41CL Quiz: Determinant of 30x30 anti-Identity matrix
|
06-05-2018, 07:42 AM
(This post was last modified: 06-05-2018 07:52 AM by Ángel Martin.)
Post: #51
|
|||
|
|||
RE: 41CL Quiz: Determinant of 30x30 anti-Identity matrix
Yes, I'm listening :-D
in fact I've been busy fixing a few glitches in the MCODE as a result of the beta testing, which this test case has also contributed to... Thanks for the example, it is indeed an interesting one (but of course it *had* to be given its source ;-). First off, there's a few remarks to be made:
With those two remarks made, here's how I did it. First, the program used to create the matrices and calculate the determinant: 01 LBL "DDN" - expects n in X-Reg 02 RCL X 03 E3 04 / 05 + 06 "Y" - matrix will use the Y-Registers, starting at Y_001 07 MATDIM 08 1 09 MCON - in the SandMatrix 10 CLX 11 MSIJA 12 2 13 LBL 00 14 1 15 + 16 MSC+ 17 SF 25 18 J+ 19 FS?C 25 20 GTO 00 21 MDET 22 END And these are the results: D(11) = 1,486,442,880.0 - in 1.8 seconds D(13) = 2.834656472 E11 - in 2.01 seconds D(30) = 3.311538747 E34 - in about 11 seconds Not bad again, in fact surprisingly good. Using Gerson's values as the reference, these results are practically as accurate as those obtained using the equivalent formula D(n) = (n+1)!*H(n+1). - which can be easily programmed using the Harmonic function in the SandMath (step #4 in the trivial program below): 01 LBL "DN" - n is expected in X 02 1 03 + 04 HARM 05 LASTX 06 FACT 07 * 08 END which returns respectively (in just a few milliseconds this time): D(11) = 1,486,442,880.0 D(13) = 2.834656474 E11 D(30) = 3.311538746 E34 Thanks for the case study, it's helped with the beta testing and adds another interesting problem to the list. Saludos, Á. "To live or die by your own sword one must first learn to wield it aptly." |
|||
« Next Oldest | Next Newest »
|
User(s) browsing this thread: 1 Guest(s)