This plant is the most common Hoya species and a very popular house plant, so you wouldn't think that yet another example of it blooming would be a big deal. However, this particular plant has a lot of history in my family, making it a big deal to me. I got it as a cutting off a plant that was owned by my grandmother Letha. Letha's plant was in turn taken as a cutting off a plant that was owned by her grandmother Augusta.

Letha Marie Rau (Born) [1908 - 1991] My grandmother	Augusta Charlotta Born (Raasch) [1841 - 1907] My great, great grandmother

The plant's full lineage detailed below for future reference:

• The original plant - "Hoya A" - was owned by Augusta. Acquisition date/place unknown.
• When Augusta died in 1907, the plant was transfered to two of her daughters: Emilia Augusta Born [1868 - 1952] and Lydia Louise Born [1881 - 1969]
• Sometime around 1945, Letha received a cutting of "Hoya A" from Emilia or Lydia. This cutting - "Hoya B" - remained with Letha until her death.
• After Lydia's death in 1969, Berta Bohnsack (Rau) acquired "Hoya A" from Lydia. It remains with her to this day. A cutting of "Hoya A" was also given to Barbara Born Kraai at this time. She reportedly still has it and her two daughters have cuttings of it.
• After Letha's death in 1991, Carol Rau acquired "Hoya B" and has it to this day.
• Sometime before 2006, Carol took a cutting off of "Hoya B", creating "Hoya C".
• Carol gave me "Hoya C" when I visited Watertown for the 2006 Tour de Kota.

I have two photos of Letha's "Hoya B" in her rural Watertown, SD home [one, two]. I don't know when they were taken. Probably in the mid 1970s to late 1980s.

In December, my brother Craig and I made a plant stand for "Hoya C", attempting to partially emulate the plant stand that my Grandfather Carl made for Letha's "Hoya B":

And so Augusta, born in Pomerania in 1841, has in me a fifth generation ancestor who's made an offshoot of her house plant bloom in New Mexico more than one hundred years after she first made it bloom in South Dakota. This is very important to me.

It becomes even more important now that Elisa and I have our first child on the way. We'll try to be good parents and good care takers of the plant. Then, as a result, I hope that one of our ancestors can continue the line by someday treasuring a related Hoya blooming for the first time in their own home.

Here's the document.

Update 2/18/2008:

I created a new example that's similar to the previous one, except that it demonstrates a) The use of jQuery, b) some of jQuery's easy build in effects, c) the use of jQuery's $(document).ready(function() ...) and $(...).bind(...) to totally remove all Javascript from the markup.

I really like jQuery. I highly recommend that you have a look at it, if you're looking for a "web 2.0" Javascript library.

Many photos are available here. A few select ones are shown below:

General People, Places, and Things of India

Team Members and Me in India

Vechicles

What Marxism Gets You

For example,

This is easy enough in a many GUI image manipulation packages, but it tends to be tedious for a large number of images. Further, if you're not careful to use the right options, the rotation process will uncompress and recompress the images, degrading their quality.

Thankfully, there's an awesome command line utility called jhead that does lossless JPEG rotation and a whole lot more. Here's an example. Quick, easy, and scriptable:

$ jhead -autorot *
Modified: IMG_1007.JPG
Modified: IMG_1037.JPG
Modified: IMG_1038.JPG
Modified: IMG_1039.JPG
...

jhead's author has a very interesting home page, including information on a home-made mechanical organ and a link to Raphi Giangiulio's Homemade Pipe Organ, which is one of the coolest things I've ever seen.

Here's a screen shot of me working on the final project:

I enjoyed the class's back propagation learning assignment. My solution was implemented with Ruby, which resulted in a very elegant implementation (IMHO).

More photos.

I'll have to fix those 69 warnings that show up on this blog's front page sometime soon.

There are quite a few steps involved, but I think the results are worth it. If you read the Gnuplot epslatex documentation, you'll see that there are some techniques that could be used simplify the process (e.g., setting font information in gnuplot.cfg, but this in how I'm doing things for now:

Step #1 - Get plot working with a simple terminal, e.g., x11

The first thing I do is make a standalone script to do the plot with the x11 terminal. This allows me to very rapidly experiment with things like axis labels and the general form of the plot. Here's an example that I put in an executable script file - something like myplot.gnuplot:

#!/usr/bin/gnuplot -persist
# myplot.gnuplot
set term x11
set xlabel "x"
set ylabel "y"
set xrange [-1000 : 3000 ]
set yrange [-1000 : 3000 ]
set title  "Positions of the three object types in FlatWorld"
set grid
plot "type1.dat" title "Type 1: delta = -0.1" with points 11, \
     "type2.dat" title "Type 2: delta =  0.0" with points 12, \
     "type3.dat" title "Type 3: delta = +0.1" with points 9

Once this is in place, I go through an edit file, run script (./myplot.gnuplot), and repeat process, until I converge on a plot that does what I want. Then I move on to the next step. Here's what the x11 step looks like:

Download type1.dat, type2.dat, and type3.dat, if you want to follow along.

Step #2 - Get your plot working with epslatex terminal

Next I take the x11 terminal script and convert it to epslatex terminal. Something like the following saved to plot.eps.gnuplot:

#!/usr/bin/gnuplot
# plot.eps.gnuplot
set term epslatex color
set format "$%g$"
set xlabel "$x$"
set ylabel "$y$"
set key box
set xrange [-1000 : 2500 ]
set yrange [-1000 : 3500 ]
set grid
set title  "Positions of the three object types in FlatWorld"
set output "./position.tex"
plot "type1.dat" title "$\\Delta = -0.1$" with points 11, \
     "type2.dat" title "$\\Delta =  0.0$" with points 12, \
     "type3.dat" title "$\\Delta =  0.1$" with points  9

Note that the strings used to set labels, titles, etc. take LaTeX syntax, e.g., $x$ for an 'x' in the math font.

When you run this new script, two files are produced - an EPS file that contains lines, ticks, etc, but no text (position.eps), and a LaTeX file that includes all the text (position.tex).

Step #3 - Convert EPS file to PDF

To take advantage of packages like microtype, and hyperref, I use pdflatex to compile my LaTeX documents, so the EPS file must be converted to a PDF, before I can use it:

$ epstopdf position.eps

The resultant PDF (position.pdf) will look something like the following. Note that there's no text. That won't come until later:

Step #4 - Create a LaTeX file for your personal settings

Next, I create a LaTeX file (positions.tex) to encapsulate position.eps and position.tex and make the LaTeX settings I want, e.g., font type:

% positions.tex
\documentclass{article}
\usepackage{graphics}
\usepackage{nopageno}
%\usepackage{mathpazo}
\usepackage{txfonts}
\usepackage[usenames]{color}
	
\begin{document}
	
\begin{center}
  \input{./position.tex}
\end{center}
	
\end{document}

Step #5 - Compile and Crop Custom LaTeX File

Finally, we're able to combine the vector plot with text and symbols in the correct LaTeX font:

$ pdflatex  positions.tex

The result is positions.pdf, which is almost complete, except that it needs to be cropped down to just the plot and not a letter sized page. I use the pdfcrop package for this:

$ pdfcrop.pl positions.pdf
$ mv positions-crop.pdf positions.pdf

There you have it! A vector plot with fonts that will exactly match the rest of the text in your LaTeX document.

If you wanted to make the plot have a different font (say LaTeX's default Computer Modern font), go back to step #4 and comment out the \usepackage{txfonts} line in positions.tex which sets up a Times font.

Of course, all these steps should be put into a Makefile or shell script so you don't have to repeat yourself when making changes to the plot or its data.

I left Sandia and CDS last month after almost five years of service to take a new position. For two weeks now, I've been serving as Vice President of Development and Implementation for Pay Lynxs, Inc. - a software startup in headquartered in Albuquerque, with an exciting product and business model centered around banking money transfers for international remittance. That is, immigrants electronically sending money to their home countries.

In this role, I'll be concentrating on scalable software development and quality assurance processes to ensure that the company's products get to market correctly and quickly. In addition, I'll be working on system architecture, feature development, and project management. It's a very challenging and exciting opportunity for me.

I'm still riding my bike to work, except for Fridays when Elisa drops me off and picks me up so we can more efficiently execute our weekly go-out-to-eat night. The ride is ~ 23 miles round trip or about twice the distance it was to Sandia and back, but I'm very much enjoying the extra distance, as it gives me time to contemplate the challenges that face me in my new position and much of the route is either on bike paths or passes through beautiful neighborhoods that I normally don't get the chance to see. I'm very happy that Albuquerque has done such an excellent job making the city bicycle accessible.

The mathpazo (Palatino) LaTeX fonts are shown above, but I've also created PDF versions with txfonts (Times) and the the default LaTeX font (Computer Modern) for comparison.

There are two really great things about these plots:

1. They were created with a completely free / opensource tool chain, and they look quite good.
2. Using Gnuplot's epslatex color terminal allows me to combine vector EPS output with custom LaTeX post-processing so that, for example, the fonts of the plots exactly match those of my LaTeX documents, including all of the fancy math and other special symbols I can imagine.

The ability to create publication-quality plots like this will be a great benefit when I start working on my MS Thesis in earnest next semester. The process of creating these plots is admittedly somewhat arcane (the subject of an upcoming blog post), but this can be automated with a Makefile, and I think the results are worth the effort.

The images below were taken on top of the UNM bookstore looking out at the new Architecture building, the Frontier restaurant, and Central Avenue.

Before: 2,336 × 3,504 pixel (~ 8.1 MegaPixel) source images arranged in a 3x15 matrix

The digital panoramic photo mosaic process involves taking multiple overlapping source photographs of a scene from slightly different vantage points and then digitally "stitching" these images together to form a single composite image that appears as if is was taken from a higher resolution and a possibly wider angle camera.

An example of photo mosaic of UNM's Zimmerman Library follows. A 2×11 matrix of source photographs is shown, followed by the processed mosaic:

Before: 2,336 × 3,504 pixel (~ 8.1 MegaPixel) source images arranged in a 2x11 matrix

After: photo mosaic: 16,961 × 5,792 pixels (~ 98 MegaPixels)

This 16,961 × 5,792 pixel image is big!

One way to think about its size would be to print it out at a relatively standard 200 pixels per inch. If you did so, the print would about 7 × 2.4 feet.

Another way to think its size is to look at a small 100%-scale cropped portion of the image like the one shown below:

If you look carefully, you can see this student crossing the plaza in the right hand portion of the full mosaic.

For full-resolution high-quality output, creating one of these composite panoramas is not a trivial computational task. An example like the one shown above can take hours on a modern workstation, while processing extremely large examples like Max Lyon's GigaPixel image can take many days.

I propose an MS Thesis that will demonstrate how the performance of digital panoramic photo mosaic compositing can be improved through parallel computing. The goal will be to take a multi-day real-world serial example and after writing parallel modifications to existing serial codes, accomplish the same task much quicker in parallel (given access to sufficient computational resources).

After motivating the need for this work with some real-world examples, I intend to break down a typical compositing work-flow into processing steps that can be more easily analyzed, develop some test workloads, and then measure the serial performance of those workloads on each processing step using existing Open Source software (Panotools, Autopano-sift, Enblend, Hugin, etc.).

Based on these measurements, I will select the processing steps that constitute the largest part of the total processing time for further analysis. Their algorithms will be explored and their theoretical time and space complexity will be determined in the serial case. Next, parallel algorithms will be proposed, formally analyzed, and implemented with Single Program Multiple Data (SMPD) parallelism using the Message Passing Interface (MPI). That is, the existing Open Source software will be be parallelized using MPI.

Then, the parallel improvements will be tested on parallel computers, with both strong and weak scalability studies being performed on up to hundreds of microprocessor cores. The scalability achieved will be analyzed and compared to theoretical expectations.

Finally, the original serial workloads will be run in parallel and the their serial run times will be compared to the best possible parallel run times.

Administrative Detail

I intend to work on this Thesis for two semesters: fall 2007 and spring 2008. I will take three credits of Thesis during the first semester and then three more credits of the same the following semester. I plan on defending my Thesis in the spring and graduating in May of 2008.

My advisor is Professor Greg Heileman. For now Professor Tom Caudell is the only other person on my thesis committee, but I'm still working on that.

In addition to partially meeting the ECE department's MS requirements, the Thesis is also indented to meet the requirements of the Computational Science and Engineering Certificate Program, which stipulates that the Thesis contain "a significant computational component".

A Summary of the Basic Workflow

The basic steps of the digital mosaic panorama workflow that will be examined and in some cases improved through parallel processing are as follows:

• Capturing the images as raw images.
• Converting the raw images to tiff, jpg, png, or other format, setting exposure, white balance, etc.
• Pre-processing images with a geometric transformation to correct known lens flaws
• Finding control points. I.e., common points between overlapping images
• Optimizing position, yaw, pitch, and other parameters based on control points, to make the images fit together
• Transformation of images, based on optimum parameters calculated above and a mapping onto a selected projection type.
• Stitching the transformed images together
• Blending the seams of the stitched images

Milestones

It's difficult to make progress on a project as large as this one without writing down a plan with incremental goals. Here's a first cut at some milestones:

Fall 2007

• Get Thesis committee in place
• Capture source images for ~ 100 MegaPixel mosaic
• Capture source images for ~ 500 MegaPixel mosaic
• Get LaTeX template and start Thesis with introduction chapter
• Write workflow chapter without complexity analysis
• Carefully specify build environment and software versions for serial runs
• Conduct timed serial runs and do analysis of which steps take the most processing
• Identify parallel computing resources that can be used for parallel runs
• Analyze time/space complexity of control point finding activity
• Propose parallel version of control point finding activity
• Implement parallel version of control point finding activity

Spring 2008

• Analyze time/space complexity of geometric transformation activity
• Propose parallel version of geometric transformation activity
• Implement parallel version of geometric transformation activity
• Analyze time/space complexity of blending activity
• Propose parallel version of blending activity
• Implement parallel version of blending activity
• Parallel runs
• Finish writing Thesis
• Thesis review and revisions
• Write defense slides / presentation
• Take care of Thesis submittal / graduation details
• Make large prints for defense
• Defense

Here's what they looked like:

The wedding took place in Sauk Center, Minnesota which is the boyhood home of Sinclair Lewis and the inspiration for his "Main Street". The whole general region of central Minnesota is also the inspiration for Garrison Keillor's Lake Wobegon.

They live rurally, where things look pretty much like this:

There's a whole photo-essay book on the region that I highly recommend:

We went fishing for Walleye on Mille Lacs:

There was a rehearsal dinner where a very healthy amount of pork was consumed. The meat was personally butchered by Kayla's uncle Paul shown manning the grill on the right.

More photos of the dinner:

So... new comments can be attached to this post, while you can continue to view old comments on those other posts.

My Grad School Curriculum page has been updated to show:

• I'm taking ECE 547 (Neural Networks) with Thomas Caudell this semester
• I'm taking 3 credits of ECE 599 (MS Thesis) with Greg Heileman this semester
• I'm finishing my thesis next semester (Spring 2008) with three more credits of ECE 599
• I'm hoping to get CS 481 (Operating System Principles) to apply as a credit for my Computational Science and Engineering Certificate Program elective. At least one professor has recommended that this be allowed, so we'll see. If it's not allowed as an elective, I'll try and also take CS 564 - Introduction to Database Management in the Spring 2008 semester, if it's offered.

Important next steps:

• Get my "Program of Studies for Master's Degree" form submitted and approved.
• Get my thesis proposal in a fair state
• Meet with Dr. Heileman regarding the thesis

If everything works out, I'll have my MS is May 2008.

Through a wonderful display of Turing Equivalence, it reimplements The Little Schemer, which teaches recursive thinking via Scheme, in Javascript.

Really, really cool. I'm increasingly interested in languages and the transformations through which they can be made to solve equivalent problems. There's an awful lot down this rabbit hole!

Nothing quite helps in initiating a healthy new habit like not having an alternative!

As of the middle of last month, I'm done waiting, an have placed my bet on Blu-ray, purchasing a Sony BDP-S300. Costco has them for $469.99:

Overall, I'm very pleased. The picture quality is truly awesome, Blu-ray has outsold HD DVD 2-to-1 this year, and Netflix has an ample supply of Blu-ray disks.

Specific entries include: Montana Vacation Picture Essay, Montana sings to me sweetly, Cumbres & Toltec Railroad, Ojo Caliente Spa and Resort, and The Northern New Mexico Little Train Who Could.

I posted a few photos from Montana. Here are two I like: