I have two functions which are the linear and non-linear solution of an equation. I'm trying to calculate where does the linear solution is a good solution of my equation, when compared with the non-linear. To do so I am using $$\int_{\tfrac{a}{2}}^\infty\left(\sigma^{NL}(x)-\sigma^{L}(x)\right)^2\mathrm dx$$ The variables I'm using are T = 298.15(*K*); k1 = 1.38064852*10^-23 (*J/K*); ϵ0 = 8.85418781761*10^-12 (*C^2/N m^2||F/m||C/V m*); ϵ = 78.5; ϵR = ϵ ϵ0 (*C^2/N m^2||F/m||C/Vm*); c = 2.99792458*10^8(*m/s*); z = 1; e1 = 1.60217733*10^-19 (*C*); NA = 6.022140857*10^23 (*mol^-1*); a = 4.25(*Å*); a1 = a*10^-10 (*m*); x1 = x*10^-10 (*m*); And the main code is nMKS[ρ0_] = ρ0 *10^3 NA (*m^-3*); κMKS[ρ0_] = Sqrt[(2 nMKS[ρ0] (e1^2) (z^2) )/(ϵR k1 T)](*m^-1*); LσMMKS[ρ0_, ψH_] = ϵR κMKS[ρ0] ψH/1000 (*C/m^2*); LψPBMKS[ρ0_, ψH_, x_] = ψH Exp[κMKS[ρ0] (a1/2 - x1)] (*mV*); LσMKS[ρ0_, ψH_, x_] = Piecewise[{ {0, 0 <= x1 < a1/2}, {ϵR κMKS[ρ0] LψPBMKS[ρ0, ψH, x]/1000, x1 >= a1/2} }](*C/m^2*); NLσMMKS[ρ0_, ψH_] = (2 ϵR κMKS[ρ0] k1 T)/( z e1) Sinh[(z e1 )/(2 k1 T) ψH/1000] (*C/m^2*); ZMKS[ψH_] = (z e1 ψH/1000)/(2 k1 T); NLσMKS[ρ0_, ψH_, x_] = Piecewise[{ {0, 0 <= x1 < a1/2}, {(4 ϵR κMKS[ρ0] k1 T)/(z e1) (((Exp[2 ZMKS[ψH]] - 1) Exp[κMKS[ρ0] (a1/2 - x1)])/( (Exp[ZMKS[ψH]] + 1)^2 - ((Exp[ZMKS[ψH]] - 1) Exp[κMKS[ρ0] (a1/2 - x1)])^2)), x1 >= a1/2} }](*C/m^2*); Hence, when I try to integrate my equation with Dif2σ[ρ0_, ψH_, x_] = FullSimplify[(NLσMKS[ρ0, ψH, x] - LσMKS[ρ0, ψH, x])^2] Vσ[ρ0_, ψH_] = FullSimplify[Sqrt[Integrate[Dif2σ[ρ0, ψH, x], {x, a/2, Infinity}]]] Mathematica is not able to solve it. How can I force Mathematica to actually solve this and thereafter define a function from its solution?