ВЫПУКЛОСТЬ_EXAM.mw     
   Another easier way. Intersects  the surface by the plane. Is obtained red curve. We are building a plane perpendicular to the red curve at every point of red curve. These planes  intersects  the surface. Obtained blue curves. At the points of blue curves we construct equidistant surface (previous examples).
   If we consider that the solution is very accurate, and the number of points is not particularly limited, such an approach is applicable for practical purposes.

I thank all who responded and supported.
I got the impression that this is the best forum that I know.

     The situation is particular: how to "flagging the post as 'other' and providing the details", if the moderator deleted this same post?

@Markiyan Hirnyk   Print  Screen  

@Markiyan Hirnyk  I remind you, any communication with you from my side impossible.

Print  Screen  

Community help me! This man is quite mad.
He writes, "0.1e-1 * exp * x1 / (0.94e-2 + x1 ^ 4 + x2 ^ 4)" instead of my
"X3 = 0.01exp (x1) / (0.0094 + x1 ^ 4 + x2 ^ 4)." Then he says that it is wrong and makes me change something, otherwise he threatened with the removal of the message. What is it ?
http://www.mapleprimes.com/posts/203796-Equidistant-Surface-?submit=comment 

@Markiyan Hirnyk
x3 = 0.01exp (x1) / (0.0094 + x1 ^ 4 + x2 ^ 4)

To some extent I have recovered one of the deleted messages 

http://www.mapleprimes.com/posts/203796-Equidistant-Surface-#comment201542 

An attempt to restore the removed  message  (not by me). The message was from on 17.05 / 2016.

Example with explanations.

In the drawing of we have a set of curves, integrated into the surface of indigo. Curves lie at a distance 0.4 from the surface

x3 = 0.01exp (x1) / (0.0094 + x1^4 + x2 ^ 4 )                                        (1)

We move the curve along the surface (1). This curve in the initial position lies at the intersection x1 = 1 and (1). At the points of this curve we put the distance of 0.4 along the normal to the surface (1) and gradually get the surface of indigo.

And yet another embodiment. The text of the programs the same.

Алсу, очень хорошие работы.

he had just removed my comment in the topic http://www.mapleprimes.com/posts/204275-Presentation

@Preben Alsholm I'm clear, you know that there is a shooting method.

I would like all the same to clarify for himself: is unclear the approach to the solution, or solution can not be done in Maple?

@Preben Alsholm  Yes, everything is correct. (I think we can minimize y (1) -Y1 without squares.)
      Regarding the shooting method, the boundary conditions are different, for example, y '(1) = y (1), and then there will be two unknown variables. More, boundary conditions can have a nonlinear relationship. That is, the shooting method is suitable for simple conditions.
      Yes, with optimization, but how to use, for example, numerical dsolve to obtain boundary discrepancies? It is important that is possible to manage numerical differentiation.