThomasKlimpel oh BTW, thank you for helping me... others just don't have the time to help...
Orthogonal polynomials are a typical examples of this procedure. Here the operator which is tridiagonalized is the multiplication by x. The tridiagonalization translates into a three term recurence formula here.
@ThomasKlimpel any good links please?
vzn
vzn
Jan 30, 2015 23:42
TA this is advanced stuff. why not get a textbook?
bye..
@vzn like what?
vzn
vzn
its "linear algebra". afaik gaussian elimination will work on the problem.
The Lanczos algorithm is an iterative algorithm devised by Cornelius Lanczos that is an adaptation of power methods to find the most useful eigenvalues and eigenvectors of an order linear system with a limited number of operations, , where is much smaller than . Although computationally efficient in principle, the method as initially formulated was not useful, due to its numerical instability. In 1970, Ojalvo and Newman showed how to make the method numerically stable and applied it to the solution of very large engineering structures subjected to dynamic loading. This was achieved using a method...
vzn
vzn
there are a bazillion textbooks on it.
@vzn that's what I heard from my math professor...
vzn
vzn
Jan 30, 2015 23:46
& why not believe him? & me? :p
@vzn I know its linear algebra, a lot of QM is linear algebra...
vzn
vzn
exactly, was just about to say that myself
@vzn I would like to try all routes possible.. don't get me wrong...I am not trying to take a shortcut...just want multiple opinions, u know..
vzn
vzn
you might like to play with some kind of simulator or solver, think some are out there.
@vzn really? any names?
@ThomasKlimpel looks like the method you were describing.. thanks...
@ThomasKlimpel looks like the method you were describing.. thanks...
vzn
vzn
Jan 30, 2015 23:48
uh hunted for simulators years ago, read about a few... it depends on the purpose....
your goals are not clear right now.
@vzn any example search terms? maybe "QM simulator" ?
vzn
vzn
lol exactly
@vzn I thought it was clear.. I am trying to answer the question that I asked.. the "Plants and Quantum Mechanics" one...
vzn
vzn
this looks pretty good for QM basics
dude your plants & QM question cannot be answered completely even by experts.
@vzn so I should be looking qm simulators for lattices?
@vzn lol... I know, I want to answer as best as I can or look into the leading theories...
vzn
vzn
Jan 30, 2015 23:51
you should be learning QM basics before jumping into lattices, probably, as the guys over on the physics site are recommending.
get a good QM book.
did you say you have one?
@vzn cool, might learn quantum mechanics from this website too... I like visual stuff, u know... there's even spin chains, what MM was trying to tell me...
vzn
vzn
cool
@vzn yep...
vzn
vzn
QM is a vast subject.
u know what.. whatever...
Jan 30, 2015 23:53
@TAbraham Maybe one additional comment: The Lanczos algorithm is normally what is at work in the cases where you can find an analytic solution. For your Hamiltonian however, it seems that the Lanczos algorithm will take n^3 operations, so that there is no advantage over a purely numerical algorithm like
Der QR-Algorithmus ist ein numerisches Verfahren zur Berechnung aller Eigenwerte und eventuell der Eigenvektoren einer quadratischen Matrix. Das auch QR-Verfahren oder QR-Iteration genannte Verfahren basiert auf der QR-Zerlegung und wurde im Jahre 1961–1962 unabhängig voneinander von John G. F. Francis und Wera Nikolajewna Kublanowskaja eingeführt. Ein Vorläufer war der LR-Algorithmus von Heinz Rutishauser (1958), der aber weniger stabil ist und auf der LR-Zerlegung basiert. Oft konvergieren die Iterierten aus dem QR-Algorithmus gegen die Schur-Form der Matrix. Das originale Verfahren ist recht…
@ThomasKlimpel note, it's in German i think..
vzn
vzn
re TKs notes. all relevant but diagonalization algorithms are all implemented in basic math pkgs. you might start playing with those. did the physics guys mention that yet? octave (similar to matlab) is free for example.
In numerical linear algebra, the QR algorithm is an eigenvalue algorithm: that is, a procedure to calculate the eigenvalues and eigenvectors of a matrix. The QR transformation was developed in the late 1950s by John G.F. Francis (England) and by Vera N. Kublanovskaya (USSR), working independently. The basic idea is to perform a QR decomposition, writing the matrix as a product of an orthogonal matrix and an upper triangular matrix, multiply the factors in the reverse order, and iterate. == The practical QR algorithm == Formally, let A be a real matrix of which we want to compute the eigenvalues...
vzn
vzn
theres also a lot of matrix algebra in some python libraries. TK have you ever used anything like that? do you use a math app/ pkg?
Intel MKL is fast, so I use that. Ever heard of boost.uBlas.bindinds? Well, they adapt LAPACK/Intel MKL to normal C++ language.
vzn
vzn
Jan 30, 2015 23:57
yes heard of "BOOST" years ago.
@vzn builtin?
vzn
vzn
there is probably even a QM related python library lying around somewhere, maybe heard of one once....
aha look at that
qutip QuTiP : Quantum Toolbox in Python
2
@ThomasKlimpel thanks...any programs that do this algorithim?
@vzn let's rejoice!! I am a python programmer, so a python module would be perfect for me... THANKS SO MUCH!!!! have to look into it...
octave/scilab/matlab are probably the easiest ones in this direction, but there are many many more options.
vzn
vzn
=) it looks cool.
maybe not so hard to implement the hamiltonian setup youre interested in, after learning all the framework, for an expert....
Jan 31, 2015 00:00
@ThomasKlimpel what's octave? anything online or python?
@ThomasKlimpel what's that?
finally, we all three talk together... I have been waiting for this..lol!
vzn
vzn
dont worry TK will probably bail soon :p
Intel® Math Kernel Library (Intel® MKL) accelerates math processing routines that increase application performance and reduce development time. Intel® MKL includes highly vectorized and threaded Linear Algebra, Fast Fourier Transforms (FFT), Vector Math and Statistics functions.
@ThomasKlimpel like matlab?
vzn
vzn
TA, not much need to worry about performance right now probably. MKL is highly optimized linear algebra. there is a lot of effort to highly optimize it on various platforms.
as a python programmer, maybe this may help?:
@vzn might look into that as well...
Jan 31, 2015 00:04
octave is mainly a direct matlab clone. scilab is a bit more independent with qualities of its own
 
Conversation ended Jan 31, 2015 at 0:05.