PORTER: SOFAR PROPAGATION OF WIDE-BAND SIGNALS TO LONG RANGES 



a non-ray tracing. This program realizes the fact that there are 

 discrepancies among various propagation loss models, which is the 

 basis of these types of programs. It sort of allows you to choose 

 what type of propagation loss model you want to use. It's just a 

 bookkeeping system. It just didn't seem to fall into any of the 

 workshop discussions that we have had. 



Dr. Hersey: Possibly what we need, I know there's been some 

 discussion of this within the Maury Center, is to focus on what we 

 can do with broadband programs. Does anyone feel any great ^enthusiasm 

 for that? 



Dr. F. D. Tappert (Courant Institute of Mathematical Sciences 

 of New York University) : This issue actually came up at the last 

 AESD workshop in the form of how one would write programs to get 

 impulse response functions. On one level it was a trivial thing. 

 You take any normal mode program or parabolic-equation program and 

 run it at a thousand frequencies and get out the impulse response. 

 It just takes an enormous amourit of computer time. 



I think what we have to learn is how to use the most modern 

 computers. For example, the Illiac computer is especially designed 

 to process in parallel 64 different frequencies and I think there is 

 a need to get access to computers like this. If we had time on these 

 machines, we could, by very straightforward methods, get impulse 

 response curves. 



Dr. Hersey: 1 think you have brought up a great point. 

 Dr. Weinberg. 



Dr. Smith: Excuse me. May I demur on principle? No matter 

 what computer you had, there will always be something that you can't 

 compute with it. You will want a stronger one. I think we have to 



662 



