Quote:
Originally Posted by Ken G
Probably not. Each subdiscipline has its own culture, and the areas I know best are stellar physics. Still, I'd be surprised if the culture in accretion-disk simulations is so much different. The equation I often see is "observation + theoretical simulation that agrees with observation = completed observational/theoretical synergy, next problem." We can leave the "what happened" issues up to the textbooks for students, and a cartoon level explanation should suffice in many cases.
|
Maybe one difference is that no one believes that any simulations are good enough to capture everything important, yet (well, if they are honest they wouldn't) ... the physics is just sooo darn complex, and by necessity the simulations, models, etc are just too obviously restrictive to be 'good enough'; For example: 'relativistic MHD in strong field conditions' is something no one would dare claim they could model!
Aside from that, there's the rapid pace of improvements/extensions in observations - compare 2MASS to UKIDSS, for example, or Compton to GLAST - so whatever goodness of fit you have today, in a year or ten there'll be a dozen kinds of new observation that you can apply. Of course, this applies to a great many areas in astronomy, perhaps most.
IIRC, there was a bit of a mini-crisis in stellar atmospheres some time ago; when good data from new wavebands/windows (I think it was the UV, but it may have been the NIR or FIR) became available, 'the textbook' on stellar atmospheres had to be rewritten (I'm exaggerating, of course), because the models gave the wrong answers. Back then, if I read your comments correctly, there were few 'kitchen sink' models, so it was a different kind of shortcoming than what you're describing here, but at some level I think the principle is the same (IIRC, the resolution involved first gaining a deeper understanding of what physical processes were
actually involved in photospheres, chromospheres, etc).
Quote:
Now, one can certainly file this under the heading of "pet peeves" and note that this description is not always applicable. Whether it is mostly applicable or mostly unfair is also not so obvious, and that question is likely where our disagreement centers. The more conservative way to sum it all up would be to say "let's be on guard against this phenomenon, in case it should rear its head", and paint me as someone akin to the child in the story "the Emperor's New Clothes". It is not my intention to take to task the whole astrophysical community, but rather to point out that sometimes the most fertile possible soil for discovery is found after one gets simulations that agree with observations, not necessarily while the observations remain mysterious-- even though the latter is where much of the attention gets focused while the former often gets relegated to a kind of "niche" market.
|
Well that's something I agree with ... "completed" simulations/models are great for pushing research forward! If you skim astro-ph regularly, you'll find a steady trickle of preprints that seem to have originated in someone looking over some 'case closed' files (kitchen sink simulations or otherwise), taking a fresh look, and finding something most curious. Not many such of course, but certainly non-zero.
Quote:
I'm not saying the kitchen sink models are weak, indeed I think they must be very strong that they make so many correct predictions. The problem lies in what was done with them once they succeeded in making those predictions. The two main issues that then cropped up were:
1) Since people did not really analyze why they got the results they did, it was not clear how much flexibility existed to fit systematic changes in the observations. There is a tendency to think a theory should fit "the observations", but there really isn't any such animal. The observers don't realize all the optional tweaks the theorists had at their disposal to try and get that agreement, and the theorists don't realize the potential for systematic errors in the observations. (There's an old joke that the observer is the only one who doesn't believe her observations, and the theorist is the only one who does. But I think the first part is actually the one that applies in both cases.)
So what you find is, some new observational effect is interpreted differently and all the datapoints shift, and sometimes this makes it "in better agreement" with the theory, as if that meant anything, and everyone is happy, or sometimes it makes it in worse agreement, and people get all bothered. But how can this be a source of either contentment or concern until we know what the observational systematics that are still out there might do, or how many free twiddles are still available for the theory? It's kind of a shell game as long as we are dealing with black boxes instead of unifying principles.
|
And that's one place where I misunderstood your point, earlier: 'observations' today are what pops out the ends of extremely long chains of logic, physics, modelling, etc, etc, etc; the existence of unrecognised systematics is very real, and one frustration I sometimes have is with papers that (to me) clearly do not treat these seriously enough. For example, somewhere in some chain or other may be an indirect link to one of your highly successful 'kitchen sink' simulations, a link that no one paid much attention to, and an overlooked link that may come back to bite us all.
The history of the distance ladder, out to z ~0.2, is an excellent example, except that it predates kitchen sink simulations (similar principle though?).
Quote:
2) The answers people arrived at to explain "why" the kitchen sink simulations found that massive stars were more luminous varied from being quite incomplete to being demonstrably false (look for any mention of high pressure or strong gravity in hot stars, or any mention that the elevated temperature causes the luminosity to be higher because of the temperature sensitivity of nuclear burning rates). Yet they are propagated in authoritative places willy nilly, particularly on the web and in the minds of professionals who work with these stars. How can this be? These people are no fools, so it can only be that they never really tried that hard-- they never really cared to know the real explanation that actually happened in the simulation, as long as the cartoon explanation seemed to work.
That's just like people who think the phases of the Moon are caused by the shadow of the Earth-- when a cartoon works because it's never subjected to careful scrutiny, it can flourish and propagate despite being anathema to the goals of science. I'm saying we remain closer to that pitfall than we may realize, whenever we fail to subject simulations to that level of scrutiny, settling instead for cartoon descriptions that "seem to work" on the grounds that what "really matters" is that the simulation got it right.
|
And that's why, thank the FSM, we are blessed with Feynmans and Bahcalls and {insert your fave 'asks, innocently, really simple but extremely deep questions' here}, as well as Davises, Hulses, ...