<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head><body style="overflow-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;">Most of my dissertation (1968) was on numerical solution of potential problems. One of the parts was a proof that some of the known iterative methods converged. The argument loosely went something like this. Consider the 2D Poisson equation on a square. If you use an N x N approximation with the usual discretization of the Laplacian<div><br></div><div>u_ij = (u_i(j-1) + u_i(j+1) + u_(i_1)j + i_(j+1))/4 </div><div><br></div><div>i.e, the average of the surrounding points, the problem reduces to the solution of a set of N^2 linear equations</div><div><br></div><div>Ax = b </div><div><br></div><div>where x in a vector of the unknown {u_ij} arranged by rows or columns, b is determined by the boundary conditions and the right side of the Poisson equation. The interesting part is A which is block tridiagonal. With only a small error A can be made block cyclic. You can then diagonalize A with a sine transform and I was able to use that for proofs.</div><div><br></div><div>A few years later when the FFT came about, we realized that we could use the FFT to do the sine transform and the resulting numerical method was as least as efficient as any other method people had come up with.</div><div><br></div><div>Ed</div><div><br></div><div>Here’s a reference from 1986 that I think was based on paper at a Bellman Continuum</div><div><br></div><div>``From Dynamic Programming to Fast Transforms,'' E. Angel, J. Math. Anal. Appl.,119,1986. </div><div><br></div><div>Ed<br><div>
<div>__________<br><br>Ed Angel<br><br>Founding Director, Art, Research, Technology and Science Laboratory (ARTS Lab)<br>Professor Emeritus of Computer Science, University of New Mexico<br><br>1017 Sierra Pinon<br>Santa Fe, NM 87501<br>505-984-0136 (home)<span class="Apple-tab-span" style="white-space: pre;"> </span> <span class="Apple-tab-span" style="white-space: pre;"> </span>edward.angel@gmail.com<br>505-453-4944 (cell) <span class="Apple-tab-span" style="white-space: pre;"> </span><span class="Apple-tab-span" style="white-space: pre;"> </span>http://www.cs.unm.edu/~angel<br></div>
</div>
<div><br><blockquote type="cite"><div>On Apr 28, 2023, at 8:18 AM, Stephen Guerin <stephen.guerin@simtable.com> wrote:</div><br class="Apple-interchange-newline"><div><div dir="ltr">Special Unitary Groups and Quaternions <br><br>Mostly for Ed from the context of last week's Physical Friam if you're coming today.<br><br>Discussion was around potential ways of visualizing the dynamics of SU(3), SU(2), (SU1) that highlights Special Unitary Groups. (<a href="https://en.wikipedia.org/wiki/Special_unitary_group">wiki link from Frank</a>), and can we foreground how quaternions are used in this process.<br><br>and a related bit on forces, I'm searching for ways to visualize/understand how <a href="https://www.codeproject.com/Articles/5308623/Solving-Poisson-Equation">FFTs with Poisson equation</a> are used to compute the forces from scalar fields (eg gravitational force from mass density, electric force from charge, etc) and if there's any relation to Special Unitary Groups.<br><div><br></div><div>-S</div></div>
-. --- - / ...- .- .-.. .. -.. / -- --- .-. ... . / -.-. --- -.. .<br>FRIAM Applied Complexity Group listserv<br>Fridays 9a-12p Friday St. Johns Cafe / Thursdays 9a-12p Zoom https://bit.ly/virtualfriam<br>to (un)subscribe http://redfish.com/mailman/listinfo/friam_redfish.com<br>FRIAM-COMIC http://friam-comic.blogspot.com/<br>archives: 5/2017 thru present https://redfish.com/pipermail/friam_redfish.com/<br> 1/2003 thru 6/2021 http://friam.383.s1.nabble.com/<br></div></blockquote></div><br></div></body></html>