DSA Notes, 10/28/09
Beaudry
Progress on Deliverables
Beau reported difficulty with decision-making of agents in a particular situation (farmers trying to decide whether to buy more quota or pay a fine), such that it no longer makes sense to make the decision based on the instantaneous cost-benefit ratio, but rather requires looking into the future for distant opportunity costs and benefits. Ben suggested swallowing programmer’s pride and hacking a work-around that is not general, thus not elegant, but deals with this particular exception to the general way of making decisions. In the spirit of the spiral, the workaround could be the a first pass and the full implementation of forward-thinking decision making could be the next pass (time permitting).
In the writing support group, Beau realized he has learnt something after all—the others did not seem to understand that hypotheses need to be tested incrementally in modeling, and that’s something you don’t understand unless you’ve worked through it yourself.
Debugging: using the demo version of the debugger improved things tremendously. Previously, Beau had to write output statements for everything he wanted to check in the code—now it’s automatic. Has sped up progress substantially. Got done what he wanted to get done, fixed a lot of bugs that wouldn’t otherwise have been discovered. Excellent!
OmniPlan: Beau presented his plan for the last phase of the PhD—very impressive. Has scheduled first draft of the dissertation to be submitted to the committee by the end of February, which leaves March and April for back-and-forth iterations of the thesis with the committee; defense to be held some time in May (in theory). Looks good!
One worry: currently not running model with allocations from CSU, but canned allocations. But this can be addressed in the calibration stage.
Deliverables
- Write CSU presentation. Re-do linked OmniGraffle flowcharts explaining structure of the model.
- Report on result of CSU meeting—what things they suggested can be reasonably incorporated/modified in the model? (In Denver all next week—at CSU Tuesday, U Denver Wednesday-Thursday). Report via blog post from Denver.
Ben
Progress on Deliverables
- Getting to grips with R: good progress on this, probably ahead of schedule. Looks promising, particularly relative to MATLAB. Worked through first 2 tutorials. Started to realize that it was teaching more basic syntax which Ben already knew. So figured out he’d just start doing stuff and then work out what syntax was needed.
- Planning: completed three plans, but hasn’t got around to translating each OmniPlan tasks to OmniFocus projects.
- Ohio samples: Ben has his own lab coat now! = real scientist. Donned it (excellent!), and found that the material is not quite dissolved yet. Topped up with acid (may not be necessary). Teachable moment: particular samples he has may not be suitable, too much iron or organic matter, heavy dolomitization. Andy seemed to think that the Australia samples might have radiolarians. Could lead to another chapter.
Deliverables
- Start to dissolve Australia samples (not more than a couple of hours work).
- Complete translation of OmniPlan tasks to OmniFocus projects.
- Diversity project: continue working on R. Have a first look through Dan’s code.
- Radiolarian project: start making a list of samples to request.
- Morphospace project: find ODP database. Start making a list of samples to ask for. Also write back to Zoe and ask for samples (also whether she has residues or prepared samples, etc)
- previous:
- OmniPlanning Ahead
- next:
- Where The Time Is Going
Beaudry Kock
October 28, 2009 @ 11:42 am
A few reflections on Ben’s report: I’m really impressed at the systematic, thoughtful way you’ve gone about the planning process. One of the reasons I enjoy interacting with you on the meta-PhD stuff is that you frequently take angles on things that I just would never have thought of myself. For example, organizing your OmniPlan calendar so that you make it sync with your iCal, and so more in keeping with the actual time you allocate to various projects – so simple, yet such a great idea! But most impressive has been the business-like way you’ve gone about it. Not that you don’t normally approach things with due seriousness, but I think this round of planning has a much heavier dose of realism built in. All your tasks seem grounded in (perhaps bitter) experience, which suggests to me you stand a good chance of getting what you want to get done, done.
It’s challenging to conduct three mini- (or not so mini) projects simultaneously, and so you should be proud of managing to pull that off. Regardless of whether or not you actually work on all three projects in a given week, the fact that you’re marshaling the projects in your head and in OmniPlan, and keeping intellectual tabs on it all – that’s really quite something.
OK, ’nuff rambling. My general impression of your demeanor and your concrete achievements since restarting things post-Australia: you’re focused, you’re motivated, and you’ve done everything organizationally that you needed to do. You’ve even started work. Keep on keepin’ on!
kotrc
October 28, 2009 @ 3:05 pm
Well, it’s going to start sounding like a broken record, and it’s no different from what I said to you in person today: I’m impressed with both your level of effort and your progress. Putting in late nights every day is energy-sapping, and you’re to be admired for being so persistent in putting in the hours it takes to get the work done. In your day-to-day frustrations with hurdles in the code, don’t forget that you are more than on-track in the big-picture work plan. You’ve got a working model to show the folks at CSU, which is far more that you’d hoped for a couple of months back, and you have a realistic plan (with plenty of time for the inevitable revisions your committee will demand) for getting your thesis done. As ever, I am inspired by your work ethic and your capacity to GRD. Just keep at it.