This week I spent a majority of my time reading up on data depth. I didn't feel that my knowledge base was enough to start research, having had only conversational exposure to the topic. I used the Tufts Comp Geo website to find appropriate papers, as well as following citations in the papers I read. Overally, my brain is feeling very very full of new information. I'm trying to stem the feeling of inadequacy that must come to everyone the first time they enter the world of research.
Tuesday I had a short meeting with Diane just to get me officially started. Wednesday afternoon was a meeting of the research group, followed by a more substantial meeting between Diane, Eynat, and me. They were very supportive and we decided to meet early next week to decide specifics of what I wanted to work on for the rest of the summer.
As for more personal things, I spent the evenings moving into my new room and laying outside reading. I've started studying for the GRE's, because grad applications and testing are coming up within the next few months. Boston summers are much hotter than I anticipated, and without air conditioning it's been a bit interesting, but I'm getting used to it. Getting back and seeing friends has been great, and adds an extra bonus to spending every day doing something I love.
Monday: Today I spent all morning doing more reading, mainly focusing on the Tukey median and ways of either calculating it or approximating it. I also looked a bit at rank contours, because I'd like to find a way of breaking ties in the ranking of points. After lunch with a friend, I went back to the lab and studied Hamiltonian triangulations, a topic I'm presenting at this week's research meeting on Wednesday. With my last few hours I wrote this website up to here, which was interesting because my html has gotten quite rusty since my last website class. I have to admit I had a moment of doubt today, as ideas kept flying into my head and I kept finding papers already published on the topic. I'm feeling better now, especially with a meeting with Eynat scheduled for tomorrow.
Tuesday: Today I spent the morning looking over more papers, and had an idea about approximations of certain depth measures. I met with Eynat, who liked my suggestion, so now I have my official specific subjects. I will be looking at sorting methods for rank contours and approximation of depths using proximity graphs. I then had a library information session that showed which databases are best for researching computer science papers. Afternoon was spent with more paper research and some updates to the website. Tonight I went out to trivia night at a local bar, where my team lost spectacularly. We can't compete on random trivia from the 1970's.
Wednesday: I have very specific topics now, as I'll be using relative neighbor graphs and Gabriel graphs to create a new data depth measure. I met with Eynat again today, followed by the group research meeting, which I didn't have to present for after all. That will come next week. Tonight there was a barbeque for everyone on campus doing research, and I was surprised by how many people I knew there. It was good (free) food and good times.
Thursday & Friday: Both days I spent looking at characteristics of a couple different ideas of depth measures. Both of them are really interesting because they combine data density with the depth for a kind of hybrid. Because of this, though, they don't adhere to many of the "desirable traits" of data depth. I can't wait to meet with Eynat and Diane and see what they think, because this could be really cool. I spent all afternoon on Thursday where my friend works in Davis Square researching there so that she could have some company. I was super excited to get my desk, but it was also nice to get away from it for a bit but still be doing the work.
Tuesday: Well, I spent the weekend in NYC with my best friend, watching the most amazing fireworks ever from her roof in Queens. Now it's back to work. Today I looked at a third kind of possible depth measure, although one I don't have a lot of faith in. Then, after lunch I switched to my other topic, breaking ties in the ranking of points. This one I don't feel like I have the same handle on, so it's slower going and harder thinking. The really interesting thing is that so many different ways of ranking points have really identical results except in extreme cases. I also started a class I'm taking, called Literature of Chaos. Much as I hate to admit it, it has potential to be really good.
Wednesday: I spent time today preparing the presentation I have to make to the research group tomorrow afternoon. I have the whole hour, and am doing a hard paper about Hamiltonian triangulations, so this should be interesting. I also had a really good idea about proving that angles as rank tie-breakers are valid, but I was shot down by one of the grad students here. He was very gentle, but it was still a shoot-down. Tonight is another of the research dinners, this one with a speaker, so that should be nice.
Thursday:This morning I looked into extending the idea of Delaunay Triangulation distance to the convex hull into cover contours using the Voronoi diagram or a derivative of it. I met with both Diane and Eynat, both of whom seemed excited about what I've been working on. Then we had the research group meeting, where I presented a paper about Hamiltonian Triangulations. Afterwards I talked to Diane about life in general. It's really nice to have someone who knows exactly what I'm doing and has been there before.
Friday: Today was great. I spent the whole day programming a Gabriel Graph visualization and had lunch at a little cafe where my best friend works. I had really missed writing code during my weeks of reading and thinking and drawing, so it was a nice change of pace. I spent more time on the cover contours, but they're turning out to be somewhat problematic. This weekend will be my first one here on campus, which will be nice and relaxing. No traveling this time. Instead I'll finally get my room settled and get to go out with my friends.
Monday:Well, the weekend was really good. Being on campus and getting all my crazy errands done was good, plus I got to go out with friends at night. It got super hot, but I'm getting used to it finally and the fan helps. Today I finished my program. I don't think I've ever written something this complicated this quickly, and I'm really proud of it. So far it calculates the distance to the Convex Hull of every point along either the Gabriel graph or the Relative Neighbor graph. I'd like to add the Delaunay triangulation, but it's a beast to code, so we'll see about that. Now I'm going to play with my new software and then read up on linear time triangulation of simple polygons for the group meeting on Thursday.
Tuesday: Today I solved the mystery of how to create a Delaunay triangulation I could actually work with. It's interesting really, because turns out LEDA's built-in functions aren't so terrible to use, once you get used to looking up everything you need. I've also been exchanging emails with Diane about a computational geometry conference in August that I really really want to go to. We'll see...
Thursday: Well, I'm officially going to the conference! We'll be flying up on the morning of Wednesday August 10th and coming back the night of the 12th. I'm so excited! It's the perfect opportunity to get to know Computational Geometry people from other departments and scout out where I might want to go next year for grad school. Yesterday I spent even more time programming up my algorithms, adding file input as an option. I met with Eynat in the afternoon to go over the creation of data sets and options for contours. My friend's 21st birthday is today, so went out last night to celebrate.
Today I had an early meeting with Diane. It went really well, and she seems excited about the depth measure I'm looking into. I spent the early afternoon reading up on linear time triangulation of simple polygons because it's the subject of our meetings for the next 3 weeks. Then we had the meeting, which was quite interesting because it was 5 of us digging through really a difficult paper. I'm looking forward to sleeping a lot tonight, something I haven't done for while.
Friday: I spent this morning adding file input to my program and creating a few data sets to test on. That actually indicated a bug which required most of the morning to track down. After lunch, I started adding the ability to find local maxima in the depths, but that proved extremely problematic and I left this evening still not having found the problem and very annoyed.
Tonight is the release of Harry Potter and the Half Blood Prince, so my friend and I are going into Harvard square to the midnight magic party. Yes, we're very big dorks. The plan right now is to spend a vast majority of the weekend reading the book, because realistically I won't be able to put it down until I finish it. We're also going to the best burger place in New England for dinner (I'll be having a veggie burger) which I'm looking forward to a whole lot.
Monday: Well, the weekend was full of Harry Potter as expected. Friday night was absolutely amazing, and I bumped into someone I went to a high school camp with back in Missouri, which was very very strange. I finished the book last night after savoring it as much as possible, and am now frustrated because very few people have finished, so I have to watch what I say. I really just want to talk about it constantly.
Today I spent the morning fixing the small bugs in my program, creating more interesting data sets, and adding the ability for multiple file inputs. I also found the bug and got the local maxima working. This afternoon I added a really interesting ability to see the edges that are traversed when assigning depths to the Delaunay measure (see the research page for a picture) and have been thinking about ways of finding an in-between graph of Gabriel and Delaunay. I guess we'll see.
Tuesday: Today was overall really productive. I got contours working for the Delaunay depth and fixed a few bugs that popped up. I made the program more user-friendly and found a long-standing bug because of an idea that I got from my contour bugs. I've been thinking more about in-between graphs and at the same time trying to find a way of fixing the Gabriel graph so that it works a little better but still has the same low time complexity.
Wednesday: I mostly brain-stormed today, and took a break from Halligan to work in the campus center for a while. I had a chiropracter appointment this morning and feel SO much better, which is definitely a bonus. I met with Eynat for our weekly catch-up, and things are looking really really good. Tonight is one of our research dinners on campus, this one dressy and catered so I have to go prepare a little bit early. Tonight I have to organize my thoughts for my meeting with Diane, as well as get to reading the linear triangulation paper we're working on again tomorrow at our group meeting. I also talked to Elena, a PhD student here, a lot today about applications for grad school. It's really nice to have someone that's been here so recently and can help, because it's seeming really overwhelming right now.
Friday: Yesterday I added the feature of having a level of removal when calculating the Gabriel graph seeds, in order to fix the fact that there were too many. We had our weekly group meeting and I had my weekly meeting with Diane. The linear time triangulation paper is interesting as ever. Today I worked on my progress report, summed up all my work so far so that it'll be organized when I start writing my paper, and took care of other house-keeping type things. I also programmed my new idea for a proximity graph and am playing with the specifics of it.
Monday: Well, I'm officially over half way through. Kinda scary, really, it has flown by. Today I added my new graph to my general program and got it working pretty well. I then programmed up the seeds for that graph, and spent the afternoon trying to figure out what is going wrong. So far I have completely failed on that front. Michael played chess, which was interesting because he is a chess God and I haven't consistently played since I was about 12. Hopefully I can pick it back up and get decent, it's always been a game that I liked and wanted to be better at.
Tuesday: Today I found so many bugs in my seed assignment that it's absolutely pathetic. I'm somewhat confident now that they're out, and have been running tests to see which of my hybrid graphs performs the best. I compiled the data and made a graph so that it's easily visualizable which are best. That actually took quite a while, because I had to generate lots of sets and get the data. I also met with Eynat today and talked about my paper and Latex, which I've never used before. I'm starting to feel overwhelmed by the amount of work I'll be doing in the next month, but I know it'll be really satisfying. During lunch I played chess and got severely beat again. Eh, I'll get better...eventually...
Wednesday: I had so much fun today. I spent the early morning making last minute fixes to my data on the graphs, then had a really productive meeting with Diane and Eynat. The three of us and Mike went out to lunch then instead of having our group meeting for the week, and I completely stuffed myself on good food. After that I started making a stripped down version of my program that I can use for analysis purposes on a ton of data sets. I still am finding other things to do rather than work on the paper...it could get interesting toward the end. Two weeks to the CCCG conference!
Friday: Yesterday I spent the morning reading over Michael's master's thesis and giving suggestions. That was interesting, because a lot of his work I haven't been introduced to before. Once I was too tired to keep plugging through that, I worked at stripping down my code and making it generate its own point sets. I'm going to get it finished and run it on 100 point sets today to see how the different graphs compare and function. I'm planning on then determining the level of separation required for the bimodality to come into effect, but that will take at least a few days.
Hopefully this weekend I'll get to see Amy, my swim coach from home who's taking a trip up here. Haven't heard from her for sure yet, but we'll see. My friend Dave is also back from Germany, and I get to hang out with him tomorrow.
Monday: Wow, I can't believe how fast this is going! The weekend was nice, I got to hang out with my friends at nights and I spent the days studying for the GRE...joy joy. Amy did come for a couple hours, but refused to take the T anywhere (being solidly from the midwest) so we couldn't really do much. It's sad how attached to our cars we are, really.
All morning I've been working on getting out this random bug I discovered on Friday. Will they never end? It's not totally gone, but its frequency is considerably lowered, and that part of my program I'm starting to not like anyway, so I'm leaving it in. I also started writing up a new idea Eynat had about ways of weeding out inaccurate local maxima. It's not completely up and running, but I'm working on it.
Thursday: Well, the last three days have been essentially identical. I spend the mornings setting up the repeated run of my program that then takes all afternoon to complete. I've shot down a couple of ideas, had a couple more. Overall it's ok, though, because the ideas that have been problematic are my ways of viewing the characteristics of the depth measure, not the measure itself. Afternoons I've been spending working on my paper. It's really hard to have to explain everything so carefully from the beginning. It's also interesting because I have to actually name the things I've been creating, rather than using my dumb names for them. This morning was a disaster and I'm not running code right now because I couldn't get what I wanted working. Instead I'm trying to learn how to use Latex so that I can make my paper actually look like a paper.
Friday: I got one of my analysis things working this morning, only to find out it told me almost nothing. Overall, not my best week. Everything's going wrong. I spent most of today again working on my paper and figuring out Latex.
Overall, an amazing summer. I got to experience research before heading off to my grad-school studies, I got to know Diane and Eynat so much better, and I found out that I'm much more capable than I had thought. I'm also excited that my interest in computational geometry hasn't waned even through my hours and hours spent on it over the summer. I've found my field. With that, I'm off to begin a new year of classes and many many grad school applications.