About bkpark

I am a graduate student at UC Berkeley.

Anti-antispam-filter filter

I use Gmail (Even my a...@bkpark.com is actually a Gmail account, through the free (well, set up when it was free and then grandfathered in) Google Apps), and my recent complaint has been that its spam filter got really aggressive—as if the people working on Gmail’s spam filter didn’t realize that false positive is a deadly thing in a spam filter. I would tolerate 100 false negatives (i.e. spam that gets through the filter) before I accept a single false positive in a spam filter.

This didn’t used to be a big problem; I check my spam folder occasionally, and all the other email accounts I have (mostly from various systems at UC Berkeley) that forward to my main account at a...@bkpark.com didn’t have such an aggressive spam filter. It became a problem when I migrated my CalMail account to bMail, Google-hosted email for UC Berkeley.

I realized this when a student claimed—even against (unfair) accusation of lying—that he sent an email that I never received. When I finally thought to check the spam folder on my abkp...@berkeley.edu account, there I found his email, along with a couple other emails from students and 100 or so actual spam. This was an untenable situation; I was never going to check this spam folder, because I never check my abkp...@berkeley.edu account; that’s why I had my mail forwarded. I couldn’t tolerate a single false positive—and when I thought about it, this must’ve been responsible for one incident last semester when the instructor said he emailed me a draft of the final for my comments that I never received.

Apparently Gmail spam filter cannot be deactivated. You can, however, put in filter rules that effectively de-activates the spam filter. When you create a filter (under Settings; you might have done this to automatically apply labels, etc.), there is a checkmark for “Never send it to Spam”. So all you have to do is create a rule that is virtually guaranteed to match all incoming email, and check the option never to send it to Spam.

There are a couple ways to do this: One is to match for the at-sign (@) in the “To” field; the other is to match for a word that you are sure that legitimate email would never have in the “Doesn’t have” field (I personaly use عَرَبِي/عَرَبِى).

Once you set up such a rule, you will find a nice (for me, anyway) side effect—apparently the filter rules apply to outgoing emails, and what Gmail means by “Never send it to Spam” is “Always send it to Inbox”—you will find your outgoing emails in your inbox, in addition to the “Sent” folder. For someone like me, who compulsively labels all email (in fact, my email backup system is set up to back up only email that’s been labeled), this is a nice side effect—it reminds me to label the sent email before archiving them.

P.S. It goes without saying, but this is a kludge—and it may not continue to work in the future (hopefully in that future, Google gave us an option to turn off the spam filter …).

Two-Lane System

I like driving on the I-5—and it’s not just because I end up liking the things I do often enough (I’ve been visiting my parents more often since I, ah hem, learned to drive long distance, rather than flying; ‘turns out the hours saved in flying between Norcal and Socal are minimal, and I’ve recently driven on the I-5, attending the nEDM collaboration meeting). It’s because, well, because I like analyzing the two-lane system that is the I-5 (for most of the way).

It’s a simple system with just enough complexity to make it interesting—while making the problems tractable (I believe it was Prof. Commins who pointed out that 3 Nobel prizes have been awarded for works on two-level systems). There is an interesting dynamic there. Here’s a couple:

Generally, there are at least 3 different types of cars and drivers on the I-5: the trucks (which go at the speed limit of 55 mph but usually around 60 to 65 mph), the … lawful drivers (who either drive at the speed limit of 70 mph or just above), and the speedsters (go at 100 mph, that’s usually the top of the range most of the day). But there are only 2 lanes, with the right lane being occupied by trucks. Because of this, the following dynamic develops. The careful, lawful drivers tend to take the left lane and stay there—so those who drive (and want to drive) fast are more often found on the right lane, trying to pass the cars on the left in-between the trucks. Never mind the driver’s ed class that teaches if you are getting passed from the right, you should change lane to the right!

Sometimes, there are constructions, and the two lanes merge into one lane for a bit. Before the point of merge, do you know which lane moves faster? It’s the right lane, the lane with trucks. I’ve hit this traffic jam a couple times last year while visiting my parents, and I had it reasoned out this way: at the point of merger, the cars in two lanes alternate in merging into one lane—after all, that’s the courteous thing to do. However, while the truck gets counted as one car in that accounting, physically, they are much longer than other cars, so, right lane simply ends up moving more meters per merger—sometimes I wonder why the cars in the left lane stay in the left lane.

Anyways. I admit I’m a bit of an aggressive—and fast—driver (I’m a safe driver; no ticket—that I didn’t fight successfully in court—or accident in 10 years of driving). And I’ve found that in weaving between the traffics, the key is in figuring out the intent of the driver, as quickly as possible. I mean, all cars on the I-5 (that are not trucks) can go at 70 mph like the speed limit says, or at 80, 90, or even 100 mph. What limits the speed is how much the driver is willing to step on the pedal. Where this matters is this: you are in a row of 3 cars, just passing a truck on the right, and the car in front of you changes the lane to the right after passing the truck—do you follow that car, or do you stay in the left lane, which is the ironically slow lane? The problem here is that there are two equally valid reason for the car in front of you to switch the lane to the right—he perceived the cars behind him wanted to go faster and yielded at the earliest chance; or he thought that the cars in front of him were to slow and wanted to pass them.

Although this hasn’t been right 100% of the time, I’ve found that the cars that don’t accelerate immediately after switching lane belong to the former—and I’m better off staying in the left lane, at least until we pass that car, because a driver wanting to drive faster will start accelerating immediately when he has a clear space to accelerate.

Anyways. I don’t want to go on and on—but I just like driving on the I-5—although I think I’ve started getting the hang of the wider freeways with 4 or more lanes; one thing I’ve found when hitting a moderate traffic: switch to the right lane as necessary; never decelerate to a complete stop (unless you want to get stuck in the lane, unable to switch because the cars in the other lane are coming too fast).

Pickle?!

Huh. Apparently the whole object can be saved into a file in Python. This is going to make programming task so much easier; when it comes to data-saving, I now only have to worry about: (1) doing most of the data analysis in Python, the preferred method in the first place; (2) saving analyzed results into some sort of CSV file for further use by plotting programs, etc.

When it comes to needing rawdata later, I can simply reload the pickled object in Python.

Ringing in the New Year

Here are some things I’m looking forward to in the new year:

  • My qualifying exam: it’s coming up in less than 3 weeks, so I ought to be prepared for it—but I’m woefully unprepared as I haven’t even made the presentation yet.
  • Meaningful progress in the nEDM project: the cell-charging study subproject we’ve undertaken should come to some meaningful conclusion—regardless of other things going on with the nEDM apparatus. This should be a nice change of pace from the rest of the project (where, the rest of the project being integration into the entire apparatus, it’s quite dependent on everything else going on in the project).
  • Being the head GSI for 7C: I’ve been wanting to be a head GSI for a while (at least until I found out that head GSIs don’t … really teach unless they hold 30-hr position), and well, I get to be that this semester. I think I’ll end up spending less time than I did as summer 7C or H7C GSI—even if not, this should be a fun experience.

And, last but not least, I will continue to enjoy the success of my resolution to make no resolutions.

Update (Feb. 16): item 1 done; item 3 in good progress; item 2 just got harder, but if there’s one thing I like about myself, it’s that I get things done—no matter what.

Merry Christmas!

… and a happy network outage.

It’s been going on for last 15 minutes or so (that I noticed), but none of the group servers have been accessible over the net.

bkpark@gorgo:~$ nslookup budker.berkeley.edu
Server:         8.8.8.8
Address:        8.8.8.8#53

Non-authoritative answer:
budker.berkeley.edu     canonical name = birge-426-275-003-d.Physics.berkeley.edu.
Name:   birge-426-275-003-d.Physics.berkeley.edu
Address: 128.32.239.6

bkpark@gorgo:~$ nmap -sn 128.32.239.1-254

Starting Nmap 6.00 ( http://nmap.org ) at 2012-12-25 14:33 PST
Nmap scan report for g1-18.inr-220-tan.Berkeley.EDU (128.32.239.1)
Host is up (0.041s latency).
Nmap done: 254 IP addresses (1 host up) scanned in 38.03 seconds

Today being Christmas, this probably won’t get resolved soon—oh, well. I hope no one’s trying to access our group web page today.

Update Dec. 26: Apparently it’s a power outage. Birge Hall is without power (except for emergency power). No wonder IST hasn’t said anything about a network problem yet.

HV relay box

The HV power supplies that Spellman advertises as “polarity-reversible” turned out not to be (at least not in the sense any reasonable experimentalist would use the word “reversible”), and the model that they have offered up for exchange (the old CZE1000R models) after the fact turned out to be woefully inadequate as well. We already had the exact same models; never mind that they are advertised as “auto-reversible”, using the sense of “auto” that no experimentalist would use—so far as I can gather, what they mean by “reversible” is that polarity can be changed by opening up the power supply and messing with its insides (exchanging the voltage multiplier), and what they mean by “auto-reversible” is that polarity can be changed without that tremendous hassle (totally ignoring the Greek root of “auto”, meaning “self”, not “easily” or “by manual flip of a switch”).

Anyways. I’m not writing a blog post to complain. The upshot of all this is I have to build a HV relay box—one that does a couple additional things that no commercially available HV supply seems to do (in particular, disconnecting the HV supply from the electrodes; since this is particular to our experimental setup, I wouldn’t expect commercial supplies to have this feature). I spent nearly a week machining and building a box to hold all the mechanical HV relays that we found around the lab. It’s still a work in progress, but here’s a pic:
relay_box_in_progress
I still need to make all the electrical connections and build the control circuit; I do hope these things take less time than building the box itself.

BTW, based on the quotes I got for similar mechanical relays, apparently these Kilovac relays we found around the lab are … worth quite a bit. Let’s just say that if we broke them and had to replace them all, it would cost somewhere in the $10k in today’s dollars (since they apparently date from the 80s or 90s, they probably cost somewhat less back then).

Starting a new site

Apparently, UC Berkeley moving to the new bMail thing means everyone on bMail system emailing me gets to see my main email address (a...@bkpark.com) front and center, instead of the Berkeley email address that I usually give out (byun...@berkeley.edu).

So, well, since I can’t keep people—people who know me in my professional capacity—off of bkpark.com, I’m re-launching bkpark.com as the more … professionally oriented site. My personal sites are still available elsewhere, if you care to look.