For some reason I have a lot of things on my mind this morning. It’s pretty cluttered!
I started working on the informs thing last night, and instead of actually Getting Shit Done, so to speak, I got stuck on this issue of smartly dealing with symmetric matrices. Basically, as far as I can tell, nobody has really bothered to do anything with this. My thinking is, if you know that A_i_j = A_j_i, then why bother storing A_j_i ? A real conondrum.
After messing with it for a while though, I didn’t bother really doing anything about it either. This is because I wanted something that looked and acted like a numpy array but knew it was symmetric underneath the surface. Couldn’t really figure that one out though. Nyoro~n
Which reminds me! I’m going to the SciPy 2010 conference in about a week! EXCITING
The reality of having to travel somewhere exotic (Austin is exotic shut up) hasn’t really hit me yet though. That, and I’m probably gonna be in way over my head as far as relative skill levels goes. But hey, hopefully it’ll be fun!
Not that I know what catamorphism means. But since Pete Krumins can explain polymorphisms in C++, maybe he does. Unfortunately, I think Proggit kinda missed the point. When I read the article, I got the feeling that it was more about What Polymorphism Is, and how it relates to something you may already know from C++. I think Proggit, on the other hand, was like, “But we already know C++!” Silly peeps. I’m probably biased though, since I chat with pkrumins on irc from time to time.
On a completely different note, I started using gmsh with my research. Or, well, trying. I’m starting to think that there’s no such thing as a “good” FEM solution, and that you really just end up trading one obnoxious quality for another. For example, COMSOL 3.5a is HORRIBLE for scripting. I mean, just horrible. It’s so bad that I tried to get Sikuli running so that I could automate a gui build process. ( Unfortunately, Sukuli is still pretty tempermental in linux, especially for 64-bit. Makes me sad, because Sikuli looks rad. I guess I’ll have to leave it to those Mac users. )
In contrast, gmsh is fairly easy to script up, though it does take a while to get the hang of it. The gui is handy, but it’s like Dreamweaver in that you really want to switch back and forth between the gui and scripting. Also, the examples are all old, from a time when gmsh had a pretty different syntax.
The big thing I’ve noticed with gmsh though is that it doesn’t seem to fail gracefully. My current model has some scaling issues, and while comsol also had issues with characteristic length oddities, it was also able to fail at meshing without dragging the rest of the computer with it. I mean, it would segfault on a problem solve all the time, but at least I could write a crappy supervisor script to restart it. For the record, I have a list of more supervisors to try out in the future. I don’t know why, but my google-fu was failing me when I went looking for them before.
I have some older odds-and-ends on my old web page that I’m not dragging along with me, and the OLD STUFF sidebar is kinda dumb so I’m gonna type about it here and then feature things I think maybe actually deserve to be featured, if I care to.
First: Exam II from Industrial Processes. I had a Hell of a time with that course. I thought it was really boring, to be honest, and, I mean, I like Making Stuff. That wasn’t the issue. I really can’t do rote memorization of the parts names of machine X that I’ve never seen/used though, nor do I really want to spend too long talking about the importance of hardness in tool selection. I mean, yeah, that’s important (because soft tools get pwned by hard materials), and I get that, but the course was a lot like a more useless version of high school biology to me (and I didn’t like vocab then either—one time I didn’t look up anything when doing a vocab assignment and wrote that a gestation period == digestation period. Got some laughs from the teacher and some embarrassment for myself. GOOD TIMES) and, tbh, I was going through Hell with my personal life right around the same time. So I was kinda hosed. I tried to study. I really did. But it’s hard when you don’t care about anything and don’t like the material anyway.
So what did I do? Well, I was a total dick and smartass about it, as maybe you can see. My classmates thought it was the funniest thing, and thought the rest of the internet would like it too. I’m less sure.
Second: Matrix solving in Excel. Engineers loove Excel and hate MATLAB. So, I explored the possibilities once. I don’t even really remember what this particular example has. I think it might be two examples—one which uses matrix inversion, and another which uses a least squares goalseek algorithm.
Third: I tried doing interactive fiction once on some forums I used to frequent. Of course, enthusiasm sputtered out pretty quickly.
Speaking of ComicGenesis, I drew this webcomic in high school. It was a good learning experience, but like most high school projects it’s pretty embarrassing now. Still, some people seem to extract lulz from it.
Which reminds me: Jekyll and autokeen have a lot in common. Well, except that Jekyll has more features and is blog-oriented instead of comics-oriented. I think Jekyll could totally be modified to be more comics-friendly though! Just an idea.