Re: Coefficient

*To*: mathgroup at yoda.ncsa.uiuc.edu*Subject*: Re: Coefficient*From*: CAMERON at midd.cc.middlebury.edu*Date*: Tue, 11 Dec 90 11:21:35 -0600

I mailed this response privately to the original poster, Keith Slavin (keith at videovax.tv.tek.com), but since then I have seen so many other comments posted to MathGroup that I thought perhaps I should broadcast this as well as sending it privately, because I haven't seen anyone else mention this approach. The original question: > I seem to be having a problem with the behaviour of Coefficient[] in > Mathematica, namely if I have a series such as > > sum = ... + a n + b + c/n +.... > > then Coefficient[sum,n] gives a, Coefficient[sum,1/n] gives c, but > Coefficient[sum,1] does NOT give b, but zero instead! This is causing me > problems, and I cannot see a simple work-around. Well, it seems to me that Coefficient[sum,1] can't know to give you "b" instead of "0" in "sum", because how does it know that you are interested in powers of "n" rather than (say) powers of "b"? It's already been pointed out that Coefficient[sum,n,0] doesn't work right for the given "sum" because of the negative powers of n. (Personal opinion: I don't regard as satisfactory any solution that has this constraint.) Mathematica's "FreeQ" is useful here... try "Select[sum,FreeQ[#,n]&]". That gives you just the terms of "sum" that don't have an "n" in them, which presumably is the "coefficient of 1" in a series in the variable "n"? It works if "sum" is actually an expression of the form "Plus[ ... , .... ]". Of course you referred to the value of "sum" as a "series", and the "FreeQ" approach will *not* work if the value of "sum" actually is a "SeriesData" object (as produced by Mma's "Series" operator). In that case, you could use "Normal" to convert the "SeriesData" object to a "Plus" object. Hope this helps-- --Cameron Smith Mathematica consultant CAMERON at MIDD.BITNET --or-- cameron at midd.cc.middlebury.edu

**Coefficient[]**

**Power[] corrupts, Absolute[Power[]] corrupts absolutely**

**Coefficient**

**D and sinh'; s/c vs. s*c^(-1).**