Advanced Search

Question: ... use codegen without redundency in the code output...

Question:... use codegen without redundency in the code output...

sarcasmo 0 Maple
array
August 25 2006
0
We use maple to generate all of our computational bits, dump this out to header files, and include this in our code base. A lot of the code that is generated is redundant so it would be nice to have codegen[C] know, or at least be able to be told, to optimize the entire code being dumped to the same header file. Is this possible? Examples follow... " ..... DCb := array(1..3): lp:='lp':ip:='ip':m:='m':ap:='ap':jp:='jp':a:='a':l:='l':j:='j':i:='i':k:='k':kp:='kp': for ip from 1 to 3 do DCb[ip] := sum(sum(sum(sum(sum(gd[ip,jp]*gu[ap,lp]*jacu[m,lp]*jacu[a,ap]*dJ[jp,m,a],m=1..3),jp=1..3),lp=1..3),ap=1..3),a=1..3): od: DCx := array(1..3): for i from 1 to 3 do DCx[i] := DCA[i] - DCa[i]: od: DCr := array(1..3): for i from 1 to 3 do DCr[i] := DCB[i] - DCb[i]: od: codegen[C]([dest_comp[0] = DCx[1]],optimized,precision=double,filename=`transform_.h`): codegen[C]([dest_comp[1] = DCx[2]],optimized,precision=double,filename=`transform_.h`): codegen[C]([dest_comp[2] = DCx[3]],optimized,precision=double,filename=`transform_.h`): ..... " and when the code comes out, I have something like: " t1 = metric_fields[GXX]; t2 = metric_fields[GZZ]; t4 = metric_fields[GXZ]; t5 = t4*t4; t6 = t1*t2-t5; t7 = t1*t6; t8 = metric_fields[GYY]; t9 = t1*t8; t11 = metric_fields[GYZ]; t12 = t11*t11; t14 = metric_fields[GXY]; t15 = t14*t14; t17 = t14*t4; t22 = 1/(t9*t2-t1*t12-t15*t2+2.0*t17*t11-t5*t8); .... t303 = t146*t31; t331 = -t291*t115*t28*t241+2.0*t300*t301*t40*t303+2.0*t300*t301*t37*t303- t129*t115*t28*t169-t260*t186*t71-t260*t186*t100-t265*t221*t26*t90-t270*t186* t157-t270*t186*t170-t106*t35*t37*t108-t275*t221*t26*t180; dest_comp[0] = -t24*t27*t71+t7*t22*t75*t34*t90-t24*t27*t100-t106*t35*t40* t108+t114*t115*t39*t70-t121*t62*t122-t114*t115*t28*t99+t129*t115*t39*t137+2.0* t142*t143*t40*t147+orig_comp[2]*spher_a*t27+t245+t295+t331; " Now when it goes to generate dest_comp[1], the code is ~70% the same. How do I avoid this and just have the entire code bit optimized? Thanks, jon
Please Wait...