I obtained a numerical solution from the following code with NDSolve

L = 20;
tmax = 27;
\[Sigma] = 2;
myfun = First[h /. NDSolve[{D[h[x, y, t], t] +
Div[h[x, y, t]^3*Grad[Laplacian[h[x, y, t], {x, y}], {x, y}], {x, y}] + 
Div[h[x, y, t]^3*Grad[h[x, y, t], {x, y}], {x, y}] == 0, 
h[x, y, 0] == 1 +
1/(2*\[Pi]*\[Sigma]^2)*Exp[-((x - 10)^2/(2*\[Sigma]^2) + (y - 10)^2/(2*\[Sigma]^2))],
h[0, y, t] == h[L, y, t], h[x, 0, t] == h[x, L, t]}, 
h, {x, 0, L}, {y, 0, L}, {t, 0, tmax}, 
Method -> {"MethodOfLines", 
"SpatialDiscretization" -> {"TensorProductGrid", "MinPoints" -> 60, "MaxPoints" -> 60, 
"DifferenceOrder" -> 4}}, StepMonitor :> Print[t]]]

(It took about 7 sec to be solved on my old laplop.)

Next, I am trying to make an animation and export a .gif file to present its evolution as follows: (taking about 50 sec)

mpl = Table[Plot3D[myfun[x, y, t], {x, 0, L}, {y, 0, L}, PlotRange -> All, 
PlotPoints -> 40, ImageSize -> 400, 
PlotLabel -> Style["t = " <> ToString[t], Bold, 18]], {t, 0, 27, 1}];
Export["test.gif", mpl, "DisplayDurations" -> 1, "AnimationRepetitions" -> Infinity]

enter image description here

Here are my questions:

As you may see, during the evolution (1) the box(frame) of the animation is shrinking and expanding, though slightly, (2) the augment in the amplitude is shown through increasing the vertical coordinate. If one neglects the scaling in this coordinate and only observe the middle peak, he may do not feel its growth. This is a problem in make a presentation.

I don't know the reason for the fist observation, for the second one, while, I think MMA try to highlight the surface variation at every instant by scaling the vertical axis synchronously.

Can anyone please help me to suppress the oscillation of the 3Dbox and hold the coordinate of vertical axis as the final frame at $t_\text{max}=27$ (i.e. about z=6 here) because I want to show the surface evolution form a small fluctuation to the final big amplitude. Thanks!

  • $\begingroup$ Out of curiosity, what is the physics you are capturing with your PDE? $\endgroup$
    – dearN
    Nov 7, 2016 at 0:42
  • $\begingroup$ Hi, @drN, I remember that it was just a simplified version of a film equation from your post and others' papers :) $\endgroup$
    – lxy
    Nov 8, 2016 at 5:01
  • $\begingroup$ nice! And you are trying to create a pendant drop due to gravity and surface tension? You should check out our work from the 2010 Wolfram tech conference. Search for Allan Struthers. He had similar code that creates pretty much the same kind of spatiotemporal pattern as you. $\endgroup$
    – dearN
    Nov 8, 2016 at 11:34
  • $\begingroup$ @drN. Hi, drN, yep, really appreciate your information! I am still new to MMA though have used it for one year. I am not used to the the Wolfram language logic. I think I can learn something from you and your post. $\endgroup$
    – lxy
    Nov 9, 2016 at 9:53

2 Answers 2


To prevent shaking, try to add ImagePadding and for the other issue, you can fix the vertical plot range.

mpl = Table[
   Plot3D[myfun[x, y, t], {x, 0, L}, {y, 0, L}, 
    PlotRange -> {Automatic, Automatic, {0, 6}}, PlotPoints -> 40, 
    ImageSize -> 400, 
    PlotLabel -> Style["t = " <> ToString[t], Bold, 18], 
    ImagePadding -> 30], {t, 0, 27, 1}];

enter image description here


I only post this as a way (without distortion or dealing with multiple scales)to illustrate the initial Gaussian (flat relative to final range) with MeshFunctions and using ColorFunction. I have voted for Nasser's answer.

fun[t_] := Legended[Show[
   Plot3D[myfun[x, y, t], {x, 0, 20}, {y, 0, 20}, 
    MeshFunctions -> (#3 &), Mesh -> {{0, 1, 1.1, 1.2, 2, 3, 4, 5}}, 
    MeshStyle -> Thick, 
    ColorFunction -> Function[{x, y, z}, ColorData["Rainbow"][z/6]],
    ColorFunctionScaling -> False,
    PlotRange -> {0, 6},
    PlotPoints -> 40,
    PerformanceGoal -> "Quality",
    PlotLabel -> Style[Row[{"t= ", t}], 20, White, Bold]], 
   Background -> Black], BarLegend[{"Rainbow", {0, 6}}]]

The gif was exported from fun/@Range[0,27,0.5] at 8 frames per second (using "DisplayDurations".

enter image description here

  • $\begingroup$ your method is quite good for a presentation. Thank! If it can accept two answers, I will definitely choose your as well. $\endgroup$
    – lxy
    Nov 6, 2016 at 9:46
  • 1
    $\begingroup$ @jsxs Nasser provides the direct answer. I would not change your accept. I voted for Nasser as well. I just aimed to illustrate one way to show you start with a flat Gaussian. There are other ways and play is one of the best teachers (for me at least). :) $\endgroup$
    – ubpdqn
    Nov 6, 2016 at 9:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.