Thank you, Mike. I'll get on it right away.

If you use my code generator, you'll get a better result from the NN code if you run part5.prg last so the order will be 1-2-3-4-6-7-8-9-10-11-5

>Grady!
>
>My neural net was percolating while I was driving in the snow storm the other night! Rather than using a single table to handle 35,000 x,y coordinate pairs, you can use multiple tables/cursors. VFP allows 32,000 cursors to be opened. I think that means you could have a single 2-D "array" that is approximately 32000 * 1.2 billion elements.
>
>>>>>>>>From my testing, I have found that I can set up an array as big as 1000 by 1000 elements and maybe bigger. Does anyone know the limits to an array size or is an array only limited by memory?
>>>>>>>
>>>>>>>In VFP9 array size limit is 2GB. It's also limited by available memory. In VFP8 and earlier, the limit is 65,000 items.
>>>>>>
>>>>>>Thanks Sergey. It's nice to find that much room to play with.
>>>>>
>>>>>Better check message: 1366284
>>>>>
>>>>>Unless you know exactly which element # you need to access, an indexed cursor can be better. There's a good write up on the fox wiki, but it's not responding right now.
>>>>
>>>>How many columns/fields will an indexed cursor handle?
>>>
>>>Well, the typical answer is 254. But, if you made a table with x, y, content fields you can make an indexed cursor with a maximum of 1.2 billion rows (until the 2Gb limit) which could be at used to make an "array" approximately 35000 x 35000 "elements".
>>>
>>>Why are you trying to push these limits? Neural net stuff again? Cool!
>>
>>Yes, Grasshopper. Neural Net stuff again. This time a self adjusting version that may be applied to any numeric column in any VFP table. . . . if all goes well.