One More Epicycle

A few weeks ago, Carl Murray gave a public talk on the history of the Cassini mission. I thought it would be a good idea to go, not only because he’s a good speaker, but because I was giving a lecture on ring dynamics the next day and might pick up some helpful analogies.

During question period, one guy particularly annoyed me. First he introduced himself (Someone So-and-So) and explained that he was the chair of the local society of amateur astronomers. Then he offered Carl some fawning praise, and said that he hoped that no one would mind if he digressed into politics.. at which point he explained how much of a waste he thought the London Olympics would be and how much better use the money could have been put to by investing in several more Cassini-like missions. He asked Carl’s opinions of this idea, and his response — the correct one, of course — was “No comment.”

I thought of asking the next question myself: “I’m new to this country, so please excuse my ignorance of the local dialect. Is there a word for self-important jerks who hijack public discussions because the world desperately needs to know more about their pet causes even in an unrelated forum?”

This guy scored the trey. He introduced himself, to make it about him. He offered inappropriately-timed, awkwardly ingratiating praise (that it was accurate isn’t relevant), embarrassing the speaker. And then he used the “question” as a soapbox to rant. I can’t speak for anyone else, but it certainly made me uncomfortable.. and I’ve long since thought that the Olympics have grown too large and could do with being cut down by a factor of ten.

No one was going to be convinced by his “question”, and he must’ve known it, so it wasn’t about actually effecting any change. It was about the performance. He reminds me of someone I used to know distantly at Queen’s who was infamous across several departments for the same sorts of things.

Similarly, at Mass last night, an old woman shouted something — no quote for you! — in response to a very unobjectionable comment by the priest during the homily. (That only God could judge the heart; the reading was the story of the woman taken in adultery: “Go and sin no more”.) Did she expect the priest to start a conversation with her in the middle of the service in an enormous cathedral with hundreds in attendance?

And on the way to the church, there was a guy on the Underground who thought that headphones were beneath him, and instead let his music blare out of the player so that we could all be entertained by it. And a fine choice it was too, with the shouting and bad guitar and the drum line which would have shamed the autoplay features on a twenty-dollar keyboard.

Today’s peeve: people who can’t respect social boundaries involving speech and noise. Thank Heaven we’ve finally invented blogs, so that everyone in the world can publish all of their silly thoughts to the rest of the human race, and none but the willing need read them.

Yesterday turned out to be surprisingly productive: by the end of the day I was up by two thousand lines of code or so.

.. oh, all right, if you’re insisting on total disclosure. Admittedly some of that was duplicated code which I hope to get rid of eventually. Generally speaking, code duplication is just a subtle bug waiting to happen, but C’s a pretty weak language as far as inbuilt data structures and methods go.. and being both strongly- and statically-typed, it “sets” pretty quickly in the sense that it dries into something hard and difficult to modify. So if you’re practicing bottom-up development, there can be advantages to some short-term duplication when you haven’t figured out what the right level of abstraction is. I do hope to get rid of a lot of the code from yesterday soon enough, but for the moment it works nicely.

(How’s that? Too obvious for programmers and too vague for nonprogrammers. We’ve achieved zero-knowledge transfer, Doctor! Prepare to reverse the polarity of the neutron flow!)

Apart from my main multiscale code, I’ve also made progress on some other fronts. The regime of planet formation I study is known as “oligarchy”, where the system divides roughly into two: a small number of large protoplanetary embryos which control the dynamics, and a large number of much smaller rocky planetesimals in which the embryos are embedded. Think of a much denser asteroid belt (looking more like Saturn’s rings) with a bunch of large moons flying around.

This is understood reasonably well at low masses, which means that I can use semianalytic models to fast-forward through this early phase and start my numerical simulations much later when there are far fewer objects I need to track. In my thesis, I used a modified version of a model due to Ed Thommes, Martin’s grad student before me. More recently, John Chambers has developed a far more sophisticated model and I’ve been implementing it from his description in the paper.. unfortunately, there are a few missing pieces in his explanation which make it hard to know exactly what numbers he’s chosen to put in certain equations. It looks like I’m getting the shape of the curve right but there’s an offset, so it’s tough to know if there’s a problem or not.. I’ll have to compare with his previous papers.

I managed to get quite a lot of the Chambers model running yesterday, though, and I think that has a lot to do with the fact I wrote the Chambers code in python and not in C. I’m roughly a factor of five-to-ten more productive in python, although that’s a pretty handwavy comparison because there are problems which are trivial in python but would take forever to code in C, and so I’m not sure how long it would actually take..

I’m a big believer in the use of VHLL (very high level languages) for programming in general and scientific programming in particular. As a rule, if you find yourself needing to write in C or fortran it’s usually a sign you don’t understand your problem well enough, and you’re wasting large amounts of time you should be spending at the pub.

Like most rules, of course, this one has its exceptions.. such as high-performance computation where the task doesn’t break nicely into discrete compute chunks it’s easy to optimize. Computations dominated by FFTs or eigenvalue finding, for example, can often be handled nicely even in VHLLs because 99% of the time is spent in routines which can be pushed out of the interpreter into tuned low-level libraries, and your VHLL code is basically scripting. Computations where the work is more spread out, where you have to take a lot of loop steps, are far more challenging.

Guess which kind of simulations I do?

Just my luck to work in a subject (HPC studies of planet formation) where I have to break my own policies to accomplish things. I look forward to the day — when PyPy is complete, maybe? — when I’ll never have to write in as frustrating a language as C again. Maybe if CASCA ever lets me rant during the education section I can help save the next generation from the brainrot that use of C and fortran causes.

Preparations for tomorrow’s lecture continue, I decided that a crazy idea which has been lurking at the back of my mind for the last few years might actually work, and I’ve started on the enormous backlog of emails which somehow built up. Frankly I’m astonished anyone still bothers to write, but it’s appreciated!

Rumours of my death have been greatly exaggerated.

(That’s one of my favourite lines ever.  I’ve considered spreading rumours of my death just to have a chance to use it.)

The last few weeks have been unexpectedly hectic, which explains the lack of posting.  Some stuff — like work — has been going very well.  Code is progressing nicely, new nodes have arrived for our simulations, etc, etc.  Other stuff — like, well, life — has been less enjoyable.

To give you a taste of the flavour: now that I’m no longer a student, it’s time to start repaying my student loans.  They don’t amount to much, and frankly I could pay them all down right now with what I have in the account.. but that would blow away my rainy-day fund (should I need to leave the country in a hurry and change my name to Miguel Sanchez).  And it rains a lot here..

Anyhow, the wonderful folks at CIBC drove me crazy.  They took to calling my parents, which I use as my secondary number because of its permanence, at 6:00 in the bloody morning to get them to get me to call.. even though I already had, and had cleared everything up.  When I called to complain, they had no record of the call!  I explained, patiently at first, and then less so, that since this was the number they’d left they should bloody well be able to explain to me why they called at six A.M.

Finally it turned out that they have a semi-automated “COURTESY CALL”, whose database wasn’t correctly getting the right information for some reason.  “You keep using this word.  I do not think it means what you think it means.”  But they thought they’d sorted things out. I was complaining about this whole saga to a friend of mine (the one who keeps losing his mittens) when I got an email from Mom saying that she had just had ANOTHER call, making six or seven in total.. so I called CIBC again to yell at them, determined that I wouldn’t get off the line until it was dealt with, whoever I had to speak to.

(While I was on hold, they played “Baby, it ain’t over till it’s over.”  That was not the appropriate lyric given the situation.  I’d quote the right one but this is a family-friendly blog.)

Finally it turned out that they wouldn’t let me speak directly with the guy in charge of calling me — I’m unimpressed by his cowardice even as I have to respect his instinct for self-preservation — but insisted that things were now fixed.  Why?  Well, because apparently now the calling dep’t had a note saying they’d contacted me.  Which of course they hadn’t done, even though they have my London number.

No, what had happened is that my easygoing mother had told them in no uncertain terms that they’d screwed up and apparently this was enough.  So I speak to them a half-dozen times and this is insufficient to get them to toggle the “spoken-to-Doug” flag in my file.  On the other hand, they speak to someone who isn’t me at an address I haven’t lived at in a decade and a half in another country, and that apparently counts as a successful contact.

There’s a reason I switched from CIBC all those years ago, and now I remember why.

(“So let me get this straight.  You’re calling my elderly parents at six o’clock in the morning every day despite the fact that I’ve contacted you many times and everything is fine with my file.  Moreover, you have no record of the calls, although you think it’s because I’m on an autocontact list and they’re screwing up the timeshift. You can’t stop the calls, because it’s not your department making them, and you can’t put me in contact with the people who are making the calls because you’re the only group authorized to speak with the public.  So pretty much you’re going to keep calling until you feel like not calling any more, and there’s nothing I can do short of having the phone company forbid calls from CIBC to that number to get you to stop.  This is insane.”

“I can see why you’re frustrated..”)

Why not just turn off the ringer before bedtime as an interim solution?

Well, that takes us to the second unfun saga of late, involving my father, who’s had several operations over the past few months.  When they did the second surgery to drain some fluid which had accumulated after the first surgery on his back, the doctors were astonished that he was still walking around: the pain should have completely incapacitated him.

Unfortunately, or maybe not, after decades of continual anguish his pain tolerance has reached inhuman levels.  This isn’t actually a superpower, but a superweakness, because it means that he doesn’t get the information from the fact that something hurts which you’d normally expect.  A few years ago, I broke a bone in my right finger during football practice — and went on to score a couple of touchdowns afterwards, I’ll have you know! — and didn’t think it was broken because it didn’t hurt much the next day, so it took a month or two before I bothered to go to the doctor.  This is pretty much what it’s like for Dad all the time.

Pain has its purpose: it’s a message.  By studying patients with Hansen’s disease (leprosy), it’s been established that much of the tissue damage isn’t directly due to the Hansen’s but due to the resulting insensitivity to pain, which means the patients are continually harming themselves unknowingly..  (See the work of Dr Paul Brand.  Or that recent House episode about the girl with a similar condition.)

Well, even Dad finally decided to check himself into the hospital, at which point the doctors decided to remove his gall bladder.  Remember that car accident I mentioned months back?  It might have ruptured it — or not, who knows, maybe it was more recent.  In any event they decided it had to go.

Or that’s what they thought.  It turned out when they opened him up that he didn’t have a gall bladder any more.  It had burst and gone completely septic and gangrenous.  Ordinarily that spreads and kills you, but his body had formed this strange cocoon to keep it isolated, and completely encased.  The doctors had never seen anything quite like it, although I gather that similar things are known to happen.  Not unheard of, just infrequent: the sort of thing you might bump into once in a career.  They had great difficulty believing that he wasn’t already dead, much less that he’d been up and about.

(My father, the zombie.  Huh.)

He’s now home, and recovering.  For many years I was convinced that he was just too stubborn-mean to die, whatever the laws of biology would say about it, but over the last few years he’s definitely been mellowing out.. which worries me.  If he’s lost his anger, he’s lost his fuel.

So if you’re reading this, Dad, go yell at the doctors for not catching the fact they were operating on a half-dead man over the last few surgeries or something.  We have a heritage to live up to: “conquer or die“.

On a happier note, my friend planetary atmospheres guy James Cho and his wife now have their first daughter.. and while he’s on leave I’m covering his lectures, which has also been a serious timesink.. especially because I don’t know as much ring dynamics as I ought to.  This week I’m covering planet formation, though, so that should go much better.

Anyway, in every direction there have been antiblogging distractions.  Fortunately things look to have settled down somewhat now, and I should return to my usual off-centre snarking.

I also have a confession to make, and no, it’s not that I broke my no-beer-during-Lent rule for St Patrick’s day, as tempting as that was.. but you’ll have to wait for that.

Two weeks or so ago, during a meeting with Richard, I moved to the whiteboard to sketch out a new scheme I had in mind.  As I was about to start writing, he stopped me by pointing out that the marker I was holding was black-coloured, which meant it was permanent.

Then I realized that the whole board was covered in green permanent marker, most of which was my fault.

There are lots of ways to get rid of permanent marker (and nonpermanent that was left too long) but the one I decided I liked best was simply writing over the stuff with nonpermanent marker.  It’s easy — it just dissolves away underneath it — but it doesn’t scale well.   If you write a word and then notice that you’ve used the wrong marker, it’s quick to use the right marker as an eraser.  However, cleaning an entire board with it is awfully inefficient..

So of course that’s what I did, because all the other methods involved shopping.  It probably took me three hours, all told: I’ve been doing it when I’m bored and I’ve just finished it.  Feel like I’ve actually accomplished something!

I think doing it the slow way was just an excuse to take breaks now and then from coding, and listen to K-Os’ last album, Atlantis: Hymns for Disco, which I liked.  Not quite as brilliant as Joyful Rebellion, but still excellent.

I can sense it already: it’s going to be another exciting Monday.  Today’s physics seminar is on complexity and criticality, and the pure math talk is on mutually orthogonal terraces, so at least I’ll have something to break up the tedium.