[41, Example 3.15] After incorporating the surgery of the brane structures, Joyce’s LMCF is no longer identical to the LMCF of the underlying Lagrangian. The most extreme case is to start with the union $L$ of two unobstructed special Lagrangians $L_{1},L_{2}$ with phase angles $\theta_{L_{1}}<\theta_{L_{2}}$, with an intersection point $b\in CF^{1}(L_{2},L_{1})$ defining a closed morphism. We regard $L$ as an immersed Lagrangian with bounding cochain $b$. This fits into the distinguished triangle

$$L_{1}\to L\to L_{2}\to L_{1}[1],$$

which is not destabilizing for $L$. This configuration is stationary in ordinary LMCF. However, under Joyce’s LMCF, the bounding cochain $b$ evolves in time, and loses positivity in the Novikov ring in finite time, after which one is supposed to ‘open up the neck’ to continue the flow in a nontrivial fashion.