[Help-glpk] intermediate solutions in the dual Simplex

[HBuesching04]

I am currently working on a project, which includes the study of adding 
a constraint to a given LP. I want to reuse the old dual optimal 
solution for finding the next solution. I am only interested in dual 
solutions. So in fact I want to use the dual simplex algorithm. 
Furthermore I am interested also in the intermediate dual solutions. 
That means that each step where dual solution is improved shall be 
noticed with the those dual variable values.

I want to use GLPK for this purpose and I made small investigations, 
which I want to verify by contacting this forum:

1. Starting the dual Simplex with a given starting point is not 
possible via glpsol.
2. I should take the function spx_dual_opt as an example, modify it and 
furthermore write a data providing routine for this function.
3. For each intermediate solution I will have to recalculate the dual 
variables, because of the usage of slack variables in the dual simplex 
algorithm.

Are all three investigations true?
Has someone further ideas? For instance it is still not too clear for 
me how to manage point 3.

Help is welcome,
Harald.