Intel's compiler is actually one of the best optimizing compilers out there (when it doesn't detect an AMD processor and not bother doing the optimizations...).
It's that to say that some random blogger likely doesn't have the tools to correctly analyze the data
Most of the palaeoclimatology stuff is just curve fitting, they use Matlab (Octave) and R (read the emails), so yeah, some random blogger actually would have the tools that the 'pros' use quite readily available. The silly thing with the paleo stuff is that it is so easily reproducible if you have access to the data, that's why there's a bit of empire building by data hoarding in this community.
(from the wiki): The Augustine Commission also stated that Ares I and Orion would have an estimated recurring cost of almost $1 billion per flight.
(from SpaceX site): 44-49.5M depending on the orbit you want (LEO or GTO), that's the 'out the door' price (not the cost), sure that doesn't include the cost of the payload
The costs are not even in the same order of magnitude, you really think SpaceX's Dragon will add additional recurring costs of $950M? The performance improvement from Falcon 9/Dragon to Ares I/Orion is incremental, certainly not enough to justify the price difference (unless you happen to work at MSFC / Boeing / Lock-Mart).
Yeah, those DoD contracts where he actually (attempted) to put stuff in orbit...what pork! They weren't paying him for power point slides...
Apparently Falcon 1 / SpaceX startup costs are around $450M, which is about what that recent Ares I-X test flight costs. You think there might be a little difference in the overhead of the two operations?
I'm not arguing against the conservation of energy, (yeah lots of energy to get something to LEO), just that there might be a better way.
You don't need decades of experience to have an opinion about the usefulness of giving a select few joy rides into space.
The assumption that any project NASA attempts needs to take deca-years and giga-bucks is part of the problem. Small is beautiful, those decades long development projects you (and the big aero-defense contractors) love are not.
A little 'buck-up' by NASA management is a good thing.
The deviation since 1960 doesn't automatically mean that the records are wrong before 1960, as the instrumental records validate a large chunk of the pre 1960 period tree ring proxy data as correct within a given error bar. Noone knows the reasons why the tree ring proxy data is wrong "recently", but it is entirely possible that the cause is something like "more recent rings on trees take time to dry out" or something like that. It would be interesting to find out the cause.
And no one knows if they were wrong in the past in a similar way; that is the danger of chasing correlations without a firm grasp of the physical mechanisms (which provide model structure and make extrapolation beyond the calibration region somewhat safe).
I doubt that the "Put a guy on a hill and have him look up line of defense" would work.
From the wiki:
"It is believed that the SA-3 crews and spotters were able to locate and track F-117A 82-806 visually, probably with infra-red and night vision systems."
The message is "Fuck you, China; we'll talk as though we're your friends, but we own your airspace and can see every hair on your bare asses, so don't try anything."
Have you ever taken even the most basic of statistics classes?
This is the problem with the whole argument right here. You have people that don't have the first clue what they're talking about telling people how they're supposed to do their job.
Did Wegman and North not have a clue either? Both were pretty critical of the statistical methodologies used by this team of scientists.
That shouldn't be too surprising, most of us making measurements and fitting models would benefit from a consultation from a mathematical statistician. And the standard criticism from statisticians of the physics and engineering community is usually "consult with statisticians more". How you take the criticism says a lot about your professionalism though.
Hi Friend,
These two quotes don't answer all your points, but it does talk to the fact that several of the scientists involved understood the precarious position they were in from withholding data and acting as 'gatekeepers'.
I'd encourage you to get a hold of the emails and read them yourself; the coverage from RealClimate is hardly uninterested.
Cheers.
I've got access to four climate models, from four competing organizations, ranging from middle-school simple to research grade. And they all give about the same results.
'Should we believe model predictions of future climate change':
'Climate projections: Past performance no guarantee of future skill?':
http://www.leif.org/EOS/2009GL038082.pdf
I'd encourage you to check out P.J. Roache's work on model verification / validation. Comparing a bunch of models to each-other is not validation (it's not even verification).
In my field verification = checking that the code is correct, validation = checking that you picked the right governing equations. Your terminology may vary.
In some sense the stability provided by the under relaxation scheme is related to the same CFL condition.
Not at all. You use under-relaxation because you aren't in the 'ball of convergence' for your Newton method, so you don't take the full step. The CFL condition is a condition on stability of the discretization. Two completely different things.
(Recalling from the classic text book in CFD by my guru Dale Anderson with Tannehill and Pletcher.)
TAP is a great comprehensive CFD reference (I still refer to it often), but it is a bit dated, so its coverage of some of the more modern stuff is limited or missing, I'd highly recommend Blazek's book as a supplement. Not as comprehensive, but good coverage of the modern developments.
Yeah, I always get a laugh out of people who think that we're ever going to beat down turbulent flow with higher resolution. It's vortices all the way down,
But at some point you ought to be able to model the 'sub-grid' eddies (homogeneous, isotropic), so you just need a calibrated turbulence model for those.
and no matter how clever your implicit scheme you still have to be able to propogate information through the grid at less than the speed of sound to prevent numerical shock waves from blowing up your solution.
This sounds like word-salad to me, if the scheme is a stable discretization of the governing equations there's no need to worry about propogating info at less than the speed of sound. That's why implicit schemes allow you to take large time-steps, maybe you're thinking of CFL limits on explicit schemes based on cell-size?
Regularization schemes that throw away information are good, but that reduces the value of going to higher resolutions.
Higher orders and higher resolutions still pay-off even with filtering and artificial viscosity.
So I'm doubtful that we'll be predicting the weather, or the climate, with significantly greater accuracy ten years from now than we are today.
Me too. The limit on our predictive ability is probably the resolution and accuracy of our observations of the system rather than the models, the models are already at significantly higher resolution than the measurements, which should give us cause to re-evaluate where we put our money: more simulation or more satellites?
The programming techniques and mathematical formulations needed to take advantage of such very large number of processors continue to be the main stumbling blocks.
True.
The CFL condition that limits the maximum time step one can take shows no sign of relenting. Score has been Courant (the C in CFL) 1, Moore 0 for the last three decades.
There are more suitable methods for stiff, multi-scale problems (implicit time integration, preconditioning, multigrid) that remove those CFL constraints and alleviate the convergence problems (ill-conditioning) with large, high-resolution grids. They may be harder to parallelize, but they make those big problems more tractable. I think most spectral/pseudo-spectral global circulation models (the summary mentions climate modeling) use some sort of implicit time-stepping at least.
I'm curious, by what mechanism would the liquid propellant in the *second stage* have any positive feedback (not damping) to thrust oscillations in the *first stage*?
In case you hadn't noticed - the second stage is attached to the first stage.
Touche, I've clearly been trolled.
The reason longitudinal oscillations (call them what you like) can be really huge in a liquid is because of the positive feedback loop in which a thrust oscillation causes a reinforcing change in the inlet pressure of the pumps, this mechanism does not exist in a solid
That's one mechanism for Pogo. It's not the only one.
Sure there may be another mechanism that causes dynamic instability of longitudinal modes in a rocket, but that would have a different name. There are two published mechanisms for 'pogo', both involve coupling between a liquid propellant feed system and vibrations of the rockets structure.
Actually if you read the last two paragraphs I cut and pasted Rubin describes the two types of pogo oscillations; both involve coupling between the propellant feed and structural oscillations. This only occurs in liquids (or perhaps a large hybrid), but certainly not a solid! How could it be any clearer?
Slashdot Mirror
Intel's compiler is actually one of the best optimizing compilers out there (when it doesn't detect an AMD processor and not bother doing the optimizations...).
I'll second that, here's an example from the Octave mailing lists:
To conclude, on my computer, for this test, Octave is approximately as fast as C, gfortran is a little bit faster and ifort is 10 times as fast.
For scientific computing it's tough to beat ifort on intel iron.
It's that to say that some random blogger likely doesn't have the tools to correctly analyze the data
Most of the palaeoclimatology stuff is just curve fitting, they use Matlab (Octave) and R (read the emails), so yeah, some random blogger actually would have the tools that the 'pros' use quite readily available. The silly thing with the paleo stuff is that it is so easily reproducible if you have access to the data, that's why there's a bit of empire building by data hoarding in this community.
What was the the design failure rate for the shuttle when it was in development? How did that turn out?
On claimed rocket reliability
The costs are not even in the same order of magnitude, you really think SpaceX's Dragon will add additional recurring costs of $950M? The performance improvement from Falcon 9/Dragon to Ares I/Orion is incremental, certainly not enough to justify the price difference (unless you happen to work at MSFC / Boeing / Lock-Mart).
Yeah, those DoD contracts where he actually (attempted) to put stuff in orbit...what pork! They weren't paying him for power point slides...
Apparently Falcon 1 / SpaceX startup costs are around $450M, which is about what that recent Ares I-X test flight costs. You think there might be a little difference in the overhead of the two operations?
I'm not arguing against the conservation of energy, (yeah lots of energy to get something to LEO), just that there might be a better way.
Check your assumptions, that's all Bolden's been asked by his boss to do, you should too.
You don't need decades of experience to have an opinion about the usefulness of giving a select few joy rides into space.
The assumption that any project NASA attempts needs to take deca-years and giga-bucks is part of the problem. Small is beautiful, those decades long development projects you (and the big aero-defense contractors) love are not.
A little 'buck-up' by NASA management is a good thing.
The deviation since 1960 doesn't automatically mean that the records are wrong before 1960, as the instrumental records validate a large chunk of the pre 1960 period tree ring proxy data as correct within a given error bar. Noone knows the reasons why the tree ring proxy data is wrong "recently", but it is entirely possible that the cause is something like "more recent rings on trees take time to dry out" or something like that. It would be interesting to find out the cause.
And no one knows if they were wrong in the past in a similar way; that is the danger of chasing correlations without a firm grasp of the physical mechanisms (which provide model structure and make extrapolation beyond the calibration region somewhat safe).
I doubt that the "Put a guy on a hill and have him look up line of defense" would work.
From the wiki: "It is believed that the SA-3 crews and spotters were able to locate and track F-117A 82-806 visually, probably with infra-red and night vision systems."
...but nothing looks as badass as the F-117.
Really? That's close to the ugliest aircraft ever (except maybe for the Boeing JSF variant). They didn't call it the 'cockroach' for nuthin...
The message is "Fuck you, China; we'll talk as though we're your friends, but we own your airspace and can see every hair on your bare asses, so don't try anything."
FTFY
Have you ever taken even the most basic of statistics classes? This is the problem with the whole argument right here. You have people that don't have the first clue what they're talking about telling people how they're supposed to do their job.
Did Wegman and North not have a clue either? Both were pretty critical of the statistical methodologies used by this team of scientists.
That shouldn't be too surprising, most of us making measurements and fitting models would benefit from a consultation from a mathematical statistician. And the standard criticism from statisticians of the physics and engineering community is usually "consult with statisticians more". How you take the criticism says a lot about your professionalism though.
Hi Friend, These two quotes don't answer all your points, but it does talk to the fact that several of the scientists involved understood the precarious position they were in from withholding data and acting as 'gatekeepers'. I'd encourage you to get a hold of the emails and read them yourself; the coverage from RealClimate is hardly uninterested. Cheers.
I've got access to four climate models, from four competing organizations, ranging from middle-school simple to research grade. And they all give about the same results.
'Should we believe model predictions of future climate change':
http://www.iac.ethz.ch/people/knuttir/papers/knutti08ptrs.pdf
'Climate projections: Past performance no guarantee of future skill?':
http://www.leif.org/EOS/2009GL038082.pdf
I'd encourage you to check out P.J. Roache's work on model verification / validation. Comparing a bunch of models to each-other is not validation (it's not even verification).
In my field verification = checking that the code is correct, validation = checking that you picked the right governing equations. Your terminology may vary.
How many of you have the capacity to deal with hundreds of terabytes of data and run models that take days on a supercomputer?
They aren't talking about running GCMs; they are talking about temperature station and tree ring data; you can process that with R on your laptop.
In some sense the stability provided by the under relaxation scheme is related to the same CFL condition.
Not at all. You use under-relaxation because you aren't in the 'ball of convergence' for your Newton method, so you don't take the full step. The CFL condition is a condition on stability of the discretization. Two completely different things.
Your ideas are intriguing to me and I wish to subscribe to your newsletter.
...you have to calculate the time steps in order.
Foiled again by that damned 2nd Law (shakes fist in air)!
(Recalling from the classic text book in CFD by my guru Dale Anderson with Tannehill and Pletcher.)
TAP is a great comprehensive CFD reference (I still refer to it often), but it is a bit dated, so its coverage of some of the more modern stuff is limited or missing, I'd highly recommend Blazek's book as a supplement. Not as comprehensive, but good coverage of the modern developments.
How many problems can these systems really solve?
Well, only the ones where you need to conserve mass, momentum and energy; pretty niche market really...
Yeah, I always get a laugh out of people who think that we're ever going to beat down turbulent flow with higher resolution. It's vortices all the way down,
But at some point you ought to be able to model the 'sub-grid' eddies (homogeneous, isotropic), so you just need a calibrated turbulence model for those.
and no matter how clever your implicit scheme you still have to be able to propogate information through the grid at less than the speed of sound to prevent numerical shock waves from blowing up your solution.
This sounds like word-salad to me, if the scheme is a stable discretization of the governing equations there's no need to worry about propogating info at less than the speed of sound. That's why implicit schemes allow you to take large time-steps, maybe you're thinking of CFL limits on explicit schemes based on cell-size?
Regularization schemes that throw away information are good, but that reduces the value of going to higher resolutions.
Higher orders and higher resolutions still pay-off even with filtering and artificial viscosity.
So I'm doubtful that we'll be predicting the weather, or the climate, with significantly greater accuracy ten years from now than we are today.
Me too. The limit on our predictive ability is probably the resolution and accuracy of our observations of the system rather than the models, the models are already at significantly higher resolution than the measurements, which should give us cause to re-evaluate where we put our money: more simulation or more satellites?
The programming techniques and mathematical formulations needed to take advantage of such very large number of processors continue to be the main stumbling blocks.
True.
The CFL condition that limits the maximum time step one can take shows no sign of relenting. Score has been Courant (the C in CFL) 1, Moore 0 for the last three decades.
There are more suitable methods for stiff, multi-scale problems (implicit time integration, preconditioning, multigrid) that remove those CFL constraints and alleviate the convergence problems (ill-conditioning) with large, high-resolution grids. They may be harder to parallelize, but they make those big problems more tractable. I think most spectral/pseudo-spectral global circulation models (the summary mentions climate modeling) use some sort of implicit time-stepping at least.
In case you hadn't noticed - the second stage is attached to the first stage.
Touche, I've clearly been trolled.
That's one mechanism for Pogo. It's not the only one.
Sure there may be another mechanism that causes dynamic instability of longitudinal modes in a rocket, but that would have a different name. There are two published mechanisms for 'pogo', both involve coupling between a liquid propellant feed system and vibrations of the rockets structure.
Read up on what 'pogo' refers to:
http://www.aero.org/publications/crosslink/winter2004/05.html
http://www.itea.org/files/2008/2008%20Journal%20Files/September%202008/jite-29-03-03.pdf
http://en.wikipedia.org/wiki/Pogo_oscillation
Actually if you read the last two paragraphs I cut and pasted Rubin describes the two types of pogo oscillations; both involve coupling between the propellant feed and structural oscillations. This only occurs in liquids (or perhaps a large hybrid), but certainly not a solid! How could it be any clearer?
Pogo is any oscillation along the vehicle's longitudinal thrust/flight axis.
Not so, I ripped some figures and a couple paragraphs from a 1970 report on what pogo is and how to prevent it in case you are interested.