There’s obviously an element of inferiority-complex here.  In a world where everyone is known by his google ID and it’s all but impossible to acquire a static IP address, I am in the extreme (maybe even six-sigma) minority to be administrating my own e-mail server.  I hold a physical copy of all my emails on my home server (and Abbie’s, too) and I never see ads.

But lately, particularly since moving from a DSL connection to a Cable connection (hosted by Cox Business), I’ve started to notice that emails from my domain were frequently spurned, often winding up in spam folders or being rejected altogether, particularly by Yahoo servers.

This is a question of net neutrality: Yahoo, being an email provider, has a business incentive to harrass small-time service providers such as myself, which, if our numbers suddenly multiplied, would pose a threat to its email revenue stream.  Of course, such a thing is extremely unlikely to happen, because people don’t even know how to keep their antivirus software up to date, much less operate a standards-compliant mail relay.  I take pride in this accomplishment, even though it’s somewhat akin to the accomplishment of writing an entire parametric, interactive feedback-control and data acquisition system in PIC ASM, which is something else I take pride in having done, even though it’s not something that very many people would mistake for being useful.

And so I take it personally when my logs report something like:

Feb 25 19:16:18 ocean postfix/smtp[98301]: 0F7E6B826:
host a.mx.mail.yahoo.com[67.195.168.31] refused to talk to me: 421 Message from (24.249.152.92) temporarily deferred - 4.16.50. Please refer to
http://help.yahoo.com/help/us/mail/defer/defer-06.html


I filed a help request with their abuse-admin service and received a canned response, from which I learned about DomainKeys.  The concept is quite elegant.  I modify my server’s DNS record to include a public encrpytion key, and store the private key secretly on my mail server.  I then stamp every message I send with a signature which is encoded with the private key.  The recipient can check my DNS record, use the public key to decode the signature, and verify that the message is legitimate.  DomainKeys were invented by Yahoo, but are now an accepted open standard.

Following this howto (and adapting to FreeBSD), I managed to convert my Postfix install to use domain keys in just under three hours (specifically, the hours from 11 PM to 2 AM on February 24-25), thus demonstrating that I am still qualified to operate my own mail server. It remains to be seen whether Yahoo will take its own medicine: at the time of writing, it’s still deferring my messages, even though they are now possessed of DomainKey signatures.

For what it’s worth, I never get false-positives in my Spam folder.  There is a single exception in the past year, and that was due to an embarrassing goof on the part of the SpamAssassin maintainers.

You can view my DKIM public key with the following command:

dig -t TXT 301south._domainkey.301south.net


if you use a computer with access to dig(1).

This can be corroborated with CO2 measurements from around the world and is really not contested.

However, some people seem to still believe that this change could not be caused by human activity.  Here I present a simple, back-of-the-envelope computation to measure the total weight of CO2 in the atmosphere, compared to the total weight of CO2 released by burning fossil fuels.

First, the easy part: the weight of CO2 in the atmosphere.  This page has a nice description of how to quickly estimate the weight of the air we breathe.  Basically, atmospheric pressure at sea level is a direct measurement of the weight of atmospheric air.  Multiply it by the surface area of the globe and you’ve got your answer (in Newtons).

The partial pressure goes with the mole fraction, so CO2 specifically accounts for 320 parts per million of that (in 1970) or 380 parts per million (in 2010).

for a net gain of about kg CO2 (also known as 300 billion tons).

Where could all of that carbon have come from?

(images link to source) These numbers have been converted to “tons of oil equivalent” for energy purposes, which is convenient for us because a ton of oil, when burned, releases a pretty reliable amount of carbon dioxide.  Coal is an inferior fuel: since its heat of combustion is much lower you have to burn almost twice as much to get the same amount of energy, which means you release twice as much carbon.  Since this chart is showing energy-equivalence and not CO2-equivalence, we can assume that it is a significant under-estimate for coal (slight over-estimate for natural gas).  But rather than tease that out, we’ll assume we had gotten all of the energy from oil.

Taking a quick-and-dirty average, I’m going to go ahead and say we’ve burned an average of 3000 million tons of oil, 2000 million oil-equivalent-tons of coal, and 1400 million oil-equivalent-tons of natural gas per year since 1970.  I’ll even be generous to the denialists and call them short tons, and round down, to get about 5,500 million metric tons of oil or kg.  Burning a barrel of oil (about 140 kg) releases about 430 kg of CO2 – that’s roughly 3.08 kg CO2/kg oil.  Let’s round down and call it 3.  Same as pi.

So we’re talking about kg of CO2, directly from anthropogenic sources, per year for the last 40 years.  This is just from burning fossil fuels, not from land use change (deforestation and development).  That’s over 600 billion tons in 40 years, more than twice the observed increase.  The difference between what we spew out and what sticks around in the atmosphere is, of course, getting soaked into the oceans, among other sinks, where it is converted to carbonic acid, lowering the pH of the sea with detrimental effects on sea life.

We clearly don’t fully understand the biogeochemical dynamics of carbon circulation, but it is abundantly clear that anthropogenic emissions are far more than sufficient to account for observed changes.  Face it: we are messing with the atmosphere on a grand scale.

In order to do conventional least-squares fitting to you need to determine .

Step response of NIH 3T3 fibroblasts to ionomycin

For instance, the figure above shows the fluorescent intensity of some 3T3 cells to ionomycin.   The traces are normalized to the initial values.  Notice that the cells don’t recover to the levels at which they began.  I wanted to know the time constant of the decay on the tail of the trace.

I solved this problem by performing the fit inside an iterative minimum-finding function.  First I identified the point of fastest decline and selected that as the point.  Then I iterated to find the value for which minimized the norm of the error:

for best-fit A and .

y0_fit = fminbnd( @(y0) ...
                 norm( y-y0 - exp( polyval( polyfit( x, log(y-y0), 1), x))) ,...
                 -10*y(1), y(1) );

This, along with other analysis:

For a population of cells (Showing mean and standard deviation for each test point. numbers in parentheses indicate number of cells in the sample.):

TeX source

Figure [above] shows how hazards, risks, and costs are considered together in an alternatives analysis.  A successful alternatives analysis should develop and evaluate several different approaches for solving the problem at hand.  The analysis should make conspicuous consideration of hazards, the sources of hazards, and opportunities to avoid hazards, thus ensuring a precautionary footing.  The risks presented by unavoidable hazards should be analyzed using conventional methods.  Decision analytic tools should be used to compare the resulting options and weight the relevant criteria.  The results of the analysis should be reported to the public in a manner that enables informed decision making by consumers while encouraging responsible innovation by industry.  The process should be ongoing and the analysis should be improved continuously.

TeX source

Since large-scale manufacturing processes are highly integrative, with each stage depending on the output from the prior stage, the functional requirements of a single manufacturing stage can be extremely specific and large changes can be disruptive to the manufacturing chain.  An analyst considering a single manufacturing process (limited scope) may have limited flexibility in selecting alternatives.  On the other hand, an analyst considering an entire product system (broad scope) may be able to envision a much wider array of alternatives. Figure [above] shows a generic product life cycle and compares the minimal scope for an alternatives analysis to the maximal scope, which is the entire product system.

So I come down with a salvaged Sempron 3800 box running Ubuntu Jaunty (total cost: $0) and quickly put that in place of my mom’s dinosaur.  Using VMWare Player (total cost: $0) we installed a bootleg copy of Win2k (total cost: $0, but not legally) in order to support her Quicken habit (GnuCash, my personal (though reluctant) choice for personal finances, sadly does not rise to the level).  On John’s end, the ancient CRM database program Q&A doesn’t run under Windows 7 but it does under dosbox; and Outlook express did not seem to be capable of transferring settings to regular-old outlook under Windows 7 (I didn’t work on this at all, but John assured me he spent a good bit of time) but Thunderbird imports it just fine… and finally, an old Agfa Snapscan e-20 is not yet supported on Windows 7, but the binary driver is still usable by xsane if it is extracted from the WinXP self-extracting executable (with a little help from wine; thanks to this post).

The only disappointment of the day was in how complicated it is to transfer a thunderbird profile from one computer to another- requiring showing hidden files and editing configuration files.  I could do it just fine, but it’s too complicated for John to do for someone else.  There is a ‘managed’ way to do it, but it requires installing another application (which is windows-only, by the way). Why is there [still] not an “export profile” option? It could just store the profile folder in an archive which can be imported elsewhere.  Maybe there’s a good reason for not doing this but I can’t think of it.

So minor config issues notwithstanding, OSS is finally reaching the level of being generally usable by the broader public.

UPDATE 2009-12-28

More good news: my Dad and his fiancée will be using my Wordpress installation to host their wedding page, which will be found at monaandron.com and hosted (presently) by FreeBSD.

Then some bad: Thunderbird 3.0 has another big FAIL: clicking “get mail” doesn’t get mail??  The configuration interface has also changed in subtle ways that make it very difficult to troubleshoot problems for non-tech-savvy family members over the phone.  Very frustrating.

Dutch study (pdf in english) on the life cycle impacts of reusable vs disposable cups for coffee service found that hand-washed ceramic coffee cups are the most impactful way of consuming coffee, machine-washed porcelain coffee cups are the second-most impactful, and disposable single-use paper cups are the least impactful, with the notable exception of very high ozone depletion potential due to chemicals used in paper cup production. Polystyrene was middle-of-the-road.  But read on…

Looking closer, the assumptions of the study are a little eye-opening. In particular, the study assumes the reusable cups are washed after every use, and that hand-washing uses an always-on under-sink electrical water heater.  In this case, the environmental effects of the ceramic-mug system are “virtually entirely the result of using electricity” (46).  Electricity is also the culprit in dishwasher-cleaned porcelain mugs, although dishwashers are more energy efficient than standby electric water heaters.  Finally, the non-electricity-caused environmental impacts of the reusable systems come from eutrophication from detergents; non-phosphate detergents would mitigate this effect.

Most of these issues are considered in the sensitivity analysis (they didn’t talk about phosphate-free detergents).  All you coffee drinkers should be relieved to know that simply by washing your mug with cold water (which the study calls “hygienically questionable”), ceramic-mug drops to rock-bottom in terms of impact.  Even under their high-impact assumptions, washing with hot water is ok as long as you only wash after 4.5 or more uses.

The moral is that before you sneer at people drinking from disposable coffee cups (as I often do), you must first ask yourself, “are these the type of people who leave the hot water running when they do dishes and/or use a fresh cup every morning?”  If they are, maybe you should thank them for using disposables instead.

Now, back to work…

On the morning of 8/13 the sky was deep red and purple; then, as the evening drew in, the temperature dropped suddenly down to around 60F, and a thick blanket of fog covered the coast.

This is what I saw leaving my office that afternoon:

Your goods would be subject to a 4.4% tariff, unless you operate in Cuba or North Korea, in which case your tariff would be 60%.

This definition is interesting because of the peculiar requirements it places on garments it includes.  Say the weight fraction of down is x, the weight fraction of other waterfowl plumage is y, and the remainder of the garment (which is made primarily of man-made fibers) is z=1-x-y.  For your garment to be included in this category, the following three requirements would have to be met:

1. x + y > 0.15 (green)
   
2. x / (x+y) > 0.35, or in other terms, x > 0.538 y (blue)
   
3. x > 0.1 (red)
   

These requirements are illustrated in the figure below:

If you run awry of these requirements, and your overcoat also does not “contain 36% or more by weight of wool or fine animal hair,” then the rates go up.  The regular-folks tariff jumps to 27.7% and the dirty-commies rate hits 90%.

In 2008, by the way, we imported 36,485 dozen such coats (reported in dozens), having a total customs value of $3,472,687 and having a weight of 277 metric tons, almost 94% coming from China.  That comes out to an average customs value of $7.93 apiece for a coat weighing on average about 630 grams.

I suppose, instead of researching coats I could probably have been working just now.

p.s. My convert-figure-to-image script doesn’t work because the transparency effects require PDF rendering.  So I had to crop it manually from the PDF using gimp.

I’m quite fond of PSTricks, a LaTeX package which allows the author to make use of PostScript’s extensive computational and graphical capabilities from within a TeX document.  My project for this week was to come up with a way to give these figures to my advisor, who works exclusively in MS Word. So what I needed was a way to export TeX figures to graphics files. Here’s what I came up with.

Essentially, I create a file which includes the \pspicture environment and all the graphics commands, which I call somepicture.fig. I then \input{somepicture.fig} into my main document when I want the picture to appear. When I then want to convert the picture into a standalone graphic, I run a script called figuretops.sh(link to file) which performs the conversion by making use of the very neat “file here” notation in bash:

cat >> filename.tex <<EOF
\documentclass{article}
\usepackage{pstricks-add}
\usepackage{calc}

\begin{document}
\thispagestyle{empty}
\color{white}
\fbox{
\color{black}
\input{}
}
\end{document}
EOF

When I run LaTeX on this file, it gets to the blank \input{} directive and asks the user for the filename. I solve this by echoing the filename into the LaTeX invocation in the bash script. I then clean up temporary files and use

convert -density 300 somefigure.eps somefigure.png

to convert the eps image to a 300-DPI image.

Here’s a source figure, the bash script, and an image it produced:

Here’s another (naively crafted: TeX source) but more interesting figure:

The US environmental protection agency are in the process of developing a fuel economy standard for hybrid and plug-in electric vehicles, and according to a recent draft version of the standard, General Motors’ flagship entry into the “range-extended electric vehicle” market would clock in at 230 miles per gallon.

Recently I entered a debate in defense of the EPA’s rating.  It was a retaliation against the claim that GM and the EPA have entered into a conspiracy in which the standards are designed to the specific product available and politically palatable, and the net result would be consumers’ further disillusionment at the very notion of “clean energy” when they discovered that their gleefully acquired products failed to perform to their expectations.

It was a Master’s student, and so naturally I assumed that he hadn’t thought out his arguments.  Comically, I then undertook the very error I was presuming, and eventually after working through the question, I came to an understanding of the physical conditions of the problem which I now share with you. The essential innovation of GM’s car is its all-electric drivetrain.  The motors which turn the wheels are all electric, and they’re powered by an on-board generator which is attached to a traditional internal-combustion engine which burns gasoline.  This is more efficient for two reasons.

First, you dispense with the transmission.  Instead of having to amplify the relatively narrow dynamic operating range of a combustion engine to match the realities of vehicle speed, we make immediate use of the full flexibility of the direct-current motor.  Ultra-high torque at low speeds, minuscule power consumption at high speeds, and a direct connection to the wheels.  Electronic speed control. Second, the gas motor which still supplies all the energy (when off-grid) has gone from the center of the vehicle and the Source of All Things to a specialized component.  It has exactly one purpose– to run a generator– which means it can be highly tuned to a specific optimal operating point and switched on when needed.

My misstep in the debate was in overestimating the energy in a gallon of gas.  Gasoline has a higher heating value of about 47 megajoules per kiloram [MJ/kg], which comes out to roughly 130 MJ or 36 kilowatt-hours per gallon.  The conventional four-stroke gas engine implements the Otto cycle, a well-characterized and carefully orchestrated alternation of isochoric and adiabatic processes.  Isochoric means “constant volume” (i.e. top dead center and bottom dead center of the piston stroke) and adiabatic means “no heat transfer” (i.e. compression and power strokes).  Maybe I should write another post about that sometime.

The thermodynamic efficiency of the Otto cycle is known; it’s a function of the working fluid (in this case a mixture of gasoline and air) and the compression ratio of the engine, and its upper limit is defined by the detonation temperature of the fuel (think anti-knock, octane rating).  In the case of modern gasoline engines, it’s about 47%.  This means that if you’re burning gasoline, you give up 18 kWh/gal “when you drive it off the showroom floor.”  Add to that mechanical losses and imperfect assumptions and you take off another (say) 22-23%, which puts the gas-tank-to-flywheel efficiency of the well-tuned engine at a practical maximum of about 36%. In the case of the Volt-style transmission-free hybrids in which the combustion engine drives a special purpose alternator, something close to that efficiency might be achievable.  Perhaps an optimistic assumption would be that 13 kWh of electric power can be harnessed from the combustion of a gallon of gasoline, where it can be stored into a battery and then run through copper windings and onto the pavement and into the winds on the open highway.

According to Chevy’s engineers, their car takes 25 kWh of electric power to go 100 miles, which comes out to 52 miles per gallon of gas at 36% efficiency.  In order to get the 230-mpg figure they must have assumed it drives 40 miles ( on a / free of ) charge and then another 12 miles at 52 mpg.  This bears resemblance to a hypothetical test in which the fuel consumption of a vehicle is measured over a 52-mile course. The interesting thing to note here is the bias in measuring gasoline consumption but ignoring electric grid power as “free.”  For me, the marginal cost of 25 kWh is about $3.50, compared to $6.50 for two gallons of gas, but if I were charging my car every day that would go up to $6-7, because Southern California Edison uses tiered pricing (which I heartily endorse).

As far as greenhouse gas emissions are concerned, it’s hard to do worse than coal.  If you get your grid power from a coal plant you’re putting out about 1 kg of CO2 per kWh.  US average is more like 0.62 kg/kWh; in California we’re down around 0.45 kg/kWh, due largely to the fact that we have no coal plants, plus we’ve got all that nuclear and hydro power.  The internal combustion engine, running at its “practical maximum” efficiency of 36%, comes in at around 0.86 kg/kWh.

So any move to plug-in vehicles is an improvement, as long as it’s coupled with a move away from coal.  That’s fortunate, because the people who love coal power are the same people who are unlikely to buy plug-in hybrids, so advocacy for these two things together won’t anger any additional people.

The more interesting question surrounds the 25 kWh/100 miles figure.  What speed of travel does that represent? I wonder what theoretical lower bounds can be produced on the amount of energy required to move a car 100 miles?  Because the 13 kWh/gallon of gas applies to any power plant burning gas in a 4-stroke engine, the only way to improve energy efficiency is to reduce the energy needed to go the distance.  Or to run off the grid.

Pity how badly I botched this topic in my public discussion.  Frankly, I’m embarrassed.  The whole thing can’t go away fast enough.