Jump condition for elasticity on an interface

[fb77 (François Bouchut)]

Dear Loïc,
Instead of using intalledges you can use int1d(Th,5), and jump will be available.
For the second term, you have to take the integral “from the outside”.
For that you need to have a clockwise orientation of \Gamma.
About the int1d on an internal boundary of a discontinuous function (and the related orientation issue), see

Normal stress in a horizontal line inside of a mesh from top side only

Hello, Your approach is correct. Using buildmesh you can put a border that is inside the domain. Then it has a label (1 in your case) that can be used to compute int1d(Th,1)(). This “internal border” is considered as a boundary. In particular if you write int1d(Th) without mentioning a label, it will integrate on all boundaries, including the “internal boundary”. When you have a finite element function u that is discontinuous through the internal boundary (for example if u is P0 on Th), your…

But take care that in order to keep the inside region in you mesh (and not exclude it), for a clockwise orientation of \Gamma you will need to apply buildmesh with a negative number on border 5, like
mesh Th=buildmesh(b1(10)+b2(10)+b3(10)+b4(10)+b5(-20));
François.