Tolerance Settings in Simplex Method

Hello,

I've been using the simplex method (method=0 or method=1), for solving linear programming problems. However, I observed that setting a lower tolerance (e.g., FeasibilityTol=OptimalityTol=1e-8) resulted in faster convergence compared to a higher tolerance (e.g., FeasibilityTol=OptimalityTol=0.0001). This outcome seemed counterintuitive, especially considering the general expectation that looser tolerances might lead to quicker solutions due to fewer stringent checks. Did the observations make sense or what could be the problem?