Knowing that the problem is due to the limit function, I am using physical criteria to prescribe "by hand" the boundary values of the expressions m1, q1, k1 at r=R[1], without having to compute limits. The procedure now runs very fast. Hopefully, the explanations provided will help me to make better modifications to the code that dodge the direct use of 'limit'. The procedure's code is now modified in the following lines: Lm0c:=(eval@subs)(r=R[1], m0): Lm1c:=Lm0c: Lq0c:=(eval@subs)(r=R[1], q0): Lq1c:=Lq0c: Lk1c:=(eval@subs)(r=R[1], k0): LH1c:=sqrt(Lm1c+Lq1c-Lk1c): LS1c:=0: Lm0:=[Lm0c, seq( (eval@subs)(r=R[i],m0), i=2..n)]: Lm1:=[Lm1c, seq( (eval@subs)(r=R[i],m1), i=2..n)]: Lq0:=[Lq0c, seq( (eval@subs)(r=R[i],q0), i=2..n)]: Lq1:=[Lq1c, seq( (eval@subs)(r=R[i],q1), i=2..n)]: Lk1:=[Lk1c, seq( (eval@subs)(r=R[i],k1), i=2..n)]: LDm1:=[0.0, seq( (eval@subs)(r=R[i],Dm1), i=2..n)]: LDq1:=[0.0, seq( (eval@subs)(r=R[i],Dq1), i=2..n)]: LDk1:=[0.0, seq( (eval@subs)(r=R[i],Dk1), i=2..n)]: LH1:=[LH1c, seq( (eval@subs)(r=R[i],H1), i=2..n)]:

Scott Tanks for trying the code. To answer your question, I didn't increase the memory allocation on Maple 10. I ran the code in Maple 10 runing on an 867 mz 4 year old G4 Powerbook (12'' screen). I didn't change the original Maple 10 settings in that mac. It took a long time but was able to complete the run without the kernel error message. I was even able to plot the numeric functions. If this is not firewall related, then does it mean that it is a bug?