28 November 2010

How to do theoretical physics

I talked to a few of my colleagues about what to do when you are stuck. Here is a compilation of their suggestions.

Every theoretical physics PhD student reaches a point where mindless cranking and/or doing what their advisor tells them to do doesn't work anymore. This is a hump that the student must learn to get over. Typically the advisor doesn't know how to solve the problem either. If the advisor sat down for a week and thought about it, he/she could probably solve it, but of course, that's not what happens. Either the student solves it or the project is abandoned. Smetimes the student shows the advisor why the project is too hard.

If you are stuck:
  • Read the literature. Work out all the equations in the papers, and keep your work in a notebook.
  • Talk to lots of people. Be aggressive. It is especially helpful to talk to other students because they are at your level.
  • Figure out why you are stuck. If you quit the project, as least you know why. Sometimes you go around in circles because you don't even know why you are stuck.
Sometimes, there's no problem to solve, because it's been solved already and only a very serious publication search recovers it. Sometimes, it's very hard to find the paper where your problem was solved because many years ago, people used different language. One way to avoid this is to find some people who are experts on the type of problem you're working on, and ask them. This could be tricky, because you don't want them to steal it from you, but if you do it in an advanced stage it would probably won't happen.

A good theoretical physics PhD advisor sees his/her students multiple times during the week. Some advisors likes to pop into their student/postdoc's office and provide "moral support". If your advisor doesn't have time, it's a good idea to email him/her to talk about the project between meetings.

You shouldn't be spending all your time hammering away at one problem. Classes are over, so you should spend time talking to other people and learning about other things. There is nothing to feel guilty about. This is why you are in academia!

For the PhD student, there should be a balance of what they are good at and not good at. If a student is naturally good at numerics, he/she should do some analytical work and vice versa if the student is good at analytical work.

Remember: theoretical physics is hard!! The purpose of theoretical physics PhD is to prove that you can do theory research. Unfortunately, not everyone can pass this "test."

27 November 2010

Essential skills for a scientist

I was having a conversation with someone who wasn't sure what he wanted to study in physics graduate school. He was worried that if he chose one field, it would be very hard to make progress, but it would be important work, whereas if he chose another field, it would be easier to make fast progress, but the work might not be so fundamental.

I wasn't sure what to say, so I advised him to take a different perspective:

"Another way you could make a choice is to think about what skills you want to acquire. You want to have the skills you need when the exciting development comes along, so that you can jump into it right away."

Wolfgang Ketterle, an atomic physicist and 2001 Nobel laureate, himself said that skills are important.
Changing fields was a crucial experience for me. Amazed to see how much of what I had learnt before could be applied within the new field, I realized that general skills are much more important than specific knowledge. I thought it would take a long time before I became productive in my new environment, but within months, graduate students who had been working in this area for much longer came and sought my advice and leadership. This experience gave me the self-confidence to venture into new areas, and provided the impetus for my later decision to come to the United States and start once again in a new field.
(Quoted from the Nobel Prize website autobiography)

The question is, what are the important skills for a scientist to learn? I can think of a few:
  • Asking important questions and picking good research problems, i.e. ones that are solvable and interesting.
  • Understanding the big picture and being able to pick out what is essential and interesting from a mess of details.
  • Turning a relatively vague and abstract research question into a concrete calculation or an observable quantity that can be measured in an experiment.
  • Writing (papers and grants)
  • Speaking
  • Managing and training (teaching) students
  • Collaboration skills
  • Life balance and project management
  • Programming skills
  • Math skills
In addition, for an experimentalist in physical science, it is also important to know some signal processing. (My sister contributed a few items on the list.)

26 November 2010

Arcade gaming at home

I was playing Metal Slug on my Nintendo DS Lite the other day and my hands were dying from so much button pressing. I realized that what I really need is an arcade style joystick. Arcade style games aren't really made anymore, so I could install MAME emulation software on my PC to play retro games like 1942, Gauntlet, Street Fighter II, and Samurai Showdown. A good resource for anything related to building your own MAME arcade box is MAMEWorld. I found some cool stuff like playing Mahjong in MAME and even light guns for your Time Crisis fix!

I had no idea there are so much information about different types of joysticks and buttons. slagcoin has an extensive guide here. Apparently, there are three major joystick types: Happ (US), Sanwa (Japan), and Seimitsu (Japan). It's much easier to get high quality arcade parts from Japan because the arcade games are still popular there. Americans don't really play arcade games anymore. Many people buy parts and mod their own custom joysticks. That sounds like fun, but I'd like to play with some of the off-the-shelf stuff first so I can get a feel for what I like. Eventually, I'd like to do a complete arcade box build project as described in MaximumPC's article "How to Build a Kick-Ass MAME Arcade Cabinet from an Old PC."

There are few recommended off-the-shelf manufacturers.
  • Hori is the biggest game controller manufacturer and has a high quality Real Arcade Pro line.
  • MadCatz makes a Street Fighter IV FightStick Tournament Edition for PS3 and XBox 360.
  • X-Arcade makes US style arcade joysticks. They are lower quality than Hori or MadCatz but they are easy to mod and the PC board is supposedly 100% compatible with MAME.
After some cursory research, these joysticks seem good for a MAME setup:
  • Hori Real Arcade Pro VX-SA for Xbox 360. A top-of-the-line model which uses all real arcade parts and layout.
  • Hori Real Arcade Pro EX-SE for XBox 360. Another top-of-the-line model which uses all real arcade parts. It's unusual in that it uses Seimitsu parts, rather than Sanwa parts. All the other Hori and MadCatz joysticks are Sanwa.
  • X-Arcade Solo Joystick. A decent joystick but kind of pricey for the parts. I would get this one for doing some modding.
There are Hori sticks that are 1/3 cheaper, but they don't use authentic arcade buttons and the whole point was not to do any modding. I looked into MadCatz joysticks, but there wasn't much feedback on using them successfully for MAME emulation.

06 November 2010

Quote of the day: Getting back to who you were

In the episode "False Labor" from the TV series Caprica, Daniel Graystone has a conversation with Joseph Adama.
Daniel: It's like every decision I've made since the bombing has been a wrong turn. By now I've made so many of them I'm not sure I even know how to get back.

Joseph: Get back to where?

Daniel: To who I was, or who I thought I was.

Joseph: You can't do that alone. You need somebody. You need someone who can see the man you were inside the man you've become and forgive you.