# Problem in solving non-linear differential equation having Gaussian function

   we = 30*10^-6; P0 = 3; \[Gamma] = 0.070; g = 10; \[Rho] = 1000; c =
3*10^8;
\[Kappa][r_] := (z''[r] + z'[r]/r); \[CapitalPi][r_] := (4 P0)/(
Pi*we^2*c) Exp[-((2*r^2)/we^2)];
Solsol = NDSolve[{-\[Gamma]*\[Kappa][r] + \[Rho]*g*z[r] - \[CapitalPi][
r] == 0, z[50*we] == 0, z'[0] == 0}, z[r], {r, 0, 10 we}]
g = Plot[Evaluate[{z[r], z'[r], z''[r]} /. sol], {r, 0, 10 we},
PlotStyle -> Automatic]

• You have a differential equation, therefore you need NDSolve instead of NSolve. Further, Infinity is not a number and can not be used as function argument. Dec 25, 2022 at 9:47

You have singularity at r=0 due to (z''[r] + z'[r]/r) so can't start at zero. Try $MachineEpsilon ClearAll["Global*"] we = 30*10^-6; P0 = 3; γ = 0.070; g = 10; ρ = 1000; c = 3*10^8; ode0 = (z''[r] + z'[r]/r) h = (4 P0)/(Pi*we^2*c) Exp[-((2*r^2)/we^2)]; ode = -γ*ode0 + ρ*g*z[r] - h == 0 ic = {z[50*we] == 0, z'[$$MachineEpsilon] == 0}; sol = NDSolve[{ode, ic}, z[r], {r,$$MachineEpsilon, 10 we}]  g = Plot[Evaluate[{z[r], z'[r], z''[r]} /. sol], {r,$MachineEpsilon, 10 we},
PlotStyle -> Automatic]


Update

To plot the derivatives, you can do this. You can actually have NDSolve solve for these.

sol = NDSolve[{ode, ic}, {z, z', z''}, {r, $MachineEpsilon, 10 we}]  Then do g = Plot[ Evaluate[{z[r], z'[r], z''[r]} /. sol], {r,$MachineEpsilon, 10 we},
PlotStyle -> Automatic, PlotRange -> All]


Now 3 curves show up. But the scales are so different, hard to see them all on same plot. You can plot each on separate plots? Something like

opt = {PlotRange -> All, ImageSize -> 300, GridLines -> Automatic,
GridLinesStyle -> LightGray, PlotStyle -> Red};
p1 = Plot[Evaluate[z[r] /. sol], {r, $$MachineEpsilon, 10 we}, Evaluate@opt, PlotLabel -> "z(r)"]; p2 = Plot[Evaluate[z'[r] /. sol], {r, MachineEpsilon, 10 we}, Evaluate@opt, PlotLabel -> "z'(r)"]; p3 = Plot[Evaluate[z''[r] /. sol], {r,$$MachineEpsilon, 10 we},
Evaluate@opt, PlotLabel -> "z''(r)"];
Column[{p1, p2, p3}]


I am trying to vary [Gamma] using sol = ParametricNDSolve[{ode, ic}, z, {r, $MachineEpsilon, 200 we}, {[Gamma]}] then I want to calculate A0 = -[Gammaa^2( (z''[[Gamma]][r] - z'[[Gamma]][r]/r for two differnt value of [Gamma] ) 0.030 and 0.070. You can use ParametricNDSolve ClearAll["Global*"] we = 30*10^-6; P0 = 3; g = 10; ρ = 1000; c = 3*10^8; ode0 = (z''[r] + z'[r]/r) h = (4 P0)/(Pi*we^2*c) Exp[-((2*r^2)/we^2)]; ode = -γ*ode0 + ρ*g*z[r] - h == 0 ic = {z[50*we] == 0, z'[$$MachineEpsilon] == 0}; sol=ParametricNDSolve[{ode,ic},z,{r,$$MachineEpsilon,10 we},γ] z1 = z[0.030] /. sol; z2 = z[0.070] /. sol; Plot[{z1[r], z2[r]}, {r,$MachineEpsilon, 10 we},
PlotLegends -> {"0.03", "0.07"}]


γ = 0.030;
z1 = z[γ] /. sol
a = 1;
A0 = -γ*a^2*(z1''[r] - z1'[r]/r)

γ = 0.070;
z2 = z[γ] /. sol
A0 = -γ*a^2*(z2''[r] - z2'[r]/r)

• Thanks, but it it showing only z(r) its derivatives z'(r) z''(r) not showing. Dec 25, 2022 at 11:22
• thanks! derivative z''(r) near 0 is not so clear, Please also try to plot (z''(r)+z'(r)/r)/2 which is the mean curvature. Dec 25, 2022 at 12:13
• Please suggest how to use ParametricNDSolve so that I can vary "we" for two different values. Jan 12, 2023 at 11:27
• @GopalVerma I do not think you can use the location of initial conditions as a parameter. But why not simply make a function and pass it we value, and solve the same equation each time you want to change we by making a call to this function with different we? This is much simpler solution and you do not have to deal with complexity of ParametricNDSolve. Jan 12, 2023 at 23:00
• Thanks, I got it. Jan 13, 2023 at 6:41