Subject: Space-tech Digest #64 Contents: Peter Scott Re: Orbital Mechanics revisited Hal Mueller Re: Orbital Mechanics revisited Paul Dietz Re: Orbital Mechanics revisited John Sahr Re: Orbital Mechanics revisited Dan Briggs Re: Orbital Mechanics revisited John Sahr Re: Minicomets Henry Spencer Re: AMSAT asteroid mission? ------------------------------------------------------------ Date: Tue, 12 Jun 90 12:48:22 PST From: Peter Scott Subject: Re: Orbital Mechanics revisited To: space-tech@CS.CMU.EDU X-VMS-Mail-To: EXOS%"space-tech@cs.cmu.edu" >Yes, it would be really neat to have some public domain set of libraries >for doing these sort of thing. On the subject of which, I wonder if the >programs used by JPL, etc. are available, most government software is, if >you know how to ask... The best place to get this stuff is COSMIC, NASA's software library... they're set up with the express charter of providing NASA-written software to the public at little or no cost. They have catalogs and indices and everything. They're on the net as cosmic@uga.edu ( think; they're at the University of Georgia. Peter Scott (pjs@grouch.jpl.nasa.gov, with 7 programs in the COSMIC catalog) ------------------------------ Date: Tue, 12 Jun 90 15:07:44 CDT From: Hal Mueller To: space-tech@CS.CMU.EDU Cc: dietz@cs.rochester.edu, uunet.uu.net!cgch!jcp@wfsc4.tamu.edu Subject: Orbital Mechanics revisited T.S. Kelso's BBS in Dayton has (had? I haven't been on it on months) some Fortran and C codes based on NORAD algorithms. I believe these were LEO only, however. Hal Mueller hmueller@cs.tamu.edu n270ca@tamunix (Bitnet) Graduate Student, Department of Computer Science Research Assistant, Department of Wildlife and Fisheries Sciences Texas A&M University, College Station, TX 77843 ------------------------------ To: dbriggs@cholla.AOC.NRAO.EDU Cc: dietz@cs.rochester.edu, space-tech@CS.CMU.EDU Subject: Re: Numerical Recipies Date: Thu, 14 Jun 90 09:29:04 -0400 From: dietz@cs.rochester.edu > [ warnings that NR is not in the same class as IMSL or NAG ] What you get is related to what you pay for. The NR diskette costs $29.95. Since Joe was running this on his PC/AT, I don't imagine his budget was very large (certainly not enough to hire a numerical analyst). I'd go a step further, actually, and say that if you *really* want to solve orbital mechanics problems, get software from the professionals. Some astronomical software is even available online over Internet. One other thing I should have mentioned: NR also has code for a Bulirsch-Stoer ODE solver that it claims is better than predictor-corrector methods. However, I understand that state-of-the-art orbital mechanics simulations use some very high order predictor-corrector methods (up to twelvth order). Paul F. Dietz dietz@cs.rochester.edu ------------------------------ Date: Sun, 17 Jun 90 20:33:08 EDT From: John Sahr To: space-tech@CS.CMU.EDU Subject: Re: Orbital Mechanics revisited In response to solving the "inverse problem" stated by Pete Thomas (I paraphrase) ] "rather than integrating a trajectory, how do we optimize ] a trajectory for fuel (or flight time, or ...)?", Paul Dietz proposed that >We can set up an optimization problem. Divide the time interval >(t_0,t_n) into n-1 points. At each of the n-1 points are six >unknowns: position and velocity of the spacecraft. n is chosen to be >large enough so that the spacecraft does not move far in one step. >The "cost" from time t_i to time t_{i+1} is a function of the >acceleration the engine must provide to get the spacecraft to p_{i+1}, >v_{i+1} (which we can compute, given that the steps are small). If >this acceleration is more than the engine can supply, a very high >artificial cost is imposed. Otherwise, the cost is something like the >amount of reaction mass consumed. This is almost, but not quite, a description of the Simulated Annealing technique for estimating extrema of many-dimensional functions. With a little hand tuning, SA can be breath-takingly fast and provide remarkably good answers. "Hand tuning" comes in designing a good cost function, "quenching schedule," and making sure that you explore enough of parameter space. Every time I have used SA, I have been astounded at the ease of implementation and the speed of convergence. Disclaimer: I have never applied SA to an orbital mechanics problem. Simulated Annealing is described (briefly) in Numerical Recipes (in C, FORTRAN, Pascal) by Press et al., which also provides further references. Since there has been some discussion of Numerical Recipes, I will offer my own mini-review of the C version. NRiC is a good source of algorithms, but I don't use the code as is. The authors abhor white space, useful variable names, structure types, pointer arithmetic, zero-offset arrays, etc.; it appears that they wish C was Fortran (this is not intended as a FORTRAN slam, but it probably sounds like one). The codes work, and ultimately I consider NRiC a good buy. Dan Briggs mentioned _Advanced Mathematical Methods for Scientists and Engineers_, by Bender and Orszag. This is a _fantastic_ book, and contains all sorts of useful information, but does _not_ cover numerical integration techniques, being more interested in analytic methods of approximate solutions to ODEs, integrals, and sums. Don't let the "advanced" in the title put you off if you are a beginner; this book is loaded with examples, and the authors rate the difficulty of the different sections of the book. ------------------------------ Date: Mon, 18 Jun 90 11:20:00 MDT From: Dan Briggs To: space-tech@CS.CMU.EDU Subject: Bender & Orszag Sorry to burn the list bandwidth, but direct mail to John Sahr just bounced. (Some of the rest of you may appreciate this as well....) Just a quick factoid, FYI. A friend of mine is a grad student for Orszag. He claims that work is actively in progress for B & O II. This volume will be completely devoted to PDE's, and will have much the same flavor as the first book. Can hardly wait! While I'm here, can I make an appeal for people to include .sig's even in list traffic? (I realize that not all might agree on this.) Personally, the mail headers that arrive here are cryptic at best. It is not always clear what an appropriate personal reply address is. ---- Daniel Briggs (dbriggs@nrao.edu) New Mexico Tech / National Radio Astronomy Observatory P.O. Box O / Socorro, NM 87801 (505) 835-7360 ------------------------------ Date: Sun, 17 Jun 90 17:09:30 EDT From: John Sahr To: dlbres10@pc.usl.edu, space-tech@CS.CMU.EDU Subject: Re: Minicomets The "water comets" issue is currently quiet, but has not completely died out. As I recall, there was a "water comet" talk at a recent AGU (American Geophysical Convention), but not the most recent one (June 1990). L. A. Frank is the chief proponent of this theory [Frank et al., 1986a, b], and this controversial theory has been questioned by many [c.f. Donahue, 1986, and a response by Frank et al., 1986c]. It is a very interesting theory; even if it is wrong it has been posed and debated in a fashion that is much more satisfying than "cold fusion." The following references sum up most of the issues. Not a whole lot has happened since; it is reasonable to state that the consensus is against the "water comet" theory. GRL == Geophysical Research Letters Donahue, T. M., GRL, v. 13, pp 555-557, June 1986; Frank, L. A., J. B. Sigworth, J. D. Craven, GRL, v13, pp 303-306, 1986a; Frank, L. A., J. B. Sigworth, J. D. Craven, GRL, v13, pp 307-310, 1986b; Frank, L. A., J. B. Sigworth, J. D. Craven, GRL, v13, pp 559-560, 1986c; ------------------------------ From: henry@zoo.toronto.edu Date: Mon, 18 Jun 90 18:33:32 EDT To: space-tech@CS.CMU.EDU Subject: Re: AMSAT asteroid mission? > Is there anyone out there who has the facts on a > proposed Ariane 4 mission to the asteroid Belt by AMSAT? I don't know a lot about it, but the basic observation was that the more sophisticated of the current amateur-radio satellites are close to being useful as planetary probes, so long as you don't need atmospheric entry or operation at great distances from the Sun. (The SSEC reached a similar conclusion, which is why Mars Observer is based on an Earth-orbit weather-satellite design.) However, I don't believe anything came of the Amateur Asteroid Mission proposal. Such ideas do keep coming up. A more recent observation was that the delta-V requirements for lunar orbit are almost identical to those for Clarke orbit, so a Clarke-orbit amateur comsat could easily be a lunar orbiter instead. That hasn't gone anywhere directly, but I suspect it contributed to the Space Studies Institute's current project for a private lunar orbiter. Henry Spencer at U of Toronto Zoology henry@zoo.toronto.edu uunet!attcan!utzoo!henry ------------------------------ End of Space-tech Digest #64 *******************