From masson@misvax.plk.af.mil Wed Mar 17 09:50:12 1993 Received: from misvax.plk.af.mil ([129.238.20.35]) by amber.unm.edu (4.1/3.1) id ; Wed, 17 Mar 93 09:49:45 MST Message-Id: <9303171649.AA03486@amber.unm.edu> Date: 17 Mar 93 09:45:00 MDT From: "MASSON, BRUCE" Subject: transmitted intensity To: "vageli" Status: R Vageli, in our discussions yesterday about the anomolous behavior of e_L(N), i.e. the transmitted amplitude, I forgot to mention that e_L(N) is the quantity that the Pittsburg group says oscillates. What I'm wondering is if they weren't seeing garbage!! What do you think? Bruce From avatar Wed Mar 17 10:57:48 1993 Received: by amber.unm.edu (4.1/3.1) id ; Wed, 17 Mar 93 10:57:47 MST Date: Wed, 17 Mar 93 10:57:47 MST From: avatar@amber.unm.edu (Robert A. Salazar) Message-Id: <9303171757.AA03537@amber.unm.edu> To: vageli@amber Subject: Math504 Status: R Hi, I am having a little trouble understanding what A is in problem number 5 in the last homework assignment. What is written is A = [1, a, 1] of order n with a >= 2. What does this mean? A looks like a 1x3 matrix to me. Hence, x must be a 3x1 and b a scalar. If this is the case, I am not sure what the rest of the problem is asking me. Thanks for the help. Robert. From akaul@math.unm.edu Wed Mar 17 12:21:07 1993 Received: from deepthought.unm.edu by amber.unm.edu (4.1/3.1) id ; Wed, 17 Mar 93 12:21:05 MST Received: by deepthought.unm.edu (4.1/3.1) id ; Wed, 17 Mar 93 12:21:04 MST Date: Wed, 17 Mar 93 12:21:04 MST From: akaul@math.unm.edu (Ann M. Kaul) Message-Id: <9303171921.AA15557@deepthought.unm.edu> To: vageli@amber.unm.edu Subject: letters of reference Status: R Dr. Coutsias, I heard from some of the universities I applied to for the fall. They said that they still need a letter from you before they decide on the TA positions. I would really appreciate it if you could send them as soon as possible as I understand there are not very many positions available here at UNM for this fall. If you have already sent them, please disregard. Thank you again for taking the time to do this for me. Ann Kaul From David.C.Montgomery@dartmouth.edu Wed Mar 17 13:41:29 1993 Received: from dartvax.dartmouth.edu by amber.unm.edu (4.1/3.1) id ; Wed, 17 Mar 93 13:41:20 MST Received: from dasher.dartmouth.edu by dartvax.dartmouth.edu (5.65+D4/4.5HUB) id AA18702; Wed, 17 Mar 93 15:41:14 -0500 Message-Id: <2375414@dasher.Dartmouth.EDU> Date: 17 Mar 93 15:41:13 EST From: David.C.Montgomery@dartmouth.edu (David C. Montgomery) Subject: Re: status To: vageli@amber.unm.edu (Evangelos Coutsias) Cc: WESLEY@northstar.dartmouth.edu (Wesley B. Jones) Status: R Dear Vageli: We have about three fourths of a draft manuscript, including figures, which we are willing to send you today, if you will keep it under your hat. It is pretty long for FAXing, and we would prefer to send it by mail, but we will FAX it if you prefer. Wes says you can have the gif color plots that he has generated if you can view them, also. You can contact him about this any time. Insofar as we understand your results, we do not presently see them as statistically-mechanically explainable. If I am wrong, correct me, but I believe that for the 2D plane Couette case, the long-time steady state to which the system relaxes is simply the linear velocity profile, with no coherent vortices. Do you have reason to doubt that? In the plane Poiseuille case, the results are definitely different: there seems to be a long time 2D steady state with vortices that will persist indefinitely, and exhibit an interesting scatter plot of omega vs. psi forever. We feel certain that the rather pyrotechnical things you are seeing, with vortices jumping off the walls, are transient phenomena, interesting to be sure, but situation dependent in a way that the long time Poiseuille vortex street does not seem to be. If there is a test of the statistical mechanical stuff, it would seem to us likely that it would occur in the later stages of evolution, where there is one big vortex left waiting to die. A scatter plot for that vortex might reveal interesting connections between omega and psi, and if so, we would be interested. From our present perspective, we don't see much that encourages trying to prove anything much in common for the two cases. It might be easier all around to submit them as companion papers and thank each other for discussions, but not try to take responsibility for the things that were in each other's papers. We can change our mind about this, if you want to persuade us to. Let us hear from you, and one way or another, we will get what we have off to you. You may also want to send what you have to us by the same way. I will be gone from Friday to Monday, and I would expect we might reach some kind of decision next week. All the best, Dave From bergeron Wed Mar 17 13:45:35 1993 Received: by amber.unm.edu (4.1/3.1) id ; Wed, 17 Mar 93 13:45:35 MST Date: Wed, 17 Mar 93 13:45:35 MST From: bergeron@amber.unm.edu (Keith Bergeron) Message-Id: <9303172045.AA03759@amber.unm.edu> To: vageli@amber.unm.edu Subject: I'll be there from 20 Mar - 23,24 Mar Status: R Keith From stanly@crunch.unm.edu Thu Mar 18 09:35:30 1993 Received: from crunch.unm.edu by amber.unm.edu (4.1/3.1) id ; Thu, 18 Mar 93 09:35:29 MST Received: by crunch.unm.edu (4.1/3.1) id ; Thu, 18 Mar 93 09:35:18 MST Date: Thu, 18 Mar 93 09:35:18 MST From: stanly@crunch.unm.edu (Stanly Steinberg) Message-Id: <9303181635.AA13393@crunch.unm.edu> To: math_applied@crunch.unm.edu Subject: Sabbitical visitor for Fall Status: R Professor Golik will visit for the Fall semester and is interested in working with people in scientific computing. CURRICULUM VITA Wojciech L. Golik Assistant Professor in Mathematics Department of Mathematics and Computer Science University of Missouri-St. Louis (314)-553-6521 golik@arch.umsl.edu Research Objectives: To be a part of a research program in scientific computing with emphasis on grid generation and grid adaption for partial differential equations. Research experience includes grid adaption for two-point boundary value problems, boundary value methods for PDEs, inverse problems for heat equation. Interested in high performance computing. Education: Ph.D. Numerical Analysis, New Mexico State University, 1989 M.S. Mathematics, New Mexico State University, 1985 M.S. Mechanical Engineering, Poznan Polytechnic Institute, Poland, 1982 Employment: University of Missouri-St. Louis, Department of Mathematics and Computer Science, Assistant Professor, 1990-Present University of Missouri-St. Louis, Department of Mathematics and Computer Science, Visiting Assistant Professor, 1988-1990 KORUND, Kolo, Poland, Mechanical Engineer, 1982-1983 Academic Honors: New Mexico State University Graduate School Fellowship, 1986-1988 Research Articles: Superconvergent Grids for Two-Point Boundary Value Problems, Mat. Apl. Comput. 10(1991),43-58, (with W.Connett and A. Schwartz). Convergence of Boundary Element Methods in Numerical Solutions of Fourier Problems, Proyecciones 10(1991), 1-12, Continuous Dependence of Solutions of Some Inverse Problems in Heat Conduction, Appl. Mat. 21 (4), 1993, 491-501. Superconvergent Grids for Systems of Two-Point Boundary Value Problems, (with W.Connett and A. Schwartz), submitted. Order Increasing Grid Adaption for Runge-Kutta Methods Applied to Two-Point Boundary Value Problems}, submitted. Computing Experience: FORTRAN, MATLAB, MATHEMATICA From schrader@xochitl.unm.edu Thu Mar 18 15:10:24 1993 Received: from xochitl.unm.edu by amber.unm.edu (4.1/3.1) id ; Thu, 18 Mar 93 15:10:22 MST Received: by xochitl.unm.edu (4.1/3.1) id ; Thu, 18 Mar 93 15:10:05 MST From: schrader@xochitl.unm.edu (Ron Schrader) Message-Id: <9303182210.AA01602@xochitl.unm.edu> Subject: PVM To: archie@xochitl.unm.edu (ARCHIE Gibson), hagstrom@xochitl.unm.edu (Tom Hagstrom), vageli@xochitl.unm.edu (Evangelos Coutsias), lorenz@xochitl.unm.edu (Jens Lorenz), jgarrard@xochitl.unm.edu (Julie Garrard) Date: Thu, 18 Mar 93 15:10:04 MDT Cc: astone@xochitl.unm.edu (Alex Stone), mreisch@xochitl.unm.edu (Christian M. Reisch) X-Mailer: ELM [version 2.3test PL26] Status: R I have spoken to Mique and Julie about PVM, and our conclusion was that there is no way we are going to put this up right now. The reasons are: - The network currently is SATURATED. We see every packet sent out by almost every PC and workstation on campus. This network cannot absorb increases. - PVM must use the network as a (logical) bus. It would add substantially to network traffic. - There are no real ``slow'' times for network traffic. It would still not work to try running this only at night. - The lab machines with only 8 Meg of Ram and slow ISA busses are not worth trying to run this on. Network traffic is a consideration over there as well. When might it be reasonable to try putting this up? - When we are subnetted. Once we are isolated from chatter all over campus, there will be much more capacity on our network. - When the network has been upgraded (our daisy-chained system is inviting disaster under the best of circumstances). When might these upgrades occur? We would subnet right away if we were allowed. CIRT has the mandate for networking the campus, however, and it is pretty much up to them. It might be worth talking to Gary Bauerschmidt and stressing that important projects in this department are being held up while we wait for CIRT to upgrade us. They are working on things, but he has told us that we are low priority since we have a functioning network. If we need to run something like PVM, we could argue that the network is not functioning. In the short run, and probably better for the long run, you might consider finding grant money to buy a few more RS6000's (which are now dirt cheap), and running IBM's PVM-like system on them. You'll see a lot better performance more easily than trying to figure out how to scale computations in a heterogeneous environment like ours (SUN 3's and various levels of SPARC). I would be happy to discuss this further. From bergeron Fri Mar 19 19:07:23 1993 Received: by amber.unm.edu (4.1/3.1) id ; Fri, 19 Mar 93 19:07:22 MST Date: Fri, 19 Mar 93 19:07:22 MST From: bergeron@amber.unm.edu (Keith Bergeron) Message-Id: <9303200207.AA07054@amber.unm.edu> To: vageli@amber.unm.edu Subject: I'm here! Status: R Got in late this afternoon. Will work until approx. 2200-2230. Then be in tomorrow from 0800-1100, 1300-??? I don't have a good phone number to get a hold of me yet, but should by tomorrow afternoon. Blue skies . . . Keith From lynov@risoe.dk Mon Mar 22 10:27:03 1993 Received: from rishp1.risoe.dk by amber.unm.edu (4.1/3.1) id ; Mon, 22 Mar 93 10:26:58 MST Received: from plf-sun.risoe.dk by rishp1.risoe.dk with SMTP (16.8/16.2) id AA07169; Mon, 22 Mar 93 18:27:30 +0100 Received: by plf-sun.risoe.dk (4.1/SMI-4.1) id AA19870; Mon, 22 Mar 93 18:25:55 +0100 Date: Mon, 22 Mar 93 18:25:55 +0100 From: lynov@risoe.dk (Jens-Peter Lynov) Message-Id: <9303221725.AA19870@plf-sun.risoe.dk> To: vageli@amber.unm.edu Subject: Some progress Status: R Dear Vangelis. I hope you enjoyed your mountain trip and have had some well-earned time to think on the more important matters in life. Things here have been very hectic (for a change). I've been writing grant proposals (we're asking for 1 million kroners worth of computer time for the next year starting April 1st) and annual reports (the department has a 5 million kroner deficit for 1992! After some hectic days we found out that an income of total 2.5 million belonging to '92 is placed either under '91 or '93. The rest the directors will steal from us because we're so rich!). Anyway, I've been reading some articles on selective decay and maximum entropy (which I've learned are different things) without improving my understanding very much. In desperation a tried a few things, and I've sent you the results. I went back to our old run 52, which has the highest Re (4450) we've used (except for our very last and weird run # 64). I then made some scatter plots. They contain the following: BWlinscat52_0.ps: omega vs. psi at T=0. I only took every fourth point in the x- and y-directions in order to keep the number of points low. This goes for the other plots as well. I found it reassuring that there isn't a functional relationship buildt in from the beginning. BWlinscat52_15.ps and BWlogscat52_15.ps show the results you already have seen, but on a larger scale. It's clear that the dependence is neither linear nor logarithmic. BWspecscat52_15.ps is inspired by the Sommeria et al. paper. They suggest a linear relationship between psi and log(omega/(a-omega)), where a is the original amplitude omega. I know that their theory is for a top-hat vorticity distribution, but it couldn't hurt trying. You can see that our result is just as good as their! The value of a is simple for the middle vortex, and after a few attempts I found a=60 is reasonable for the wall created vortex. Changing the value of a changes the curvature of the upper part of the curves a lot, and I'm sure I can make the curves look even more straight. But what does theory have to say? Another progress note, which actually is a minor breaktrough, is that Jan finally got the channel code to run on the Amdahl today. Several things have been teasing him, but mainly the Cosine transformation. Anyway, the code is running, and the Amdahl is more than 10 times faster than the Convex. We`ve just launched a run corresponding to our old run 52, but with 256*1024 resolution and Re=10,000 (according to my definition) which is more than twice as large as we ever tried before. We put the noise level down to 10**-10 and hope to see the two vortex state stay longer and more pronounced. We'll go half way to this state tonight (we don't want to spend too much CPU time if something goes wrong) and will continue tomorrow. At last we have some muscle to do things, we only have to agree on what to do! That's all for now. I fill you in with the new results as soon as we get them. Jens-Peter