Difference in LVD between 1.6.7 PR4 and PR5

[abelsm2]

I've noticed a large difference in LVD between PR4 and PR5 where PR5 appears to be broken. When opening the same file, PR5 starts from a different point and is unable to arrive at the same solution after optimization. I used LVD to create a mission with a gravity assist at Eve then 2 orbits before a gravity assist at Kerbin and then finally capture at Jool. The changes you made in 1.6.7 PR4 made this almost simple! The continuity constraints were violated but only by very small amounts. However, when I opened the same file in PR5, the starting position for the trajectory is very different, as seen in the images below, with PR4 on the left and PR5 on the right. The continuity constraints are now violated significantly.

The starting point for the constraint violations in PR4:

The starting point for the constraint violations in PR5:

If I run an optimization on the file in PR5, this is about as close as I can get:

With constraints violated quite severely:

[abelsm2]

Adding Zip file that contains the MAT file that gives the output shown above.

LVD - KSS Explorer To Laythe.zip