Hello again! This is my second post on this blog. The past two weeks have been very eventful! This post will include research updates and a big review of The Owl House finale! Read on if you’re interested, and see the Conclusion for contact info.
Research
Conference
I recently started working on a new research project. As you may know, I’m currently a Masters student in the United States, and I eventually want to receive a PhD in compilers. My research advisor is very nice, and they want to help me reach this goal. They’ve said I should try and get into a better school than I’m currently going to, though, so we want to release an extra special paper that gets me accepted into an awesome PhD program. I’m not all that certain on the timeline for this paper and how long it’ll take to do, but it should take somewhere between 6 months to a year to finish. Research takes a long time, not to mention the review process and the act of sending out the papers to conferences for them to read!
I’ve worked on one paper that was accepted before. It was accepted very recently, and I was a coauthor on it. The parts I worked on were very selective (I think I really just contributed on making a single benchmark use a new programming feature, when the paper consisted of maybe 8 benchmarks), so it was very nice of the other authors to still put my name on the paper as a coauthor. Although, since the paper was accepted, we were invited to go to the conference. The conference date was a little over a month away from the time we found we were accepted, but it was across the country!
If you know me, you’d know I’m a very shy person. I’m not very good at talking to people, and a lot of times I’ll delay in responding to people out of shyness. This doesn’t mean I don’t want to talk, and I’m trying to get better at it (don’t get discouraged from sending me messages!), but the conference made me very worried. Personally, I would much rather do the exact same thing every day, and my current workday is great because I go to the lab at 9am and leave at 9pm, with very little room left in the day to do anything but sleep. That’s the kind of schedule I love because of how consistent it is! Unfortunately, a conference with lots of “networking” and “conversing” and “engaging with people” is very out of my comfort zone. Also, that piled up with needing to learn how to stay in a hotel alone, or fly on a plane alone, or use rideshare apps, and a million other things was just too much for me, so I decided to stay back at the lab for this first conference.
A new thing I learned with this experience was how social PhDs needed to be. I always imagined it as a “work very hard doing research and let the business people handle the social parts,” but it requires attending a lot of conferences and networking with people and building connections with different institutions. That doesn’t discourage me from getting a PhD though, it just means I’ll have to work on my social skills a lot more if I want to succeed. I may not have gone to this conference, but I’m slowly hyping myself up to be able to go to the next one!
Polyhedral Compilation
Besides that scary experience, my new research project requires learning a lot of different things about compilers. I don’t want to give away too many details over what I’m researching, so I’ll just say its over a specific aspect of GPU compilers. One thing I wanted to look into was specifically polyhedral compilation, a very interesting concept I heard about online and wanted to learn more about anyways. You can learn more about the topic through this link.
Basically, polyhedral compilation is a way to remove dependencies in nested loops. Imagine you have a two-dimensional for-loop, in which the inner-loop depends on values obtained from previous iterations of the outer-loop. Normally, this would be very poor for performance since the inner loop in this example cannot be parallelized. There’s a strong dependency connecting loop iterations, so they must be completed in a specific order. Polyhedral compilation essentially converts this simple program snippet into a mathematical model (e.g., vectors, matrices, equations, and linear relationships, all containing only numbers instead of code), and then solves these models using some algorithm.
It’s a very interesting way of viewing code; program instructions seem like such a “physical” thing, corresponding to commands sent to hardware to control the actions of registers, caches, and other very real objects within a computer, that it’s mind-blowing how a mathematical model could reorganize it to run better. Anything defined in terms of math must be provably true by default, as math by definition is one of the only things in the world which contains undeniably true statements. You don’t get that in many other academic domains like physics or humanities, where statements are merely approximations of real-world events so inconceivably complex they cannot be predicted exactly in every scenario (for example, the force exerted by an object sliding against another due to friction can be approximated by an equation, but the equation only takes into consideration concepts on a macro-scale, even though the individual atoms that make up each object are the largest contributing factor in the determination of force). Due to everything in math stemming from “known” axioms, we can establish that some statement is true if it can be reduced into either axioms or other already-established true statements. The fact that a program can be visualized the same way is incredible!
I’ve read three papers over polyhedral compilation, listed at the end of this section. [1] and [2] are cited in many different papers due to introducing some very important ideas, and [3] further improves the process with some neat tricks. To summarize the main process behind polyhedral compilation, you (1) identify Static Control Parts within a program, which are sequences of statements containing only for-loops and if-conditions; (2) define an iteration vector that describes the range of the loop and a polyhedron—a special set of integer—for each statement and array reference; (3) create a schedule from the iteration vectors and polyhedra, containing the times at which each statement instance is executed; (4) produce a dependence polyhedra and set of equations from it representing all constraints the schedule needs to maintain; (5) solve the set of equations for those constraints to give a transformation matrix, bounding specific constraints when applicable to avoid infinite solutions; and finally (6) plug the transformation matrix, schedules, and polyhedra into a special polyhedral code generator to give completely restructured—but semantically correct—code. That last part isn’t explained in-depth in the papers I’ve read, so I’m reading up on that process next.
Below is a nice little figure from [3] that gives a visual overview of the process. Each polyhedral code generator essentially generates the transformation matrix their own way with some algorithm, but they share lots of consistencies in the way they create and define polyhedra, schedules, and dependencies. You may notice that the transformation schedule in that figure is very simple-looking, to the extent it could be brute-forced generated by a very simple compiler, but these transformations are not guaranteed to always be semantically correct (e.g., produce a program which will always give the “correct” results), and the number of possible transformation matrices exponentially explodes in size as the program size increases.
I’m not really a big math person so a lot of this stuff went way over my head, but it was very interesting to see it from a layman’s perspective. I’m a big fan of linear algebra, calculus, and other subjects from a conceptual standpoint, and I’m great at visualizing representations of these concepts, but math papers tend to get super abstract with the stuff they try and solve. For example, a turing machine can either be formally defined as a weird 7-valued tuple, with each value being either a vector, set, or function, and the relationships between these variables being explicitly defined using only set theory, or you could imagine a turing machine graphically as a “controller” sliding across a “memory tape”, changing values to other values depending on a list of instructions within the controller. Both representations mean the same thing, but the math version is provably correct while the other version is (in my opinion) easier to visualize.
- L. Pouchet, C. Bastoul, A. Cohen, and N. Vasilache, “Iterative optimization in the polyhedral model: part 1, one-dimensional time,” in International Symposium on Code Generation and Optimization (CGO), March 2007, pp. 144-156.
- L. Pouchet, C. Bastoul, A. Cohen, and J. Cavazos, “Iterative optimization in the polyhedral model: part II, multidimensional time,” in ACM SIGPLAN Notices, vol. 43, no. 6, June 2008, pp. 90-100.
- S. Parsa and M. Hamzei, “Locality-conscious nested loops parallelization,” in Electronics and Telecommunications Research Institute (ETRI) Journal, vol. 36, no. 1, February 2014, pp. 124-133.
The Owl House
Alright, enough with the nerd stuff! This section talks about a few of my thoughts on The Owl House, which aired its finale last Saturday on 4/8/23. This section contains spoilers for The Owl House. If you want to avoid spoilers, skip this section!
To start, I want to say that The Owl House is my favorite show ever created. It’s also the only show I’ve felt the need to rewatch from start to end. I’ve seen the entire thing 6 times now, and every time I see it I’m glued to the screen. Everything about the show is wonderful to me, and its been a huge influence on my personality, identity, confidence, gender, and more.
Initially, I was very worried in the days leading up to the finale. I rewatched the entire show in the two weeks leading up to the date, and I remember crying to myself one night at the thought of the characters being separated from each other or important characters dying in the ending, since a lot of the promotional material for the show said things like “good things don’t last forever” and the showrunner Dana Terrace often talked about liking bad endings more than good ones. I tend to get very invested (obsessive?) over fiction, and I’ve realized I tend to treat these characters as real people and the conflict of the story as real events influencing the lives and futures of these characters.
The idea of Luz needing to stay in the human realm while all her friends went back home to the demon realm made me feel very sad; nevermind the fact that the show made it explicitly clear throughout the entirety of the story that it would be completely against the theme to have the characters split up at the end (they say the line “us weirdos have to stick together” maybe five separate times throughout its 43 episode run). The show expresses that Luz did not have a community in the human realm, so she built her own community of like-minded people. Initially, it seemed wrong of Luz to go off on this “escapist fantasy adventure” in a different world, abandoning her responsibilities to her mother by leaving, but season 3 showed how much Luz longed for a place to be understood around people who she felt she could relate to.
Anyways, the finale itself was absolutely wonderful. I’m a huge sucker for endings that go “and here’s what the characters look like 3 years from now, what they’re doing with their lives and the kind of new adventures they’re going on, and by the way they’re all still friends and visit each other daily and they’re all happy and healthy and very successful and consider each other as family.” As mentioned before, I like to imagine these worlds as real things happening to real people, and seeing that everything works out for everybody (e.g., its not just an abstract theme the authors are trying to convey, but a glimpse into the lives of fully-realized individuals) makes me feel very happy.
I originally watched a livestream of the finale, but halfway through the stream buffered and I didn’t want to miss anything so I decided to stop and come back later when a recorded version was uploaded. It felt very strange waiting those 90 minutes between watches, where I watched up until the pivotal climax of the story and then took a break to wonder about my own thoughts, but my viewing experience was not damaged in the end. I grew a much stronger disdain for streaming, and don’t think I’ll ever stream again for anything important anymore, but I rewatched the finale in its entirety after that break and got through it alright.
I cried an embarrassingly large amount during the finale. That Saturday I spent pretty much the entire day watching season 2. I had a plan in mind to see several episodes a day for two weeks, but my schedule didn’t really align at times so I had to sort of crunch to make the finale in time, but the ending really was perfect in every way. It connected all the themes I wanted it to, tied up every loose end I wanted tied up, leaved some mystery and confusion in places I wanted to remain mysterious, and made references to past events in the best possible ways. Going into season 3, I was originally very upset that the show didn’t get a “proper ending” and that aspects of the show would be very rushed, but after watching the finale I realized how perfect the creators were in crafting the story. Nothing felt rushed in the slightest; of course I would’ve been thrilled to have an extra 20 episodes of content, but nothing about the ending was short of exceptional.
The content of the finale was breathtaking. The Titan was finally shown as a character, as someone whose been influencing the main characters’ journey from the very beginning. Luz and the Titan’s brief conversation about Belos’s being “evil” and how that was different from what they were doing was especially impactful for me. Of course, Belos wanted to do a genocide against all the witches in the Boiling Isles, but out of all the people to want to do this, a colonizer from the 1600s who lived through actual witch trials would be the most likely person to think it as rational. Why was Belos, a person who believed his actions were pure and morally correct, different than Luz, a person who also believed her actions were pure and morally correct? How could you justifiably define one as “evil” despite them both fighting for the same core ideals? It mirrors the current political landscape very closely, and gives a very good lesson to viewers who want to explain how a bigot can justifiably be “evil”, as it revolves around intent like the Titan said. The next paragraph gives some of my ideas on the topic of bigotry through transphobia, before returning to more Owl House talk.
What reason does a transphobe have to be transphobic? Admittedly, a significant amount of regular transphobes must believe what they think is morally correct; if they don’t, then they must be a sadist, deriving pleasure from inflicting pain on other people without regard for morals. Whether or not you believe sadism makes you unconditionally evil is out of scope, so I’ll focus on the first camp of people. Transphobes at least have some personal belief that what they’re doing is correct, since if they didn’t they would make steps to change their ideals. Some people may not be transphobic but still take actions that are transphobic in some cases, like a parent who doesn’t want their child to be transgender due to the potential increase in hardship they may experience as a result due to the bigotry of other people (of course this example makes a lot of assumptions on the ignorance and emotional intelligence of the parents, as dysphoria and other factors could still lead a closeted trans child to feel more unhappy than they would if they transitioned and experienced discrimination in their lives, but the point is that parents who feel no personal transphobia towards other people would still take transphobic actions in some cases for reasons they personally feel are justifiable). Furthermore, many transphobes do not have good reasons for being transphobic (like the idea that “if we normalize transphobia then that’ll be the death of our culture”... what exactly does that mean, and how will it affect me?), but these people may maintain their ideas out of conscious ignorance or being misled by other people. After all, many, many arguments bigots make can be proven incorrect, but that requires people who are willing to listen and learn. Are those people who choose not to listen evil? What if they’ve dedicated their lives to their bigoted views; bringing this back to The Owl House, the main villain killed his brother over his bigotry. Further down the line, if his belief in his views were challenged, he would have a very tough time changing his mind because it would mean he killed his brother for nothing. Nevermind the good a lifelong bigot could bring about by changing their mind! A transphobe who dedicated their life to hurting other people may think they can’t change their mind or else it would mean they hurt those people for no good reason! It is very difficult for any person to confront that idea, but the show ultimately argues that an “evil” person deliberately chooses to ignore strong arguments that debunk their views. Some of the other denizens of the Boiling Isles who were misled by Belos are not evil because they were willing to change their mind when presented strong evidence against their ideas. This is a very powerful message I think has influenced my own perception on the political landscape: your racist, transphobic grandparents have had plenty of opportunities to do a simple google search on the stories of real people affected by their bigotry. They choose to ignore the evidence that suggests they are wrong, and maintain the idea that the existing sources they consume are always devoid of error. It is very easy to be uncritical of the information you receive, to trust that other people put in the work of verifying and proving data so you don’t have to, but The Owl House says that an evil person is one who rests on their laurels despite the good they could cause (or the bad they could prevent) had they just explored different ideas. Regardless, I hope this little tangent was coherent, but I’ll return to talking about the show now.
The way the show ended made me feel was an emotion I’ve never really felt before with a show. For instance, I loved She-Ra and the Princesses of Power, but I’ve never felt the need to rewatch it as the ending communicated what it needed to and the conflict resolved itself nicely. When I was watching She-Ra in real-time, I wasn’t really hit with any significant theming that really resonated with me personally, so when I finished it I was happy I saw the show and moved on with my life. Before finishing The Owl House, I thought I would be devastated when the show ended, but I was met with such a unique overwhelming feeling of content afterwards. Of course I cried a lot during the finale and for a few minutes afterwards, but I felt nothing but “content” in the remaining hour I was awake; a peaceful, satisfied feeling of happiness over the show I liked so much. I knew it was over, but it ended so wonderfully that it didn’t leave me wanting for any more. It was a beautiful present I could open whenever I wanted to re-experience the emotions it gave me the first time I watched it, and I could wrap up that present and put it back on my shelf to unwrap again any time in the future. The present would never degrade, it never needed to be maintained, it always was pristine on my shelf, and I could always unwrap it and experience it again whenever I wanted. Throughout it all, the show felt like it was made just for me and nobody else; the creators hand-tailored this present for me alone, and the fact that it ended made me feel so extraordinary and peaceful to the extent I don’t think anything else in the world could match.
All in all, I’ll probably talk a lot more about The Owl House in the future. I’m rewatching the show again, and I don’t know if I’ll ever get tired of thinking about it. I started watching it after the Grom episode, when I heard a Disney Channel show about magic featured gay characters. I was right off the cusp of finishing Steven Universe, so I was excited to see more content that tangentially related to me. I caught up to the show fairly quickly after that point, and I remember frequently refreshing the website I watched it on the day a new episode released. Initially, the show had a big effect on me because it motivated me to make more friends and to leave my comfort zone; I don’t have all that many friends (see the aforementioned “I’m very shy” paragraphs above), but the show made me feel it was possible (and even easy!) if I just put in the effort. Around the same time, I started visiting a club in my school, and I’ve been attending the same club weekly since then.
Another huge effect of the show on me was the idea that I could do things that were “weird” and not feel bad about it. In 2021, I started calling myself autistic because I read the experiences of other autistic people and saw a lot of correlation to myself. I’ve never gotten a diagnosis or anything (and I don’t know if I ever will), but it made me feel better about myself and my personal limitations by justifying it with a label. For this reason, there’s always a chance I could not be autistic, but the presence of the label made me feel secure and happier regardless. Anyways, growing up there were moments where I was interested in doing things that I chose not to do, due to it being “too weird”. Reading fanfiction was one of these things: it made me uncomfortable to think about reading some person’s 2000-word story of two cartoon characters going on a date, because I always thought “it would be too weird of me, a very normal person, to indulge in this activity,” even though I definitely did want to read it. The Owl House communicated to me that it is completely fine to have these kinds of interests. A core idea of Luz’s journey is that everybody is weird in some way, and that on a macro-scale, your personal weirdness sort of averages out with everybody else. The Boiling Isles is a crazy place with strange people who all think they’re normal, but when viewed by an outsider like Luz, they’re all incredible and unique and weird. After watching the show, I convinced myself that things I wanted to do were worth doing, and I silenced the thoughts that came about from wanting to do things that were “too weird”. Incidentally, this includes reading fanfiction, making mashups of the one Mr. Beast Rap Battle and Splatoon songs (see below), and barking/meowing.
I think that’s everything I want to talk about today. I couldn’t possibly enunciate everything the show means to me, so I’ll save that for another time. I haven’t even dipped into any specific interactions between characters, or individual voice lines said that mean a lot to me! To give a precursor to whats to come: when I was rewatching the show in the two weeks leading up to the finale, I kept a little notepad next to me and made a note of moments that I thought were funny or that resonated with me in some way, and I’ve counted more than 100 moments that I’ve listed from one-line jokes to scenes I think about on a daily basis. I’ll be bringing lots of this in the future (each in their own little section like how it was organized on this page, so feel free to skip it if you want!). I wish it were possible to fully put my emotions about this show into words, but I don’t think thats possible. I just really, really, really like it.
Conclusion
Thanks for reading my awesome post. The next big thing on my plate is a huge project I have to complete for a class I’m taking due this week. It’s related to training a deep learning model, which famously is something that requires lots of time to execute without me actively doing anything, so I hope it finishes in time. I sort of procrastinated a lot on it. Also, I’ll still be rewatching The Owl House. I’ve been reading a fanfiction lately that I’m just about caught up with and am looking for more. Finally, if you want to contact me about anything, you can find me on:
- Discord: URLslug#6738
- Twitter: @URLslug
-Sophie