I have a mps file given to me by someone else (our collaborators). I read the MPS file into my program with GLP_MPS_FILE as the argument to glp_mps_read function (I am using C API). In the program I change the right hand sides of some of the constraints (which are already there in the model). I am observing that the read model considers the objective function as constraint number 1, then the first actual constraint as constraint number 2, and so on. This is because, only when I consider this numbering of constraints I get the same results as reported by another solver (which I know is giving the right values). Is my observation correct ??