The first layer of the dome.

 The second layer of the dome.

Facing the biggest problem which is not the dome but how to bending the geometry following the dome’ s curvature.

In order to achieve this progress, I chose to use box morph node.

 As we know there are several functions to create a dome, including using kangaroo or using loft. Since I want an easier way in order to test my solution without file crashes, I decided to cut the dome from a mesh sphere. However, the final creation of kangaroo physics is a mesh and the final creation of kangaroo physics is an opening brep. Both of them fail to input into surface box node which can create a twisted box along the curved surface. So I find a way to convert my mesh sphere into a surface.
To begin with, I deconstructed the mesh in order to find all the vertices and faces. Those faces are some small planes which can be converted into surfaces. Then I deconstructed these faces again and getting the points of them. In this way, I can use four points surface to recreate the surface form the points of faces. 

However, there is one thing should be noticed. After deconstructing the faces, what I got are not the exact points from those faces, but the list of the first, second, third and fourth points of all faces. So without using list item to pick out the points I want, I failed to connect the generation of deconstructing faces with the 4 points surface node.

 Then I divided the domain of the surface to get each domain of a fragment of the surface. Since the surface which I created from mesh sphere is a gathering of small fragments. So I only input 1 into U and V coefficient since I did not want to cut the surfaces into smaller ones.

 Secondly, I created the small pattern in rhino and using bending box node creating the unit box if the pattern.

 After inputting all elements into box morph, I got the twisted geometry which is sphere covered by the pattern.

I can simply bake the geometry and cut it in rhino to get the dome shape.

Then I also tried to use kangaroo physics to create the dome.

 It also worked.

 However the only problem is after cutting the final mesh into a dome, I got a waving edge instead of a round edge.

Thirdly, I use voronoi 3D to create the interior layer.

 Making difference between cell geometries and the sphere to get the small openings.

For the exterior layer, the denser the pattern results in the smaller the opening area. For the interior layer, the more cells result in a smaller area for each cell. So I decided measuring maximum daylight by calculating the maximum overlap of exterior patterns and interior openings.