|
Phys(ics)Geek
|
|
"I'm a physicist in this entry, you are forewarned..."
w00t! Thus far a really good day... didn't feel the need to actually go in to work... but instead telecommuted thanks to the fact that I work on a $90,000 supercomputer and all I had to do was SSH to go to work... Secondly, a phone call resolved the problem of my getting paid on time... so I can relax and stop worrying that I won't get my check for $2000 on payday... Thirdly, I sat down at 7:00 this evening to really focus my powers on solving this long standing problem I've been having with the code, and by 7:45 I had fixed the problem, compiled the code, and saw that my new fangled "start at zero" system works!!! Booyah, bitches... And to let me show you just how kickass my numbers came out... here's a sample: Old method: Bxz = 0.77930383996832652826 + 1.53905979731185937887i Best data before fix in the new method: Bxz = -23241.15381427558412647552 + 0.00000000000000000021i; New value after fix: Bxz = 0.77835741801148272678 + 1.53872164802831412622i As one can see... the new numbers almost exactly match the "correct" value that we did have... I put correct in quotations because, honestly, the new third value is actually more theoretically correct, because instead of approximating the starting location (in the complex plane), we actually start the computation at exactly 0+0i. I did some analysis of accuracy, and according to my calculations (using a predeifined state and known experimental data), there was about a 0.00259% error in the first calculation, and I improved that to about 0.00017%. Pretty 'dern good, eh? Feeling:
|