representations.There were no such efficient representations other than the constraint graphs until the SP [1]and the BSG [2]appeared in the mid-1990s.Subsequently,several efficient representations were proposed,such as the O-tree [9],the B'-tree [10],the corner block list (CBL)[11],the TCG [3],the TCG-S [4],the twin binary sequences(TBS)[12],the CS [5], and so on.However,most stochastic optimization algorithms using these representations were SA.Moreover,these representations were designed for rectangular blocks and cannot handle rectilinear blocks directly A.Stochastic Optimization Algorithms In the field of floorplanning,SA is much more popular than EAs.Several researchers used general SA to search solution spaces [1]-[5],[10]-[12].They perturbed the current floorplan according to properties of the employed representation and decreased the temperature according to a pre-defined cooling schedule.Additionally,on the basis of the SP, [13]proposed an orthogonal SA algorithm (OSA)with an efficient generation mechanism (EGM)for solving floorplanning problems.The EGM sampled a small number of representative floorplans and then efficiently derived a high-performance floorplan using a systematic reasoning method for the next move of the OSA,based on the orthogonal experimental design. In recent years,EAs have attracted increasing attention because they are suitable for solving complex and ill-defined problems.They have been successfully applied to the fields of numerical optimization [14],constraint satisfaction problems [15],data mining and knowledge discovery [16],[17],neural networks [18],and so on. There were also a few studies on the application of EAs to floorplanning problems. 66 representations. There were no such efficient representations other than the constraint graphs until the SP [1] and the BSG [2] appeared in the mid-1990s. Subsequently, several efficient representations were proposed, such as the O-tree [9], the B* -tree [10], the corner block list (CBL) [11], the TCG [3], the TCG-S [4], the twin binary sequences (TBS) [12], the CS [5], and so on. However, most stochastic optimization algorithms using these representations were SA. Moreover, these representations were designed for rectangular blocks and cannot handle rectilinear blocks directly. A. Stochastic Optimization Algorithms In the field of floorplanning, SA is much more popular than EAs. Several researchers used general SA to search solution spaces [1]-[5], [10]-[12]. They perturbed the current floorplan according to properties of the employed representation and decreased the temperature according to a pre-defined cooling schedule. Additionally, on the basis of the SP, [13] proposed an orthogonal SA algorithm (OSA) with an efficient generation mechanism (EGM) for solving floorplanning problems. The EGM sampled a small number of representative floorplans and then efficiently derived a high-performance floorplan using a systematic reasoning method for the next move of the OSA, based on the orthogonal experimental design. In recent years, EAs have attracted increasing attention because they are suitable for solving complex and ill-defined problems. They have been successfully applied to the fields of numerical optimization [14], constraint satisfaction problems [15], data mining and knowledge discovery [16], [17], neural networks [18], and so on. There were also a few studies on the application of EAs to floorplanning problems