We derive a technique for scaling the search directions of feasible direction methods when applied to optimization problems over Cartesian product sets. It is proved that when the scaling is included in a convergent feasible direction method, also the new method will be convergent. The scaling technique is applied to the Frank-Wolfe method, the partanized Frank-Wolfe method and a heuristic Frank-Wolfe method. The performance of  these algorithms with and without scaling is evaluated on the stochastic transportation problem. It is found that the scaling technique has the ability to improve the performance of some methods. In particular we observed a huge improvement in the performance of the partanized Frank-Wolfe method, especially when the scaling is used together with an exact line search and when the number of sets in the Cartesian product is large.

<p>Traffic planning in a modern congested society is an important and time consuming procedure. Finding fast algorithms for solving traffic problems is therefore of great interest for traffic planners allover the world. </p><p>This thesis concerns solving the fixed demand traffic assignment problem (TAP) on a number of different transportation test networks. TAP is solved using the Frank-Wolfe algorithm and the shortest path problems that arise as subproblems to the Frank-Wolfe algorithm are solved using the network simplex algorithm. We evaluate how a number of existing pricing strategies to the network simplex algorithm performs with TAP. We also construct a new efficient pricing strategy, the Bucket Pricing Strategy, inspired by the heap implementation of Dijkstra's method for shortest path problems. This pricing strategy is, together with the actual use of the network simplex algorithm, the main result of the thesis and the pricing strategy is designed to take advantage of the special structure of TAP. In addition to performing tests on the conventional Frank-Wolfe algorithm, we also test how the different pricing strategies perform on Frank-Wolfe algorithms using conjugate and bi-conjugate search directions. </p><p>These test results show that the updating shortest path calculations obtained by using the network simplex outperforms the non-updating Frank-Wolfe algorithms. Comparisons with Bar-Gera's OBA show that our implementation, especially together with the bucket pricing strategy, also outperforms this algorithm for relative gaps down to 10E-6.</p>

Traffic planning in a modern congested society is an important and time consuming procedure. Finding fast algorithms for solving traffic problems is therefore of great interest for traffic planners allover the world. This thesis concerns solving the fixed demand traffic assignment problem (TAP) on a number of different transportation test networks. TAP is solved using the Frank-Wolfe algorithm and the shortest path problems that arise as subproblems to the Frank-Wolfe algorithm are solved using the network simplex algorithm. We evaluate how a number of existing pricing strategies to the network simplex algorithm performs with TAP. We also construct a new efficient pricing strategy, the Bucket Pricing Strategy, inspired by the heap implementation of Dijkstra's method for shortest path problems. This pricing strategy is, together with the actual use of the network simplex algorithm, the main result of the thesis and the pricing strategy is designed to take advantage of the special structure of TAP. In addition to performing tests on the conventional Frank-Wolfe algorithm, we also test how the different pricing strategies perform on Frank-Wolfe algorithms using conjugate and bi-conjugate search directions. These test results show that the updating shortest path calculations obtained by using the network simplex outperforms the non-updating Frank-Wolfe algorithms. Comparisons with Bar-Gera's OBA show that our implementation, especially together with the bucket pricing strategy, also outperforms this algorithm for relative gaps down to 10E-6.