Home |
Search |
Today's Posts |
|
#1
|
|||
|
|||
Computer model experiment
On May 11, 1:38*pm, Jim Lux wrote:
Ralph Mowery wrote: "tom" wrote in message et... On 5/10/2010 3:12 PM, wrote: As Clint said in the wonderful old movie, "A man's gotta know his limits". For antenna modelers it should read, "A man's gotta know the program's limits". Of course, Art thinks things have changed and the computer modelers have a better grasp upon reality than the ones even he calls "the masters". He is an example of the blind man leading himself. tom K0TAR The computer program should know its limits. yes and no. *For EM modeling codes originally intended for use by sophisticated users with a knowledge of the limitations of numerical analysis, they might assume the user knows enough to formulate models that are "well conditioned", or how to experiment to determine this. NEC is the leading example here. It doesn't do much checking of the inputs, and assumes *you know what you are doing. There were modeling articles in ARRL pubs 20 years ago that described one way to do this at a simple level: changing the number of segments in the model and seeing if the results change. *The "average gain test" is another way. In many cases, the constraints on the model are not simply representable (a lot of "it depends"), so that raises an issue for a "design rule checker" that is reasonably robust. *Some products that use NEC as the backend put a checker on the front (4nec2, for instance, warns you about length/diameter ratios, almost intersections, and the like) It's sort of like power tools vs hand tools. *The assumption is that the user of the power tool knows how to use it. * Anytine a program allows the data entered to be too large or small for the calculations, it should be flagged as being out of range. *Also many computer programs will use simplified formulars that can mast the true outcome. *Usually it is not very much, but as all errors start to add up the end results may be way off. There's whole books written on this for NEC. *Part I of the NEC documents, in particular, discusses this. *There's also a huge professional literature on various FEM computational techniques and their limitations. *NEC, like most numerical codes (for mechanics, thermal, as well as EM), is very much a chainsaw without safety guards. * It's up to the user to wear gloves and goggles and not cut their leg off. Jim Lux of NASA no less! All of the programs clearly state that they are based on Maxwells equations. The bottom line of that equation is that for accountability for all forces involved are required and where the summation of all equals zero. This is nothing new and has been followed thru for centuries. The equations requires first and formost equilibrium and what the program supplies is easily checked that it meets these requirements. It is very simple. Showing that the solution is that inside an arbitrary boundary all within as with the whole must be resonant and in equilibrium.It requires no more than that to show if the program has achieved its object. I understand your preachings but you presented no point that can be discussed. Now you will respond that I must do such and such to back the statement above despite that those requirements are the basis of physics. So to you I will supply the same that I have supplied to others which they reject, no one has stated why. A arbitrary gaussian border containing static particles ( not waves as many summize. Gauss was very clear about the presence of static particles) in equilibrium may be made dynamic by the addition of a time varying field such that Maxell's equations can be applied to solve.I have stated the over checks that can be applied to provide correctness of this procedure. You may, of course, join the poll that swells on behalf of NASA in opposition to the above but it would provide me a great deal of delight if you provided more than to just say "I am wrong". Nobody as yet provided one mathematical reason that disputes the above, so in the absence of such you will not be alone, only your credibility suffers but you will remain in the majority of the poll in the eyes of the ham radio World. Regards Art Unwin |
#2
|
|||
|
|||
Computer model experiment
Art Unwin wrote:
On May 11, 1:38 pm, Jim Lux wrote: The computer program should know its limits. yes and no. For EM modeling codes originally intended for use by sophisticated users with a knowledge of the limitations of numerical analysis, they might assume the user knows enough to formulate models that are "well conditioned", or how to experiment to determine this. NEC is the leading example here. It doesn't do much checking of the inputs, and assumes you know what you are doing. Jim Lux of NASA no less! Speaking, however, as Jim Lux, engineer, not necessarily on NASA's behalf. All of the programs clearly state that they are based on Maxwells equations. snip I understand your preachings but you presented no point that can be discussed. While NEC and its ilk are clearly based on Maxwell's equations, one should realize that they do not provide an analytical closed form solution, but, rather, are numerical approximations, and are subject to all the limitations inherent in that. They solve for the currents by the method of moments, which is but one way to find a solution, and one that happens to work quite well with things made of wires. Within the limits of computational precision, for simple cases, where analytical solutions are known to exist, the results of NEC and the analytical solution are identical. That's what validation of the code is all about. Further, where there is no analytical solution available, measured data on an actual antenna matches that predicted by the model, within experimental uncertainty. In both of the above situations, the validation has been done many times, by many people, other than the original authors of the software, so NEC fits in the category of "high quality validated modeling tools". This does not mean, however, that just because NEC is based on Maxwell's equations that you can take anything that is solvable with Maxwell and it will be equally solvable in NEC. I suspect that one could take the NEC algorithms, and implement a modeling code for, say, a dipole, using an arbitrary precision math package and get results that are accurate to any desired degree. This would be a lot of work. It's unclear that this would be useful, except perhaps as an extraordinary proof for an extraordinary claim (e.g. a magic antenna that "can't be modeled in NEC"). However, once you've done all that software development, you'd need independent verification that you correctly implemented it. This is where a lot of the newer modeling codes come from (e.g. FDTD): they are designed to model things that a method of moments code can't do effectively. |
#3
|
|||
|
|||
Computer model experiment
On May 11, 4:02*pm, Jim Lux wrote:
Art Unwin wrote: On May 11, 1:38 pm, Jim Lux wrote: The computer program should know its limits. yes and no. *For EM modeling codes originally intended for use by sophisticated users with a knowledge of the limitations of numerical analysis, they might assume the user knows enough to formulate models that are "well conditioned", or how to experiment to determine this. NEC is the leading example here. It doesn't do much checking of the inputs, and assumes *you know what you are doing. Jim Lux of NASA no less! Speaking, however, as Jim Lux, engineer, not necessarily on NASA's behalf.. All of the programs clearly state that they are based on Maxwells equations. snip I understand your preachings but you presented no point that can be discussed. While NEC and its ilk are clearly based on Maxwell's equations, one should realize that they do not provide an analytical closed form solution, but, rather, are numerical approximations, and are subject to all the limitations inherent in that. *They solve for the currents by the method of moments, which is but one way to find a solution, and one that happens to work quite well with things made of wires. Within the limits of computational precision, for simple cases, where analytical solutions are known to exist, the results of NEC and the analytical solution are identical. *That's what validation of the code is all about. Further, where there is no analytical solution available, measured data on an actual antenna matches that predicted by the model, within experimental uncertainty. In both of the above situations, the validation has been done many times, by many people, other than the original authors of the software, so NEC fits in the category of "high quality validated modeling tools". This does not mean, however, that just because NEC is based on Maxwell's equations that you can take anything that is solvable with Maxwell and it will be equally solvable in NEC. I suspect that one could take the NEC algorithms, and implement a modeling code for, say, a dipole, using an arbitrary precision math package and get results that are accurate to any desired degree. *This would be a lot of work. It's unclear that this would be useful, except perhaps as an extraordinary proof for an extraordinary claim (e.g. a magic antenna that "can't be modeled in NEC"). *However, once you've done all that software development, you'd need independent verification that you correctly implemented it. This is where a lot of the newer modeling codes come from (e.g. FDTD): they are designed to model things that a method of moments code can't do effectively. Again you preach but obviously you are not qualified to address the issue. Maxwells equations are such that all forces are accounted for when the array is in a state of equilibrium. To use such an equation for an array that is not in equilibrium requires additional input ( proximetry equations) which is where error creep in.When an array is in equilibrium then Maxwell's equations are exact. The proof of the pudding is that the resulting array is in equilibrium as is its parts. AO pro by Beasley consistently produces an array in equilibrium when the optimizer is used as well as including the presence of particles dictated by Gauss., The program is of Minninec foundation which obviously does not require the patch work aproach that NEC has. On top of all that. it sees an element as one in encapsulation as forseen by Gauss by removing the resistance of the element, which produces a loss, and thus allows dealing only with all vectors as they deal with propagation. It is only because hams use Maxwell's equation for occasions that equilibrium does not exist, such as the yagi, do errors start to creep in. Any array produced solely by the use of Maxwell's equations provides proof of association by producing an array in equilibrium which can be seen as an over check.Like you, I speak only as an engineer on behalf of myself. Clearly, Maxwell had taken advantage of the presence of particles when he added displacement current so that the principle of equilibrium would be adhered to. This being exactly the same that Faraday did when explaining the transference from a particle to a time varying current when describing the workings of the cage. Regards Art |
#4
|
|||
|
|||
Computer model experiment
On May 11, 8:30*pm, Art Unwin wrote:
When an array is in equilibrium then Maxwell's equations are exact. maxwell's equations are ALWAYS exact, it is digital models that are inexact and have limitations due to the approximations made and the numeric representations used. |
#5
|
|||
|
|||
Computer model experiment
On May 12, 12:10*pm, K1TTT wrote:
On May 11, 8:30*pm, Art Unwin wrote: When an array is in equilibrium then Maxwell's equations are exact. maxwell's equations are ALWAYS exact, it is digital models that are inexact and have limitations due to the approximations made and the numeric representations used. On this I have total agreement. The moment one strays from the concept of equilibrium is when we expose ourselves to errors. Regards Art |
#6
|
|||
|
|||
Computer model experiment
On May 12, 3:29*pm, Art Unwin wrote:
On May 12, 12:10*pm, K1TTT wrote: On May 11, 8:30*pm, Art Unwin wrote: When an array is in equilibrium then Maxwell's equations are exact. maxwell's equations are ALWAYS exact, it is digital models that are inexact and have limitations due to the approximations made and the numeric representations used. On this I have total agreement. The moment one strays from the concept of equilibrium is when we expose ourselves to errors. Regards Art ok, so you DO agree that maxwell's equations that make no mention of particles like neutrinos, gravity, coriolis forces, or levitation ARE correct! And therefor you must agree that the representation of gauss's law encapsulated in maxwell's equations, WITHOUT an explicit t in it must be correct! You must also be admitting that your optimization experiments are full of errors. wow, now its time to go and rejoice, art has finally come around to the real world! |
#7
|
|||
|
|||
Computer model experiment
K1TTT wrote:
On May 11, 8:30 pm, Art Unwin wrote: When an array is in equilibrium then Maxwell's equations are exact. maxwell's equations are ALWAYS exact, it is digital models that are inexact and have limitations due to the approximations made and the numeric representations used. Inexactness of the solution isn't because the method is digital. The field equations solved by the digital methods simply can't be solved by other methods, except for a relatively few very simple cases. Many non-digital methods were developed over the years before high speed computers to arrive at various approximate solutions, but all have shortcomings. For example, I have a thick file of papers devoted to the apparently simple problem of finding the input impedance of a dipole of arbitrary length and diameter. Even that can't be solved in closed form. Solution by digital methods is vastly superior, and is capable of giving much more accurate results, than solution by any known method. Roy Lewallen, W7EL |
#8
|
|||
|
|||
Computer model experiment
On May 13, 8:56*am, Roy Lewallen wrote:
K1TTT wrote: On May 11, 8:30 pm, Art Unwin wrote: When an array is in equilibrium then Maxwell's equations are exact. maxwell's equations are ALWAYS exact, it is digital models that are inexact and have limitations due to the approximations made and the numeric representations used. Inexactness of the solution isn't because the method is digital. The field equations solved by the digital methods simply can't be solved by other methods, except for a relatively few very simple cases. Many non-digital methods were developed over the years before high speed computers to arrive at various approximate solutions, but all have shortcomings. For example, I have a thick file of papers devoted to the apparently simple problem of finding the input impedance of a dipole of arbitrary length and diameter. Even that can't be solved in closed form. Solution by digital methods is vastly superior, and is capable of giving much more accurate results, than solution by any known method. Roy Lewallen, W7EL quantization of every number in a numeric simulation is but one of the contributions to inaccuracy. the limitations of the physical model is another, every modeling program i know of breaks the physical thing being modeled into small pieces, some with fixed sizes, some use adaptive methods, but then they all calculate using those small pieces as if they were a single homogonous piece with step changes at the edges... that also adds to inaccuracies. the robustness of the algorithm and the residual errors created are a bit part of getting more accurate results. There is no doubt that numerical methods have allowed 'solutions' of many problems that would be extremely difficult to find closed form solutions for, but they must always be examined for the acceptibility of the unavoidable errors in the method used. other non-digital methods also have their limitations. unless you are using the original differential or integral equations and satisfying all the boundary conditions, your method will introduce errors. weather that means you represent an odd shaped solid object by a sphere, or make other geometic replacements that give you simpler field configurations, you have introduced errors at some level. you must of course judge these methods by the same way to determine of the errors introduced by the simplyfied geometry or other methods used are acceptible for the problem at hand. |
#9
|
|||
|
|||
Computer model experiment
K1TTT wrote:
On May 13, 8:56 am, Roy Lewallen wrote: K1TTT wrote: On May 11, 8:30 pm, Art Unwin wrote: When an array is in equilibrium then Maxwell's equations are exact. maxwell's equations are ALWAYS exact, it is digital models that are inexact and have limitations due to the approximations made and the numeric representations used. Inexactness of the solution isn't because the method is digital. The field equations solved by the digital methods simply can't be solved by other methods, except for a relatively few very simple cases. Many non-digital methods were developed over the years before high speed computers to arrive at various approximate solutions, but all have shortcomings. For example, I have a thick file of papers devoted to the apparently simple problem of finding the input impedance of a dipole of arbitrary length and diameter. Even that can't be solved in closed form. Solution by digital methods is vastly superior, and is capable of giving much more accurate results, than solution by any known method. Roy Lewallen, W7EL quantization of every number in a numeric simulation is but one of the contributions to inaccuracy. the limitations of the physical model is another, every modeling program i know of breaks the physical thing being modeled into small pieces, some with fixed sizes, some use adaptive methods, but then they all calculate using those small pieces as if they were a single homogonous piece with step changes at the edges... Not all modeling uses step changes. Some modeling approaches use a model description that is continuous at element boundaries (at least for some number of derivatives). For example, a cubic spline has smoothly varying values, first and second derivatives. The tradeoff in the code is whether you use fewer, better (higher order modeling) chunks or more simpler chunks. For instance, NEC uses a basis function that represents the current in a segment (the chunk) as the combination of a value and two sinusoid sections. Other codes assume the current is uniform over the segment, yet others assume a sinusoidal distribution or a triangle. This leads to a tradeoff in computational resources required: numerical precision, computational complexity, etc. (lots of simple elements tends to require bigger precision) I think that for codes hams are likely to encounter, these are pretty subtle differences and irrelevant. A lot of the "computational efficiency" issues are getting smaller, as cheap processor horsepower is easy to come by. that also adds to inaccuracies. the robustness of the algorithm and the residual errors created are a bit part of getting more accurate results. There is no doubt that numerical methods have allowed 'solutions' of many problems that would be extremely difficult to find closed form solutions for, but they must always be examined for the acceptibility of the unavoidable errors in the method used. That's why there's all those "validation of modeling code X" papers out there. |
#10
|
|||
|
|||
Computer model experiment
On 5/14/2010 6:19 AM, K1TTT wrote:
quantization of every number in a numeric simulation is but one of the contributions to inaccuracy. the limitations of the physical model is another, every modeling program i know of breaks the physical thing being modeled into small pieces, some with fixed sizes, some use adaptive methods, but then they all calculate using those small pieces as if they were a single homogonous piece with step changes at the edges... that also adds to inaccuracies. the robustness of the algorithm and the residual errors created are a bit part of getting more accurate results. There is no doubt that numerical methods have allowed 'solutions' of many problems that would be extremely difficult to find closed form solutions for, but they must always be examined for the acceptibility of the unavoidable errors in the method used. I will assume that most here are familiar with Simpson's Rule Integration. This allows one to compute the "area under the curve" of a function with a fairly simple algorithm. It's as little as 7 statements using Fortran. And it is quite amazing how accurate the answer can be with even just a few slices of the curve from start to finish. If used properly. Don't think that seemingly large chunks mean poor accuracy. When the algorithm is good, and the program selects the chunk size well, the results can be very close to the true answer. tom K0TAR |
Reply |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
FA: Philbrick GAP/R Model K2-W Early Computer Tube Op-Amp | Boatanchors | |||
FA: Philbrick GAP/R Model K2-W Early Computer Tube Op-Amp | Boatanchors | |||
FA: Philbrick GAP/R Model K2-W Early Computer Tube Op-Amp | Boatanchors | |||
FA: Philbrick GAP/R Model K2-W Early Computer VacuumTube Op-Amp | Boatanchors | |||
FA: Radio Shack Model 100 laptop computer ++ | Equipment |