Life as a Physicist

Our kid does not like sleeping in her crib. But this is impressive. She can sleep all the way through a diaper change. But if you lower into the crib right after said diaper change she explodes, screaming her head off, the second any part of her touches the crib. The crib is her kryptonite!

I’ve made lots of comments on this blog that I think C++ is stagnating and all the interesting features to help programmers with their jobs are being added to languages like C#, java, python, ruby, etc. But C++ is approaching its next draft of the standard. (ISO C++0x, where x is “7″ or “8″). I read Herb Sutter’s blog, and he is on the ISO committee and he has a blog posting on what might be be included in the next version of C++.

Some of the things are fixes. For example, the horrid error messages one gets when a compilation error occurs during template instantiation. Another, mutli-threading libraries, addresses a basic portability problem. But neither of these really pushes the language into new ground. Both will help the programmer, of course!

The other thing that Herb talks about, however, is interesting: garbage collection. No worries! It sounds like an opt-in scheme. In DZERO we have had a lot of trouble with memory leaks. Heck, just look at ROOT (sorry, couldn’t resist). Even here, of course, C++ is playing catch up. But this is a welcome addition. As they point out in the transparencies linked to above, there are many reasons why you might want to have explicit memory management (low level system programming, memory hungry programs, performance critical programs). But most HEP algorithms for reconstruction do not fit this bill! I look forward to seeing if we can take advantage of this.

As a side note, I’m not sure ROOT will be able to benefit from this. This is because is makes use of void* pointers and also redefines the new operator. They say in several places in the transparencies that garbage collection as they want to specify it will not work under this circumstances. Oh well.

Things are a bit crazy around here — by the end of this week they should be a lot calmer. But yesterday I cleared out 2 hours to work on research — the first time since classes start up again here at UW. And guess what I spent the full 2 hours doing? Trying to find the work I wanted to work on! I never got to the actual work, nor did I find the old work that I’d left off!

One day this is going to get better. I have a sabbatical next year and I’m looking forward to that!

Scott left a comment on my blog which lead me to explore his blog (you know, the blog thing). He had short post on visualizing data — a passion of mine that I rarely have done anything about. But a comment in his blog led me to the Periodic Table of Visualization. Catalogs just about every type of chart, plot, or other weird visualization. Including a parameter ruler and an iceberg. Worth spending a few minutes poking around. I wonder how many of them would useful for HEP?

This reminds me of something I heard about a few months ago, called DynaVis - a display technology out of MS research. Now that my memory has been triggered, I was able to find a video demo of it (but no info on the MS research pages). I think the idea behind it is to take one set of data and represent it with several different plot types. The difference is that as shift from one plot type to the other is animated. So if you switch from a histogram to a pie chart, for example, the pie wedge shapes deform to histogram bars and fly out to a appear on an axis. The state goal is that this helps your eye track what happens to the data as you view it in different ways. While cool looking I’m having trouble wrapping my head around how that would be helpful in my work.

But, it did remind me of something else (ha!). A long time ago at a conference that I can’t remember (CHEP, possibly) I saw a presentation from a fellow running a company called Mind The Gap. Based in the UK, they took many UN population and country financial databases and put them together into very attractive plots that were animated. Instead of using time they would use something like GDP and you could see how various country’s performance varied as the plot evolved. Further, the plots were full of information: the dots size, position, and color all had significance and could all change. Now that plot I thought I might be able to use — but they never released their software as far as I can tell. And I can’t find them on the web any longer (I used to be able to find them).

Actually, there was an effort in D0 at one point using some sgi based software to build an event display and you could attach all sorts of properties to a jet or electron’s arrow (arrow head size could represent EMF, length could be the pT, etc.). That was back in the days it looked like sgi was going to take over all of HEP computing.

Then again, with HEP’s relentless push to use only PDF, and only the most basic versions of PDF, animation is disappearing from all of our talks. Not that animation is always a good thing, of course!