@Carl Love 

It is great to get this workaround - thank you! It makes life so much easier ... now I can visualise the relationships of the solutions.

Thanks (again!) for your generous support.

Thank you, Carl. 

I tried that - but doesn't the adjacency matrix has 2 non-zero diagonal entries? If so - then the routine won't run.

Perhaps I am misinterpreting your suggestion (I am new to Graph Theory) - in this instance, the solution to my LP problem is given by the square matrix,Y, and will change as certain conditions / values are altered in the model. In other words, the entries will vary on a case-by-case basis and I simply wish to show these on a graph.

Thanks again to all for addressing this question - as I don't use these methods and techniques, I find that the support you all gave is generous and well-constructed. I have learned a lot today.

Last question .... is there a convenient way to allocate a name to each i and j node?

Thanks for this - this is a great help and is very much appreciated. Nice solution!

However, can we separate each node i in the graph? For example, can we show node 2 (i=2) connected to the j-nodes, 3, 4 6 and 9 alone?

In the same way, I'd like to show node 3 (i=3) simply connected to the j-nodes and in the case where there are no connections from i to j, can we simply omit these? 

I'm only interested in the plots that show the active connections, i.e. in the case where the entries are 1's and not 0's.

@Carl Love 

Thanks for the insights here. I'll give this a go also!

@tomleslie 

Yes ... that does the trick!

Thanks a lot - I'm grateful.

@Carl Love 

Yes indeed .... I frequently lost my connection also for small p.

I figured it was pc-related rather than numerical computation issues, so I removed firewall, re-booted, etc. but it still made no difference.  

@acer 

Nice model. I varied the parameter values across my range of interest and it produces outputs in quick time.

Thanks very much .... and thanks again to all who worked on this.

@Carl Love 

This looks very interesting for small p values. I'll apply this to my problem. 

Thanks for the creative input (as usual!).

@acer 

Interesting and useful routine .. thanks for this insight.

@epostma 

At least some good arises!

I'm glad you picked this up, Erik - thanks for informing us.

Is it possible to obtain a pre-release version of the fix before the next release?

@mmcdara 

That's a beautiful method of solution .... such a simple and creative approach.

Again - thanks for this. 

@mmcdara 

Now, that's impressive! I'll take this into consideration.

Thank's for the guidance and quick response.

@mmcdara 

 I appreciate this. Nice code.

This works nicely when p=0.5.

However, when p is adjusted to p=0.05, the processing time remains excessive. I ran this for 10 minutes and decided to stop the computation. 

NBD_Issue.mw

@tomleslie 

Thanks. I hope you can access the code here.

I'll try the 'numeric' option.