Older blog entries for BrentN (starting at number 9)

Long time since my last update. I've been "heads down" at work for the past month. Sadly, I don't feel like I have much to show for it. I have a Langevin dynamics scheme implemented now. God(dess) only knows if its working correctly. It appears that "working" is a platform dependent thing. Yes, on Compaq Alphas - No, on IBM Power3's.

Been giving a lot of thought to doing an analysis of the employer/employee relationship in terms of a social contract. The inspiration for this was my reading of Death March, Edward Yourdon and Your Money or Your Life, by Joe Dominguez.

Still, its probably a waste of time.

I'll probably continue to be "heads down" at least for another 2-3 months. Unfortunately, this means no hacking on Finance::Quote for some time. Sorry, Paul.

Ahhh. I'm back from the image analysis short course and a short vacation. Now, I must figure out why my indenter code is crashy-crashy on multiple processors. I am definitely not feeling any MPI love right now.

Paul did some nice work on Finance::Quote, making it more modular. I don't think this has made release yet, but if you're following this package, check it out from the CVS tree at sourceforge.

Helped a co-worker install LinuxPPC 2k on his G4 today. As expected, BootX failed miserably. I got yaboot up and running for him, and he appears to be happy.

But then, I'd be happy with a 500Mhz G4 too. :)

Work

Finishing up a presentation on O(N) methods for density functional calculations. Exams just around the corner...

I'll be heading to Raleigh, NC in a week or so to make my annual showing at a short course on image analysis and computer-aided microscopy. It'll be a nice change of pace.

Hacked an Open Firmware script to get "nice" dual-boot capabilities on my Yosemite G3. It works. Updated my online notes on configuring yaboot for SCSI hds.

Finished up the error-checking for the tiaacref routine in Finance::Quote

Finance::Quote
Huzzah! Huzzah!
After much tinkering, I finally got CVS working with SourceForge.

LinuxPPC 2k
Still tinkering with yaboot. In the interest of having something that actually works, I'm trying again to get BootX to function properly. I'm not sanguine about my prospects though, considering the difficulties some people have had with BootX and OS 9.

Important tip to you B&W G3 owners trying to install LinuxPPX2k:
If you have a SCSI CD in addition to the IDE device that comes with the unit, the installer may not handle the setup of your /etc/fstab properly. I saw this with an old CD-R attached to mine.

Work
Hardness value of Si3N4 too high by a factor of 4. I think its a data analysis issue, not a code issue. (whew)

yaboot considered harmful

I spent most of Saturday playing with LinuxPPC 2k. In particular, I spent most of that time playing with yaboot, the next generation boot loader for LinuxPPC.

It sucks. Hard.

Actually, to be fair, yaboot itself is fairly clever, albeit not without its flaws. The instructions provided by LinuxPPC.com however do not in any way, shape, form, or fashion reflect what is actually necessary to get yaboot to function properly. After a great deal of searching on the linuxppc mailing lists, I found myself in possession of a pile of unorganized information, some of it contradictory. If you are in the same boat I was in, you might be interested in this text file, which contains some of the more useful tidbits.

At some point, I plan on organizing this better, since there is precious little online help for yaboot. But in interest of contributing to the community, here is the distillation of what I discovered. This document contains my working notes on getting LinuxPPC working on SCSI drives.

L-J 12-6 systems

This is actually a note for macricht, since I am too lazy to find his email. Two things you may want to check for in your broken L-J code, if you haven't already are (1) the starting positions of your atoms and (2) the signs on the terms in the force calculation. If two of your atoms are too close, the force between them becomes large, quickly. Since there is no damping, this plays hob with the energy of the system. Also, if you botch the signs when you take the derivative of the L-J energy expression, and both components become effectively repulsive, that can have untoward effects on the system energy. I made that mistake when I first did L-J. Good luck!

Grumble, grumble. The network connection to the machine room here is no-joy. Facilities Services was rewiring for the PC cluster - they must have botched something. I am trying to justify not going to the console to continue work...

This fellow is studying MD too! Cool stuff. Lennard-Jones systems are good toy problems - the interaction is simple enough so that you can run a fair number of atoms on modest equipment, but robust enough that you can learn all about the various statistical correlations that are important in MD.

Of course, the only real systems that L-J models well are noble gases, but hey - you can't have it all.

My life is now complete. I got moderator on Slashdot. Whoopee.

In other news, I have a "go" from the bossman to scale my silicon nitride sim up to multiple processors. Now, I'll see if I actually know jack about parallel computation or not.

pjf has informed me that my TIAA-CREF mod was rolled into GnuCash. I'm kinda tickled about that.

I saw this morning that my friend data has joined up. Cool. I'm looking forward to seeing his open source sequencer, BINARS, released. He's been putting a lot of time into it. Check out his home page for more info.

I'm considering putting some effort into GnuCash. I am currently a Quicken "poweruser", but I'd prefer to use something that wasn't quite as proprietary. Intuit doesn't even provide cross-compatibility between their Mac and their Windows versions of Quicken! I'd love to see GnuCash running on MacOS X - since the underlying kernel is BSD, the engine ought to compile without any problems - then it'd just be a matter of writing a Cocoa-based frontend.

My research is progressing. I have nice pretty pictures of my first nanoindentation simulation. I am fairly sure the indenter code is working "as advertised" on a serial job. Now I have to convince the boss of this, so I can test it on a parallel job.

Contributed a modification to Finance::Quote

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!