Hi-tree Layout Using Quadratic Programming
Horizontal placement of nodes in tree layout or layered drawings of directed graphs can be modelled as a convex quadratic program. Thus, quadratic programming provides a declarative framework for specifying such layouts which can then be solved optimally with a standard quadratic programming solver. While slower than specialized algorithms, the quadratic programming approach is fast enough for practical applications and has the great benefit of being flexible yet easy to implement with standard mathematical software. We demonstrate the utility of this approach by using it to layout hi-trees. These are a tree-like structure with compound nodes recently introduced for visualizing the logical structure of arguments and of decisions.
KeywordsQuadratic Program Convex Quadratic Program Layout Algorithm Contextual Node Horizontal Placement
Unable to display preview. Download preview PDF.
- 4.Bohanec, M.: DEXiTree: A program for pretty drawing of trees. In: Proc. Information Society IS 2007, pp. 8–11 (2007)Google Scholar
- 8.Marriott, K., Sbarski, P., Gelder, T., Prager, D., Bulka, A.: Hi-Trees and Their Layout. IEEE Transactions on Visualization and Computer Graphics (to appear)Google Scholar
- 9.Sbarski, P., Gelder, T., Marriott, K., Prager, D., Bulka, A.: Visualizing Argument Structure. In: Proceedings of the 4th International Symposium on Advances in Visual Computing, pp. 129–138 (2008)Google Scholar
- 10.Walker I, J.Q.: A node-positioning algorithm for general trees. Softw. Pract. Exper. 20(7), 685–705 (1990)Google Scholar