Day 5:

• Last day of work! So stressful. I guess I'll have to write my paper at home, a prospect that does not bode well but since I guess we have the added pressure of a real paper to submit I might have more impetus to do it in a timely fashion.
• We all went out for lunch at Kebab and Curry, which was nice and pleasant.
• I did final bookkeeping things like returning keys and changing all the directory permissions on Oscar so Marco could have access to all my files. I also had to take the code I'd been cranking out all week and make it readable/understandable so it'd be easy for Marco to work on. I'm a bit sad to be leaving and hope I can get some time to contribute more before the school semester starts.
• I wrote up a pseudo-code type thing for Marco and at the meeting with Dr. Bahar they both seemed to think it was a good idea to actually include pseudo-code in some discussion of our process of making chains, etc. I'm starting to feel a little daunted by the idea of contributing a whole section to a real life paper, but we'll see what happens!

Day 4:

• My Smartspice jobs finally ran! I got news back from the CCV people who said because I had really low priority access and more people with higher priority access might be running jobs, there'd be more of a delay. Of course, the jobs I waited for days to run ended up to have syntax errors so they didn't work anyhow. I fixed up the errors with Marco and sent them to run again but the estimated run date is like 8-22 so we'll see what actually happens.
• Still chugging away on my script. Tomorrow Dr. Bahar, Marco, and I are meeting to outline a paper and discuss what the structure should be and what we can fit into the paper.

Day 3:

• Yesterday I was having trouble getting the new chains script to actually make chains, and upon inspection I saw that the script was visiting all the nodes (and marking them as "visited") and then not ever unmarking them as visited, which was a problem. So I had to dig through the old script to figure out where exactly it marked all the "node.visited"s as false. It took me a really long time to find it. But then I did and I stuck it in and everything worked as expected!
• The next task was to then change the chain selection algorithm so it stopped "scoring" chains based on the old set of metrics and used our new simplified set of metrics and only chose ONE chain.
• There are some complications with this task, as sometimes there are nodes that are reinforced like one node away? So maybe one output reinforces an output almost entirely, but there is one inverter or something separating the outputs. This complicates the script a bit and the old code couldn't handle this case. We may need to rethink how we understand "reinforced" outputs and such.
• Also, for some reason the script was skipping including the output node in the chain. Also, because we decided to use the implications as nodes and just craft a "shortest path" algorithm rather than just make implication chains, which I think is a good choice, there needs to be some fancy footwork implemented to actually output a set of implications and not just the graph node names, which is what I'm doing now.

Day 2:

• My Smartspice jobs are still sitting in the queue. I've sent emails to a variety of helpdesks and people who may be able to help me with my problem, so we'll see what happens.
• I asked Marco what I should do and he asked if I'd fixed up the new chains script and I that's when I realized that when he showed me what he'd been working on he was doing that with the purpose of having me contribute (duh). So I sat down to improve the new script. It basically does nothing right now, but it does do some testing to make sure the sorting of implication sets is correct, so I finished that up as a "proof of concept" bit, and then set to work getting it to have all the old functionality of actually making chains.

Day 1:

• Smartspice was down today, it seemed like none of my simulations were working. I'm not sure if it's a problem with me or a problem with licenses, I have all these weird problems (no multinode license, jobs sitting in the queue, etc)
• I took the day to collect sources for my final paper and look over Marco's "improved" script for making chains. I think he wants to discard the old methodology entirely and make a script that just weighs the implication activation probability and distance, and constructs just one chain. He also wants to do some thing where you add the imps as graph edges but that starts to get confusing with node pointers and where the edges begin and end.