I have a huge series of a nonlinear function.
3240910002558383085492736408270606958592 sin(x3_) 23950603179913084375 k10 x6_ 23950603179913084375 k9 x5_
------------------------------------------------- - ---------------------------- - ---------------------------
#4 25973015655783048675328 25973015655783048675328
2
816555808152780085063004140309315584 cos(x3_) sin(x3_) 64261963098890133504 k5 x1_ 64261963098890133504 k6 x2_
+ ------------------------------------------------------- + --------------------------- + ---------------------------
#2 #5 #5
2
64261963098890133504 k7 x3_ 64261963098890133504 k8 x4_ 2414846496921978125 k9 x5_ cos(x3_)
+ --------------------------- + --------------------------- + ------------------------------------
#5 #5 10625324586456701730816
I am succeeded in simulation, the result of this simulation is around of 0.01 up to 2. But its elements and computation are so complicated. I would like to simplify it, then use that result to run in a C program. I also try to use vpa, but that is a bad result due to the difference between the huge number and the small value so it makes a wrong approximation on the result.
vpa = (2.6737e-21*(4.4485e90*sin(x3_) - 9.4197e86*x4_^2*sin(x3_) + 1.19e90*cos(x3_)*sin(x3_) - 5.9344e88*k1*x1_ ...
In the other hand, my idea is can I do some function to get the factor like below:
Assume f = a1*sin(x)*y*y + a2*x*x/(cos(x)^3).
Any function in Matlab that I can achieve a1, a2?
Moreover, in C language, the maximum number of int32 is limited, so I am looking for a solution for this problem. Please help to give me some suggestion. Thank in advance!