| Author |
Comment/Response |
Joel D
|
 10/09/12 6:11pm
In Response To 'Re: Poisson Solver'
---------
Bill, thanks for your response, your reply makes total sense as far as the machine resolution. The implementation part is what I am struggling wiht now. I'm not able to produce the same results as you mentioned. The Mathematica help on the machine precision function is not extremely intuitive on where all you need to apply it.
I went into my program and changed the convergence threshold to read 1.0`35*^-34 and I still kick out at 633 iterations. I need to be able to slowly rachet down the convergence thresholds until my Vnew data is no longer changing. I've commented out the Tracker Point part since we are getting upwards of 1000 iterations now.
When you say you changed all my coefficients to have 32 digits of precision, is there a quick way to go about doing that? And would I need to have all coefficients digits or precision, match the convergence test threshold digits of precision in order to step through incremental thresholds? Ex. I need to test at 10^-10, 10^-25, 10^-50, 10^-75, 10^-100 etc. etc. until the Vnew values stop changing.
Is there anyway you could paste your example at: http://pastebin.com/u/JoelD
Thanks again for your help.
Joel
URL: , |
|
