Performance evaluation
Customized IP method vs CVXHere, we compare the performance of our customized primal-dual interior-point algorithms with CVX. The direct algorithm uses Cholesky factorization to compute the search direction, and the indirect algorithm uses the matrix-free preconditioned conjugate gradients method (with diagonal preconditioner). We set and choose the state weight that penalizes the mean-square deviation from the network average . Figures below compare scalability of direct and indirect algorithms with CVX.
The solid lines in the following figures show ratios of the running times for CVX and our customized algorithms versus the number of nodes. As indicated by the dashed red lines, on average, algorithms based on Cholesky factorization and PCG method are about and times faster than CVX (for network sizes that can be handled by CVX).
Influence of the regularization parameter on performanceNext figure illustrates the influence of the regularization parameter on the performance of the matrix-free PCG algorithm for an Erdos-Renyi network with nodes.
IP method vs proximal algorithmsTo compare the performance of proximal gradient and Newton methods and illustrate scalability of these algorithms, we solve the problem of growing Erdos-Renyi networks with different number of nodes. Figures below compare scalability of primal-dual IP algorithm, proximal gradient (proxBB), and proximal Newton (proxN) algorithms.
The solid lines in the following figures show ratios of the solve times between the IP method and the proximal algorithms versus the number of nodes. As indicated by the dashed lines, on average, proximal gradient and proximal Newton algorithms are about and times faster than the customized IP algorithm (for network sizes that can be handled by the IP method).
Performance of different customized algorithmsThe following table compares our customized algorithms in terms of speed and the number of iterations for the problem of growing connected resistive Erdos-Renyi networks with different number of nodes and . We set the tolerances for and to and , respectively.
We see that the implementation based on PCG method is significantly more efficient than the implementation based on Cholesky factorization. Even though both of them compute the optimal solution in about interior-point (outer) iterations, the indirect algorithm is superior in terms of speed. Furthermore, for , the direct method runs out of memory and the indirect method computes the optimal solution in about seconds. Moreover, even for such a small network, proxN and proxBB are significantly faster than IP (PCG) algorithm. Larger networksWe next illustrate performance of the customized algorithms for the problem of growing connected resistive Erdos-Renyi networks with the edge probability and . The results for larger networks are shown in the following Table. In particular, for a network with nodes and edges in the controller graph, it takes about four and a half hours for IP to solve the problem. However, proxN and proxBB converge in less than and minutes, respectively.
Additional computational experiments indicate that these two algorithms can handle even larger networks; for example, the problem of growing the Erdos-Renyi network with nodes and edges in the controller graph can be solved in about and minutes for proxN and proxBB, respectively (with Matlab implementation on a PC). Implementation of these algorithms in a lower-level language (such as C) may further improve algorithmic efficiency. Alternative performance criterionWe next use our customized algorithms for adding edges to an Erdos-Renyi network with nodes and edges in the controller graph. The following shows the relative error in the objective function in logarithmic scale versus the computation time. Here, is the optimal vector of the edge weights obtained using primal-dual interior point method. We set the tolerances for and to and , respectively. Even for a problem of relatively small size, both proxN and proxBB converge much faster than the IP algorithm and proxN outperforms proxBB. Thus, a possible less sophisticated criterion for termination is the relative change in the objective function value, We have examined this criterion for our customized algorithm. Not only did not we encounter false termination but also the accuracy of the obtained solutions were very high. However, for a fair comparison with IP algorithm, we report the results using the generic stopping criteria in Remarks and . Proximal gradient vs fast greedy algorithmWe next compare our proximal gradient algorithm with the fast greedy algorithm of Summers et al., 2015. We solve problem for Erdos-Renyi networks with different number of nodes ( to ) and . After proxBB identifies the edges in the controller graph, we use the greedy method to select the same number of edges. Finally, we polish the identified edge weights for both methods.
|