Linear and quadratic objective function effects on force directed method based analog placement

This paper advocates the use of linear objective function in analytic analog placement. The role of linear and quadratic objective functions in the behavior and results of an analog placement algorithm based on the force directed method is discussed. Experimental results for a MCNC benchmark circuit and another one from text books are shown to demonstrate the effect of a linear and quadratic objective function on the analog constraints satisfaction and CPU time. By introducing linear objective function to the algorithm, we obtain better placements in terms of analog constraint satisfaction and computation cost than in case of conventional quadratic objective function.