a physics question

Just a quick off topic question for RMT. In your opinion, if you use closed source tools during the scientific process does that invalidate your efforts? I know it's no big deal when you are starting out, and calculator isn't a good example but it got me wondering, when you present a paper, is it "acceptable" to use closed source tools or have any papers been rejected? One reason I could think of is without open source software that you compile yourself, you aren't actually repeating the experiment. Have you ever heard of this, or is closed source vs open source never an issue?
Click to expand...

If the tool is used for analysis to support a design, or for non-flying software, it is generally no big deal using a closed-source tool. All the popular Computational Fluid Dynamic (CFD) programs that are used to augment wind tunnel aerodynamic data are accepted by military and FAA customers in aerospace. The ONE AREA where closed-source tools are generally NOT acceptable is when it comes to flight-critical (i.e. flight control, autopilot, and flight instrument systems) and flight-essential (navigation systems) software programs. The FAA will NEVER certify any one of these types of systems to fly on an airplane if it is based on Windows for the simple fact that MS will not give you the source code, AND MS will not submit its OS to the VERY RIGOROUS testing that these types of critical software systems must undergo to achieve FAA certification to be used on commerical airplanes. The reasons that MS would not submit it to testing is (a) they know it would likely fail some tests, and (b) it would not be worth it to them monetarily. So there are companies that make a business out of real-time operating systems and "critical-trusted" code compilers. Wind River and Greenhills are two such companies that provide RTOS and "certified" code compilers that will meet the stringent FAA certification standards.

RMT
 
Thank RMT. Do you know how long they have been "open source"? What did they use in the 80's?

I hope someday that open source becomes a requirement someday for all things related to science & engineering. Without others being able to repeat your results with the hardware and compiler of their own preference, there could be a common bug that appears when people use the same tools.
 
"The apprehension that one is someone else is all that one needs to believe that the world has changed from top to bottom."

Orhan Pamuk

ty rmt, you are right, as hard as that is to admit.
 
oh my god... i just realized that if i learned physics, i would know what all those things in my game editors REALLY mean! :eek:
 
may i ask how you calculate drag coefficient?
Click to expand...
I knew that question was coming sooner or later.
The answer is not at all simple, and touches many other fundamental areas of aerodynamic science. But let me try to give you a short answer:

"You don't calculate it. You measure it in a wind tunnel over a range of Reynolds Numbers." (And before you ask, Reynolds Number is a non-dimensional scaling parameter that relates the inertial, pressure, and viscous forces felt by a body immersed in a flowfield.)

Now, that short answer is not really satisfying, so I can augment it with the beginnings of the longer answer. Actually, you CAN begin to calculate the total drag coefficient for a 3-dimensional wing (not necessarily a whole airplane...yet) if you know the drag coefficient for the 2-dimensional airfoil that makes up the wing. You see, 2-dimensional flow is much simpler than 3-dimensional flow...and the words "wing" and "airfoil" mean very specific things to an aero engineer. The airfoil is a thin section (2-D) of a wing (3-D).

In the world of the 3-dimensional wing there is a phenomenon called "drag due to lift". As Darby is fond of relating, in engineering (as in all reality) "there is no free lunch". Such is true when it comes to a wing producing lift. The more lift you produce, the more drag you produce. This fact of the physics can be represented in an equation for the total drag coefficient on a 3-D wing as follows:

Total CD = (CD due to profile drag) + (CD due to lift production)

The first part is what we measure in a wind tunnel, and it is composed of two pieces itself: The skin friction drag piece and the pressure drag piece. The second part we can actually relate to the square (^2) of the LIFT COEFFICIENT (yep...another coefficient).

Now you can begin to see just how deep this rabbit hole goes, eh?
RMT
 
Do you know how long they have been "open source"? What did they use in the 80's?
Click to expand...
Well, they are really not totally "open source". A company like Wind River still makes their $ by selling and licensing their Real Time Operating Systems on aircraft and other safety-critical software systems. So they still keep the source under wraps. However, they WILL sign a contract with their customers (aerospace integrators, such as my employer) such that they will make the code OPEN TO INSPECTION for the purpose of overall system verification, validation, and FAA certification. An even easier (and cheaper) solution is when a specific, configured version of the RTOS itself has already been certified by the FAA, then for an aerospace integrator to simply use that version, the FAA will accept it without further verification because it has already gone thru the process. In this case, only the rest of the system (outside the RTOS) software needs to undergo certification verification testing.

Back in the 80s, any aerospace company (or their electronics vendor) basically had to write their own kernel and RTOS for the Intel, TI, or Motorola processors that were commonly used in that day and age. It was the day and age of "roll your own RTOS". Those "roll your own" solutions eventually evolved into the standards and "off the shelf RTOS" products we have today.

I hope someday that open source becomes a requirement someday for all things related to science & engineering. Without others being able to repeat your results with the hardware and compiler of their own preference, there could be a common bug that appears when people use the same tools.
Click to expand...
This is true, but you do not need to reproduce form the source code level to reproduce the results. If you run the binary version of someone's RTOS in the target processor, you can still reproduce any anomalies encountered... and then the RTOS vendor is on-the-hook to fix any problem identified as a safety-critical error. The problem with "true open source" in aerospace is that it would be even MORE troublesome to the bottom line (profit) than it would be to a company like Microsoft. The reason is that once you developed a novel approach, that may make it easier and cheaper for you to bring your airplane to market, all of your competitors would immediately be able to use it as well (because it is totally OPEN SOURCE). When you compare the number of units sold of airplanes vs. the number of units sold for PCs and game/productivity software, you can see that every single dollar saved against your competitor translates to large impacts on the eventual price of your airplane. PC and game vendors recoup their costs over tens of thousands (or even MILLIONS) of sales. In the airplane business, if you sell 1000 of any given airplane design, you have done EXCEEDINGLY well. That means you must recoup your development costs over fewer unit sales.

RMT
 
A company like Wind River still makes their $ by selling and licensing their Real Time Operating Systems on aircraft and other safety-critical software systems.
Click to expand...

I'd say their Linux stuff (which is required to be open source) isn't as popular as the vxworks stuff because people are heavily invested into vxworks. Linux wasn't around in the 80's so it has a lot of catching up to do. As for which division makes more money, Linux is free, they don't have to pay any royalties. There is more profit in open source.


but you do not need to reproduce form the source code level to reproduce the results. If you run the binary version of someone's RTOS in the target processor, you can still reproduce any anomalies encountered...
Click to expand...

But there could be a bug in common to the compiler that produces those binaries, or a bug common to the hardware that runs the binaries that could make the results of a "safety check" return "PASS" all the time. Even binaries made by different people could use the same "math library" from a "closed source 3rd party". If that math library has an undiscovered bug, everyone would see the same results. How would you know there was an error in the first place. If you don't let people compile source code with their own compiler and run it on different hardware, and make your "order of operations" completely transparent via open source tools, you haven't made every effort to ensure you are letting people reproduce the "intended" math.

Matlab for example, it's closed source, and I say it's days are numbered as a closed source product because people shouldn't be using the same binaries on the same hardware to reproduce results. Compilers can have bugs for a long time before they are discovered. CPU's have been sold to consumers with problems on their silicon that only appeared after they were used by many people. If there is an undiscovered bug in brand x model y processors everyone will see the same bug. You don't want that bug to make it look like everything in a critical system is A-OK when it's not.

http://www.ibm.com/developerworks/linux/library/l-oslab/

The FOSS movement has so much MOMENTUM and gaining, that it might even be illegal to make software available to the public without releasing the source code.


By 2012. MOOOOHAHAHAHHA
 
Total CD = (CD due to profile drag) + (CD due to lift production)

The first part is what we measure in a wind tunnel, and it is composed of two pieces itself: The skin friction drag piece and the pressure drag piece. The second part we can actually relate to the square (^2) of the LIFT COEFFICIENT (yep...another coefficient).
Click to expand...

This approach to a complex problem is similar to orbital mechanics (yet another aerospace engineering problem).

If we have a situation where one planet is orbiting a single star we can apply an analytic approach and plug in Kepler's Laws to very accurately predict where the planet will be in its orbit over periods of tens of thousands of years. This corresponds to the second part of Ray's CD problem.

If the Earth was the only body in the Solar System other than the Sun the analytic approach would suffice. But the real Solar System has many massive bodies other than the Earth orbiting the Sun. If we are to accurately calculate the position of the Earth (say within a few seconds of arc) over 100 years we have to account for the mass, relative movement, position and proximity of the Moon, Saturn and Jupiter (we can basically ignore the other planets, asteroids and comets due to their tiny masses WRT their distances from the Earth). We don't have a simple analytic math equation to accomplish this. We have to use numerical analysis. We plug in the values for the variables for the Earth at time t_0 and position x_0 and let it run for a period of time to t_1 and x_1. Then we shoot the new position of the bodies with a telescope of some sort and recalculate. That's the equivalent of the wind tunnel. The second half of the situation requires a physical experiment because the situation is extremely complex - and beyond our current math abilities (if we desire to do a straight analytic approach).

Michio Kaku refers to this as 21st Century physics confounded by 20th Century math. We know what we want and we have a good clue about what the answer should be but we just haven't figured out the proper math to accomplish the task.
 
i find this post very, very interesting. so does this mean that physics and math are...*gasp* incomplete?

if so, what is the most important thing in math and physics that needs to be figured out?
 
if so, what is the most important thing in math and physics that needs to be figured out?
Click to expand...

For me, I want to know if I should abandon all hope that local realism is supported by our universe.
 
so does this mean that physics and math are...*gasp* incomplete?

if so, what is the most important thing in math and physics that needs to be figured out?
Click to expand...

Ruthless,

Yes. Physics is "incomplete". All scientific theories, whether its physics,biology, chemistry, you-name-it describes an approximation of "reality" that is valid only in the limit of the domain that it covers. Newtonian physics is valid in the domain of relatively minor velocities; Special Relativity is valid in the limit of no gravitational fields (or more generally no accelerating forces); General Relativity is valid in the limit above the sub-atomic scale; quantum physics is valid in the limit at and below the sub-atomic scale. Each is extremely accurate within the limit of its domain. None, however, is a completely accurate description of the universe as a whole. Even though we've known for 102 years (and suspected it for about 150 years) that Newtonian mechanics is wrong it is still so accurate in our everyday existence that we use it to calculate just about everything from calculating the ballistic flight of a cannon ball to sending a space craft to Mars because the speed of light is so fast that we can normally ignore the actual velocity and just assume that it is infinite. Its not correct but its close enough most of the time.

Think of a billiard table. Put one ball on the table give it a shove and have it start bouncing off the rails. You should have no real problem predicting the path that the ball will follow. If you have sufficient information about the ball, table and the momentum that you gave the ball you should even be able to accurately predict where the ball will stop.

Now start adding balls by having people standing around the table and randomly shoving them toward the rails. By the time that you get just three balls going you'll lose any ability to accurately predict the complete paths once they begin to collide with each other.

Next, move off the table and into space in the Solar system. Instead of three billiard balls you have nine planets,a star, a couple dozen major moons, hundreds of minor moons and thousands of asteroids and comets. It's far, far too complex a system to accurately predict the future evolution of their positions over an arbitrary period of time by simply plugging the variables into a calculus equation and chugging out a solution.

That's the entire basis of thermodynamics. Thermodynamics says nothing at all about individual atoms or molecules in a system that includes billions of such particles. It makes predictions based on statistical tendencies for a given volume and density of a (gas, liquid, radiation, etc.).

What's the most important discovery to be made? Probably how to integrate General Relativity and quantum physics. Each, within its domain of accuracy, is known to be "correct". But they each fail to predict each other. Obviously there's something missing - and its not a simple problem to fix. A grand unification theory has to include something akin to quantum gravity. Einstein tried for the last 50 years of his career. Others have been trying for the 50 years since his death. So far the answer hasn't been discovered.
 
Back
Top