# Errors in evalauting an expression

I have the following:

Constants:

b3b = 2;
G = (1/eps^2);
a1 = 3;
a3 = 4;
xc = 0.3;


Functions:

y[x_] := (x*a1 - x^3*a3)/G
n[x_] := (y[x]^2)


I want g[x], gp[x], gpp[x] as functions of eps only and all of these three functions should be evaluated at x=xc :

g[x_] := (G*(-y[x]/eps)*Exp[-n[x]] - (b3b/eps)*
NIntegrate[(y[s])^4*Exp[-n[s]], {s, 0, xc}])


Please note that in the integrand above --> both n[x] and y[x] are functions of eps and so I can use NIntegrate since I am using := for g[x] even though eps is not given a numerical value as yet. Everything is perfect till this point as I am able to get a numerical value for g[xc] for a eps value. The problem happens when I take the following derivatives:

 gp[x_] := D[g[x], x]
gpp[x_] := D[gp[x], x]


since when I evaluate gp[xc] and gpp[xc] for a particular eps value, I get errors: NIntegrate::inumr and General::ivar.

Therefore, I need a way to get around this. I must be doing something wrong. Any help would be greatly appreciated. Thanks!

• @Nasser If I use the := for g[x] then isn't it fine to use NIntegrate? because at the end I have to plot a huge function that requires g[xc], gp[xc], gpp[xc] to be changing with eps..i tried the one you had up and got General::ivar can you please tell me what I can do? – Stoc Aug 24 '13 at 23:53
• @Nasser thanks for the help...apologies regarding that...can you please tell me what else I can do for my situation...can i use Integrate for g[x] instead and then take the derivatives? however, i still get an error when I evaluate the functions at x=xc for a particular eps...please help me as to what I can do – Stoc Aug 25 '13 at 0:02
• @Artes I reformulated the question as you suggested...can you please take a look now...i hope it makes more sense as to what I want to do now – Stoc Aug 25 '13 at 0:18

The problem here is the way you defined the derivatives. Mma will evaluate e.g. in this order: gp[xc] -> D[g[xc], xc] and obviously you'll get an error there. You may define the derivatives like this:

gp[x_] := D[g[t], t] /. t -> x
gpp[x_] := D[g[t], {t, 2}] /. t -> x


And you'll get it working:

Block[{eps = 0.1}, {gp[xc], gpp[xc]}]

{-19.1964, 72.004}

• you are amazing...thanks so much for this..been dying for this! – Stoc Aug 25 '13 at 3:37