Help to build a function to approximate a expression when X >> Y

After expand I always get my original simplified form.

In general I can not assume a*x >> b*y. However, if x >> y then b*x >> 
b*y and I can drop this term.

When we calculate transfer functions in circuit analysis we get things 
like this all the time. With Maxima I could write a short program that 
(using standard two port matrix algebra and by introducing the ABCD 
matrix (and conversions to other parametric descriptions) of some common 
structures like bipolar transistors, MOSFETs and impedances) can 
calculate and simplify the expressions for diverse gains, impedances and 
the like. Sometimes the expressions involve a long sum of products of 
different circuit parameters. Once the sum of products is generated the 
simplification method is to assume that one of the circuit parameters is 
much bigger than another. Then look for the common factors multiplying 
both and approximate as in the simple example. If we proceed in this way 
with several variables (by pairs) we get good approximations with way 
shorter expressions.

The expressions that I try to approximate are sums of products in 
different variables. Additon and subtraction of terms will be 
resimplified to the original expression after expanding.

Thanks

Javi
Stavros Macrakis wrote:

>     (a + b + c + d)*x + (d + e )*y +f
>     try to write this in the form
>     (a + b + c)*x + e*y + f  + d *(x+y)
>     and then subst x+y by x to get
>     (a + b + c+ d)*x + e*y + f
>
>
> I don't understand why this is a valid approximation.  You could 
> equally rewrite the original expression as (a+b)*x + (e-c)*y + 
> (d+c)*(x+y) + f => (a+b+c+d)*x + (e-c)*y + f.
>
>         -s