Line into Color, Mesh into CFD

Painter Helen Frankenthaler's work has been on my mind since seeing the exhibition Fluid Expression: The Prints of Helen Frankenthaler at Fort Worth's Amon Carter Museum of American Art earlier this week. While there I also purchased and then read John Elderfield's book on her painting, Line Into Color, Color Into Line. Both left me with a much stronger appreciation for her work.

As often happens to me, certain concepts, ideas, or statements about art trigger analogies to my work in computational fluid dynamics and mesh generation. Such was the case when Frankenthaler was quoted in the book as saying

I felt more and more that the drawing should come from what the shapes of the colors are; rather than, "I am arranging this with lines or confinements or patterns." And I do very much believe in drawing, especially when it doesn't show as drawing... When I talk about drawing, I mean "how are you getting your space," not where the pencil is going.

To put that quote in context, the book's theme was how Frankenthaler exercised three types of lines (drawn lines, the perimeter of regions of color, and the edge of the canvas) to great effect in her paintings that are more typically known for their ethereal washes of color - poured, stained, painted or otherwise.

Helen Frankenthaler, Sesame, 1970. source

But first a bit of background. Within the world of computational fluid dynamics, the mesh is the digitalized version of the object around which you wish to solve the equations of fluid motion - digitalized so the computer can understand it. Think of it as the scaffolding on which the fluid will be simulated. In the illustration below (image source), the mesh lines around a ship's hull define where the computations will be performed.

When the equations of fluid motion are solved on the mesh, the results are often presented as graphical contours of a some property of the fluid like pressure as shown below on the ship's hull (image source). In this picture, red represents high pressure and blue represents low pressure.

The analogy I'm making between Frankenthaler's paintings and computational fluid dynamics is lines are meshes and color is the CFD results. When she says "drawing should come from what the shapes of the colors are" I hear the case for mesh adaptation (closely coupling the mesh to the actual flow of fluid instead of just the geometry). When she says she believes in drawing "especially when it doesn't show as drawing" I hear the case for "invisible" mesh generation (because meshing is not and end unto itself, only a means to an end). Regardless of whether we're talking about lines in terms of mesh or lines in terms of boundaries of the regions of color, it's true that the mesh is how you're getting your space, the space within which the simulation will be performed.

Other than the fluidity with which she applies pigment to canvas, there's little that visually ties her paintings to CFD (the images above make this clear). The drawn lines in Sesame don't look anything like lines in a mesh and the regions of color don't look much like fluid flow. But it's Frankenthaler's process and approach and her way of thinking about the interplay of line and color, mesh and CFD, creating and defining space, that brings art and science, a bit closer, at least in my mind.

Not Hiring Computer Scientists?

The CEO of Dittach wrote an opinion piece in yesterday's Wall Street Journal with the title Why I'm Not Looking to Hire Computer Science Majors. (Behind paywall.)

Let's look at a few of his statements.

"Finding [software developers] is the toughest task." Not really. Finding good ones is.

"Startups have to compete with hegemons like Google and Facebook that offer extraordinary salaries for the best talent." Yes, and it's not just the money; it's the name recognition. They could work for Google or they could work for... what's your company's name again?

"Computer science departments prepare their students for academic or research careers and spurn jobs that actually pay money." Then where are Google and Facebook finding all the people you complain they hire?

"There isn't a single course in iPhone or Android development in the computer science departments of Yale or Princeton." Repeat after me: college is not a trade school. You don't go to college to learn a skill, a language, a tool. Skills can be learned. Skills can be taught (by you, the employer). You go to college to learn to learn the science of programming. Also, I have nothing against our Ivy League friends, but are Yale and Princeton really the first place you're looking for CS grads?

"Today we insist on higher education for everything." Yes, college isn't for everyone. Let's get kids into the trades when appropriate. Have you read in the WSJ about how much welders are getting paid these days?

"A serious alternative to the $100,000 four-year college degree wouldn't even need to be accredited - it would merely need to teach students the skills that startups are desperate for." Didn't you just poo-poo the 12- and 19-week programs for cranking out people only interested in career transition and not the "love of coding for its own sake"? Maybe this is the business opportunity: a 2-year, for-profit institute of programming practice. Hell, make it for women only and you'll kill two birds with one stone (i.e. attracting women to the STEM fields.)

To view this from another perspective, think about the legitimacy of the following statements. I'm not looking to hire Ivy Leaguers. Mathematicians. People with graduate degrees. Men.

So I guess in the end I don't really grok Mr. Gelernter's bias. Some of the best programmers I know are CS majors. Others are engineers. And others are mathematicians. All have college degrees. All love what they do. None are Ivy league.

But, this is just my opinion. I could be wrong. All I've done is co-founded and boot-strapped a 20-year-old 30-person software company.

P.S. In the interest of full disclosure, I am on the advisory board of my alma mater's engineering department. My oldest son has a masters degree in CS and is gainfully employed. My youngest son is an undergraduate CS major. Furthermore, I recognize that the title of Mr. Gelernter's piece is click-bait for an article that's really a jab at higher education (misplaced IMO) and not a revelation of his hiring practices.

The Arts of The Sciences

"Are nuclear explosions art?"

OK, you've piqued my curiosity.

Photograph of the Castle Bravo nuclear test in 1954 of the United States' first hydrogen bomb. Image from Wikipedia.

Alex Wellerstein (@wellerstein) asked this question at the beginning of his article Art, Destruction, Entropy on the Restricted Data blog about nuclear secrecy.

On its surface the question seems patently ludicrous. A nuclear detonation brings to mind heat, fire, blast, radiation, fallout, war, conflict, apocalypse, destruction, and death. Anything but art.

Chopping up a piano with an ax isn't art either.

Or is it?

Art, Exhibits, and Museums

Let me back up. Wellerstein's inspiration was an exhibit at Washington's Hirshhorn Museum that he saw called Damage Control: Art and Destruction Since 1950 that features both film footage of nuclear test detonations and Raphael Montanez Ortiz' piano destruction (among many other pieces by many other artists). Ortiz is not new to this, having been involved as far back as 1966 in the Destruction in Art Symposium, a reaction to the human race's "will to kill" and a recognition that death itself, not life, needed a sense of the transcendental.

Wellerstein, a historian at the American Institute of Physics, seems to take the position that the exhibition's inclusion of wall-sized looping footage of U.S. above-ground nuclear tests from the 1950s is art only because it is being shown in an art gallery and because of the innate aesthetics of the images. According to him, this focus on the aesthetics distracts the viewer from the larger context of these events such as their intended cataclysmic use, the effect of fallout on innocent bystanders, and the societal cost of the Manahattan project and subsequent decades of nuclear weapons development. He asks whether this particular presentation of nuclear force robs it of something, specifically its larger global context.

Screen shot of the Hirshhorn's webpage for the Damage Control exhibition.

I seems to me that this decontextualization, to use his word, is precisely what contributes to the film's artistic value. Or maybe, more accurately, it's the recontextualization of the work when juxtaposed with Ortiz' and others. To rely solely on aesthetics as a barometer of art misses the point. Certainly the original purpose of the detonation films was one of scientific and historic record and their aesthetic value is merely coincidental.

But those two purposes aren't necessarily at odds either. I am reminded of something William Deresiewicz wrote in The American Scholar: "We ask of a scientific proposition, “Is it true?” But of a proposition in the humanities we ask, “Is it true for me?” So while the nuclear scientists could use the films to learn scientific truth, their inclusion in the exhibit allows the public to assess their implications in a more personal light. In other words, by placing works of scientific origin within the context of Ortiz' work and the works of other artists it humanizes what might otherwise be too incalculably vast to be comprehended.

Are Squares Art?

Let's look at this from a slightly different perspective. You might just as easily ask whether squares are art. This would be coincidental because also on the Hirshhorn's website is a section dedicated to an exhibit on Josef Albers: Innovation and Inspiration. Albers is probably best known for a series of paintings called Homage to the Square that he used to explore the interaction of colors through the form of inset squares. Through the interaction of form and hue Albers was able to achieve sensation of depth through the picture plane and a shimmering vibration within the plane itself.

Josef Albers, Homage to the Square, 1966. Image from the Josef & Anni Albers Foundation website. (I chose this painting in particular because it has a certain abstract similarity, in my opinion, to the Castle Bravo photo above.)

Albers was also a great color theorist and author of Interaction of Color, a pivotal work on color. (See also the app.) In the preface of this important work Albers wrote "Just as knowledge of acoustics does not make one musical... so no color system by itself can develop one's sensitivity for color." "What counts here - first and last - is not so-called knowledge of so-called facts, but vision - seeing."

Note that it is the act of seeing - by the observer - from which value is derived. The intent of the producer may have been something totally different.

It is the relative interaction of various colors and forms - i.e. their context - that change a color into something else. Two different colors appear to be the same, the same color appears differently due to interaction with a third, two colors in proximity create a third in their midst. Many monochromatic painters achieve effects on the entire gallery in which their works are placed. 

Interaction of Color by Josef Albers. (Highly recommended but not an easy read.)

So questioning whether squares are art is similarly off target as questioning whether film footage of nuclear detonations are art. Albers' forms interact with each other through their placement and relative color allowing you to see more than shapes and hues. The detonation film puts Ortiz' piano destruction in its own new light and vice versa, allowing us to perceive each in a new way.

Multiples and MIRVs

Wellerstein also raises the issue of producer's intent, questioning whether something made specifically for one purpose can be regarded in another. From the art world, he cites work by Andy Warhol (included in the Damage Control exhibit) and exemplified by the image below.

Andy Warhol, Red Atom Bomb, 1963. Image from the C4 Contemporary Gallery. (Note: I don't know whether this work was included in Damage Control. Its inclusion here is simply for illustrative purposes.)

Wellerstein compares Warhol's work with the photographic series of the Trinity explosion taken by lead photographer Berlyn Brixner (see below).

Berlyn Brixner, TR-NN-11, 1945. Image from the Restricted Data blog. (I tried to find an independent source of this image but failed.)

Clearly, Brixner's purpose was scientific while Warhol's was artistic. A cursory interpretation of Warhol's work is that it makes  a statement on the commoditization of destruction. I would argue that his colorization - and admittedly I'm not a huge Warhol fan - draws something from Albers, if not in this particular work then in his others.

The question remains whether there is any artistic value in the Brixner photographs. Again,  I would argue that there is - or can be - depending on context. I'd make my argument by introducing the work of Edward Tufte, author of the seminal and beautifully produced text, The Visual Display of Quantitative Information. As the name of the book implies, Tufte is a master of making dense data understandable through visualization. One of his techniques is the method of small multiples.

The essence of Tufte's concept of small multiples is to contextualize data by answering the question "compared to what?" (Chess players can see a good example of small multiples on Tufte's website for Magic Knight's Tour.) As opposed to a single photograph of the Trinity explosion, Brixner's photograph above compares the fireball are various times. A single photo almost begs the question "So what?" while the array of photographs reveals details about the evolution of the explosion. And here I would make the argument that it's done in a highly visually appealing manner.

To complete the tie-in, I'll mention that it was via Tufte's recommendation, after attending his seminars in Dallas, that I purchased and read Albers' book.

So What?

So what exactly is my point? First, I have kept to this blog's intent to be a rambling stream of consciousness. Second, as frequent readers know, I have this fetish for instances where my various interests are juxtaposed. In this case we have Cold War and nuclear history crossing with modern art and data visualization.

But that's what's in it for me. What about you?

Are nuclear explosions art? The answer, like most things in life, is "it depends." They can be simple scientific and military tools to the extent they impart understanding and support national objectives. And they can be art in how they make each of us think and feel about life and death.

Like the old - really old - SNL skit said, "Sometimes a banana is just a banana." And sometimes a square is just a square. But through contextualization of a square's color and form it can become a striking work of art.

And just because science and math are quantitative, it doesn't mean their execution and explanation and insight can't also be executed beautifully.

Frederich Nietzsche is credited with saying something that seems quite apropos for the subject at hand: "We have our arts so we won't die of truth."

Insight to a "T" in Mechanical Engineering

It's that time again when I extol the virtues of some articles in Mechanical Engineering magazine (this time the April 2013 issue).

First comes The Psychology of Insight, an article about three relatively simple things that engineering professors can do to keep students from changing majors out of Mech Engr. And they do seem relatively simple but have been shown to actually work.

1. Use everyday examples in lectures. Because students will likely have had experience with these things it keeps them more engaged and promotes understanding. This idea also addresses a complaint of employers, that new graduates lack practical knowledge.
2. Develop 3D spatial/visual reasoning skills. There's something called the Purdue Spatial Visualization Test that serves as a predictor of student success in courses like graphics or CAD. Furthermore, work by students to improve their PSVT score has a direct influence on grades.
3. Ensure faculty-student interaction outside the classroom. This information interaction not only provides positive feedback but helps students see the professors as members of the profession.

Second was To a T: Recruiters Finding Breadth Outscores Depth in Engineering in which we learn that engineers with a broader skill set (the horizontal line in a letter "T") are more prized by companies than those with a deeper skill set (the vertical line in a letter "T"). Part of the reasoning is that a broad skill set promotes system-wide thinking and breaks down barriers to working with other disciplines. Personally, I believe it's paramount that an undergraduate engineering education provide students with a significant breadth of capability.

Just so that you don't think I'm a blind Mech Engr fan boi, I'll cite a letter to the editor in this issue that may be one of the most depressing things I've read about the profession. It starts out "It is true that a tiny fraction of engineers are allowed boldly creative work. The rest of us make careful calculations to effect small modifications." It's as though the author thinks most engineers are Milton from Office Space. The author believes that engineering is oversold by professors who are able to consult on the most challenging industry problems and are accordingly blinded to what most engineers actually do. (Which is contrary to the old adage that those who can, do; those who can't, teach.) Perhaps the most damning part of this letter is the statement that engineering careers can become obsolete as opposed to a psychologist who can keep practicing for an entire career based on their university education. I doubt the latter is true. And if you become obsolete, you only have yourself to blame. The author also (mistakenly, in my opinion) believes that companies are more likely to pay for entry-level engineers who have been "trained in all the hottest new methods" as opposed to someone with 30 years experience. Repeat after me - an engineering undergraduate education is not a trade school that results in training in "hot" methods. (See above about breadth.) Also, an engineer with 30 years experience should be able to learn new tools. [Hmm, why did I write more about the article I didn't like than the two that I did?]

Strategic Bomber Gets a Double Take

There's a nice article in the March 2012 issue of Aerospace America (sorry - only available online to AIAA members) about the need for a new U.S. strategic bomber ("Strategic bombers - relevant again" by Richard Aboulafia). B-1s are due to retire in 2030, B-52s in 2040 at age 80 (yes, an 80 year old aircraft), and B-2s in 2050. Now is the time to start designing and building a new bomber to enter service when its venerable older siblings start retiring.

But that's not why I'm writing this. On page 22 there's a photo montage captioned "The U.S. bomber fleet offers [emphasis mine] a wide mix of aircraft." Included in the photo with the previously mentioned aircraft are the FB-111 and B-58.

Whoa. That got a double take.

Scan of a photo from page 22 of the March 2012 issue of Aerospace America. Note the verb tense in the caption.

The F-111 Aardvarks retired in the late 1990s. I had the pleasure of seeing a couple on the factory floor at General Dynamics (now Lockheed Martin Aeronautics) in Fort Worth. The B-58 Hustlers were retired in 1970 - 42 years ago. I saw one once in a museum.

I read that caption and immediately flipped back a page to see what exactly I had missed in the article. Are they bringing them out of retirement?

Alas, it appears to be just a proofreading oversight.

Communications of the ACM Scores Big in Nov 2011

This month's issue of Communications of the ACM hits a home run with several notable articles that I thought I'd share.

Hexagonal grid with 480 km resolution.
Goal is 4 km.

Modeling Chaotic Storms starts things off with a look at simulations of global weather patterns up to and including the potential for 60-minute advance predictions of extreme weather. In fact, the authors write that future advances will come from simulation rather than increased scope of observation (i.e. satellites and radar). One of the challenges faced by modeling is the rather sparse set of data on which any computation is based - so-called initial conditions. Like most simulations of fluids like air or water, the nature of turbulence, the chaotic motion of a fluid, is also poorly understood. The computations required for weather modeling are far from trivial and researchers are turning to new technologies like GPUs (graphical processing units) that ran run 25 times faster than traditional hardware. The challenge here is that GPU programming requires a non-trivial up-front investment to modify the software.

Whether or not software engineering is truly engineering is addressed by Will Software Engineering Ever Be Engineering? The author makes a strong argument for the fact that it isn't now and likely never will be. The arguments for the latter are:

1. Engineering is ultimately about physical systems.  Software is not.
2. Software is a large profession - on the order of law or medicine - and is therefore more likely to split with engineering than join up with a profession twice its size.
3. MIS or IT might be a better home for software as its more likely to be practiced in those areas than engineering.

In an article that's one of the most explicit treatments that I've ever read, Teaching-Oriented Faculty at Research Universities lays bare a truth that many people, including students, ignore. If you think that faculty at universities are hired and promoted for their teaching, you're wrong. They are hired for their research and their ability to bring in funded research. You'd also think that despite that they can teach.  Alas, no. "Active research is not a predictor of effective teaching." Therefore, teaching-oriented faculty (TOF) provide a valuable service, not only by the act of teaching courses (especially undergraduates) but they also undertake critical functions like curriculum development, advising, and course development. And yes, they do some research too. The fact that TOF at a university is almost seen as a novel idea is quite disturbing.  Or I'm just naive.

Other interesting articles touch on diverse topics such as the rejection of the Google Book Search settlement (centered around whether scanning books for the purpose of indexing search results was fair use or copyright infringement), a call for software liability laws (most if not all software licenses include limits on liability including "merchantability and fitness for a particular purpose"), and why high performance Fortran hasn't had any staying power.

This issue alone was well worth the cost of my annual ACM membership.

I Like the Swiftpoint Mouse

I've been using the Swiftpoint wireless laptop mouse for quite some time now with my netbook and it's a pretty handy device.  I'm not coordinated enough to use a touchpad well so I was searching for an alternative and came across the Swiftpoint.

As you can see from the photo it's got a unique design where you grab the red part with your thumb, put your middle finger on the other side, and work the buttons and wheel with your index finger.  The wheel on the right can be flipped with your index finger or the whole device picked up and run alongside whatever surface you're using.  And speaking of surfaces, you can use just about anything.  I've used a conventional desktop, an armchair, carpet and the mouse works great each time.  Also, if you look closely in the photo you can see some damage where I dropped it about 6 feet onto a tile floor and it kept on working.

You can also see in the photo the Swiftpoint's charging station sticking out just a bit from a USB port.  The station is magnetic and attaches to a recess on the bottom of the mouse so you can carry around the netbook with the mouse attached - very convenient.

If you're looking for a highly portable mouse for your netbook or laptop, give the Swiftpoint a chance.  Check out the videos on their website at http://www.americas.futuremouse.com/.

My month's old Swiftpoint in action.

I received no compensation of any kind for this review.

Good Stuff in Communications of the ACM

There are some good articles in the July 2011 issue of Communications of the ACM for those of you who are computer inclined.

In Non-Myths About Programming, Moti Ben-Ari takes students' negative perceptions of computer science and rather than dispelling them mitigates them as truisms (non-myths) when compared to real life.  For example, the first non-myth is "programming is boring."  Welcome to the real world, folks.  All jobs - doctors, lawyers, airline pilots, police - become boring at a certain point, especially after you've put in several years and have developed certain skills.  The real question he asks is whether your career choice allows you to handle certain types of routine tasks.  This is an excellent article and great reading for anyone considering a career in CS.

The article Passing a Language Through the Eye of a Needle by Roberto Ierusalimschy et al present the aspects of scripting language design that supports its ability to be embedded in a host program.  The article presents the case for Lua and because a friend uses Lua in his product it was a nice read.

The cover article, Algorithmic Composition: Computational Thinking in Music by Michael Edwards, is incredibly interesting.  The music theory stuff went clear over my head but the discussion about the historical use of algorithmic techniques going back to Mozart and earlier proves yet again that what was old is now new.  Computer based algorithms.  Of course, the authors really got my attention when they cited the algorithmic composition work of Brian Eno and Steve Reich, two musicians whose work I enjoy.  Although I was surprised that Eno's generative work 77 Million Paintings wasn't named.  Regardless, the article was quite excellent.

Energy, Chocolate, and Mechanical Engineering

Mechanical Engineering magazine has been on a roll lately with articles that I think are pretty damn interesting.  I read them but also feel compelled to share them with you here.  Such is the case with their January 2011 issue.

In Myth vs. Fact, two folks from MIT's Sloan School of Management try to separate myth from fact when it comes to energy policy.  They write "Before we can embrace appropriate energy policies, we have to face the hard truths about the technologies available to us."  So here I'll just repeat their headlines verbatim for you to deliberate, discuss, and debate.

• Truth: Without significantly higher energy prices, energy demand will continue to grow - especially in developing countries. 
• Myth: The world is running out of fossil fuels.
• Myth: Renewable energy generation technologies are free from environmental concerns.
• Truth: Continued reliance on fossil fuels is altering Earth's energy balance.  The risks of extreme temperature changes and subsequent impacts will grow substantially unless concrete actions are taken.
• Myth: Getting started on reducing greenhouse gas emissions will require transformative technologies.
• Truth: There are opportunities to greatly reduce energy intensity through gains in efficiency.  These gains could be made by both technological advances and price-induced substitution.
• Myth: There is little room for improvement on fossil energy technologies - fossil fuel extraction, internal combustion engines, and so on.  Therefore, only small advances are needed before alternative energy technologies are competitive.
• Myth: Explosive growth in renewable fuels in recent years is evidence that these alternatives are economically competitive with traditional fossil fuels.
• Truth: Subsidizing renewable energy generation doesn't lower the cost of alternative energy; we simply pay the cost in other ways - and at the expense of cost-effective market-based policies.
• Truth: The true costs and benefits of available alternative technologies are highly uncertain.  Therefore we should all be careful about predicting technology winners.

The same issue provides dessert in the form of a great article on the high-tech challenges involved in making chocolate, Engineering Taste, authored by researchers at the University of Birmingham in the U.K.

The complexity here is based in the fact that molten chocolate is a non-Newtonian fluid whose viscosity changes with applied shear force.  What's important is to solidify this liquid in one of its six stable solid forms, one of which is preferred from the standpoint of mouth feel (melts at body temperature) and visual appeal (glossy).  Unfortunately, this solid form isn't stable and often the result is a less desirable solid form that looks funny and tastes gritty.  (I recognize this taste as a brand that begins with "H".)

One key to achieving the correct solid form is the cooling rate.  One method uses a very slow cooling of 1 or 2 degrees Celsius per minute.  But another technique called the "frozen cone" method chills the chocolate in 3 seconds using a cone plunger at -20 degrees Celsius.  Obviously, this works well for creating thin chocolate shells.

So now we have some insight into the science behind the fluid dynamics of something we're all familiar with but have probably not thought much about.  In addition to mouth feel and taste, another area interest for these researchers is the nutrition and digestion of this manufactured food.  My parting chocolate advice is to stay away from the "kiss" - that little pointy tip always breaks off and leaves tiny brown stains on your shirt.