Date: Fri, 13 Oct 2006 14:04:25 -0400 (EDT) From: Jin ChenTo: Guo-Yong Fu Cc: jardin@pppl.gov, jbreslau@pppl.gov, strauss@cims.nyu.edu, sugiyama@psfc.mit.edu, wpark@pppl.gov Subject: RE: comparison between version 3.9 and 3.10 Hi Guoyong, As we talked about yesterday to check the implementation of poisNeuman, I made changes according to 3.9, 3.10, and m3d_9-30-06.tgz (Linda's) as well. The new version was given to Josh yesterday and he told me it works and produces the same result (good) as before. So I personally think this part of work is done. For Linda... Hi Linda, I remember you talked about this before: m3dp, m3dp_fsymm, m3dp_fsymm_opt produced different results. I checked the implementation of corresponding slovers (poissc.c, poissc_fsymm.c, and poissc_fsymm_opt.c) and I found their implementations are different. That's how the differences came from. Then I changed poissc_fsymm.c and poissc_fsymm_opt.c according to the correct poissc.c (dated by yesterday). Now they all produce the very simular results using gmres/ilu as the common solvers. Please check the following numbers: -m3dp.x- ncy,tim,dt= 10 650.19999999995639 2.00000000000000004E-2 g 5.99306404943192143E-3 2.8574373322053303E-11 7.89595002862425552E-6 -4.46636594627447227E-8 -2.27629585422369407E-5 1.233974894998539 p,pe 3.78154268835922326E-5 0.E+0 F,lapF,si= -1.85908127074577304E-9 1.29916757406525922E-7 -2.81830986615972745E-3 chib,lap,y= -4.46636594627447227E-8 -2.40152532608343229E-6 1.37546264673087993E-7 -m3dp_fsymm.x- ncy,tim,dt= 10 650.19999999995639 2.00000000000000004E-2 g 5.9842524587005539E-3 2.85747382626055003E-11 7.89246092049707699E-6 -4.45655512487764644E-8 -2.27631189746820371E-5 1.2339748960709114 p,pe 3.7815426933910069E-5 0.E+0 F,lapF,si= -1.85448725278635027E-9 1.29918965659877251E-7 -2.81830999643064434E-3 chib,lap,y= -4.45655512487764644E-8 -2.40183059130962075E-6 1.37449354406897438E-7 -m3dp_fsymm_opt.x- ncy,tim,dt= 10 650.19999999995639 2.00000000000000004E-2 g 5.98425241322560371E-3 2.8574738262361915E-11 7.89246091983336569E-6 -4.45655512452066441E-8 -2.27631189747748245E-5 1.2339748960709116 p,pe 3.78154269339100622E-5 0.E+0 F,lapF,si= -1.85448725498485737E-9 1.29918965831581788E-7 -2.81830999643132608E-3 chib,lap,y= -4.45655512452066441E-8 -2.40183059193553998E-6 1.3744935633635993E-7 The only differences are g 5.99306404943192143E-3 for m3dp.x g 5.9842524587005539E-3 for m3dp_fsymm.x g 5.98425241322560371E-3 for m3dp_fsymm_opt.x Please let me know if you want the code after you come back.