Conclusions and future work

Hopefully, members of the acustic community will find TRACEO a useful and reliable tool of research. The model was written with modularity, stability and accuracy in mind and the examples and comparisons presented in this document are expected to speak by themselves regarding the potential of the model. There remain, however, several degrees of freedom worth of further development, namely:

• Code optimization; analysis of the subroutines will reveal repeating patterns of code, which some programmers will consider annoying. But such patterns exist in order to make the code as understanble as possible, and anybody capable of improving the code is invited to do so and to share such knowledge with the community.
• Translation to other programming languages; generally speaking Fortran statements are efficiently converted into machine code, but there are limitations of the Fortran 77 standard (like the need for initial memory allocation and no access to pointers in memory), which are higly desirably for optimization and/or for the development of interfaces^8.1.
• Parallelization (and cloud computing); ray calculation is an ideal task for distributed calculations, either way by using multi-processor machines or by distributing the calculations in a computer cloud. It is an interesting option for cases, which require a huge amount of repetitive calculations.
• Extension to the three-dimensional case; among the modelling tools used in underwater acoustics ray tracing has no match in terms of computational efficiency; thus, three-dimensional modeling with ray models represents an important alternative. TRACEO was written in order to allow easily such extension, but some elements of the code will require an ellaborate rewriting, in particular, to allow the inclusion of three-dimensional objects, to update the dynamic equations after boundary reflections and to calculate the normals and the ray influence at a given position of the array.