I have a circuit with waveguide components modeled via calculated data in a generic MDF file, accessing via a DAC. When I simulate the design, the results are as expected. However, when I optimize(over subsets of the modeled frequencies), the circuit improves as expected, but upon completion and resimulation with the updated optimized variables, the results change. In fact, while optimizing(and updating the display), the response actually shows a discontinuity, not seen during normal simulation. The MDF files seem to have plenty of points in the regions simulated, and I have tried it with all of the interpolation methods with similar results. Are there any reasons why the optimized updates are different than the regular simulation? This also occurs with all the different optimizers.