Dongarra, J. J. and Grosse, E. 1987. Distribution of mathematical software via electronic mail. Commun. ACM 30, 5 (May), 403-407.  Home/Search   Document Not in Database   Summary   ACM   TOC   Related Articles   Check

....restrictions, which includes many commercial, government, and military connections, may have restrictions that allow only e mail access. 2 The TUGLIB connection To improve the access to the T E X archives and other software at Utah, I have installed a modified version of the netlib server [1], which has been renamed tuglib. This server provides a means whereby remote users can send electronic mail messages containing service requests to a daemon program. The daemon parses the requests, logs them, and responds to them. The tuglib daemon program runs on a local UNIX system at Utah, but ....

....of a directory. 9 Acknowledgements This work was carried out by the author with the support of facilities at the Department of Mathematics at the University of Utah. My deepest thanks go to Eric Grosse and Jack Dongarra for having written the netlib system and published a paper about it [1]. I also want to thank Eric Grosse for making the netlib software available to TUG for modifications to create tuglib, and for keeping a record of everyone who has received netlib so they can get bug fixes. Thanks also go to the TUG Board of Directors for helping in the testing of tuglib, and to ....

Jack Dongarra, Eric Grosse, 1987. Distribution of Mathematical Software via Electronic Mail. CACM, 30(5), 403--407.

.... update and downdate [10] which mimics the behavior of the Linear Programming Dual Active Set al..gorithm [20] The rst matrix is 10 I A 0 A 0 , where A 0 consists of 5446 columns from a larger 6071by 12,230 matrix B with 35,632 nonzeros arising in an airline scheduling problem (DFL001) [13]. The 5446 columns correspond to the optimal solution of the linear programming problem. Starting with an initial LDL factorization of the matrix 10 I A 0 A 0 , we added columns from B (corresponding to an update) until we obtained the factors of 10 I BB . We then removed columns ....

J. J. Dongarra and E. Grosse, Distribution of mathematical software via electronic mail, Comm. ACM, 30 (1987), pp. 403-407.

....Department of Electrical Engineering, Stanford University, Stanford, CA 94305. He is now with the Division of Nuclear Medicine, University of Michigan Medical Center, Ann Arbor, MI 48109. IEEE Log Number 904231 I. If the measurement error has the (positive definite) covariance matrix G212, then we can premultiply y, and h, by I; 2. Singular covariances may be the result of linearly dependent measurements. indicating that other constraints should be incorporated. covariances) are unknown and are difficult to determine. Fur thermore, the state space formulae imply the a priori ....

....we want to choose the smoothing pa rameter x to provide good estimates of g. One method with intuitive appeal and high (statistical) efficiency (as shown in [4] in the linear case is to choose the smoothing parameter that minimizes the cross validation (CV) score, defined by g I , h( 112 (9) Y] a (gin(t) 2 dt. g, is the solution to the smoothing problem posed without data point y. Exact evaluation of (9) is impractical since it would require N iterative smoothing problems for each value of a. Motivated by the coesponding foula for the linear case [4, eq. 17)1, we ....

[Article contains additional citation context not shown here]

J. J. Dongarra and E. Grosse, "Distribution of mathematical software via electronic mail." Commun. Ass. Comput. Mach., vol. 30, pp. 403-407, Oct. 1987.

....The computational requirement of that algorithm is only O(N) due to the band matrix structure of spline In [23] the AM algorithm was applied on an ellipse by ellipse basis, and the sparseness argument was weaker. 10 VII EXPERIMENTAL RESULTS smoothing. Source code is available through NETLIB [54]. That algorithm uses the nonlinear pseudo measurement function n (x) and its derivatives. The cross validation method described in [48, 49] could be used to select the smoothing parameter # automatically in several different ways for this problem. A computationally efficient alternative to ....

J. J. Dongarra and E. Grosse, "Distribution of mathematical software via electronic mail," Comm. ACM, vol. 30, pp. 403--407, Oct. 1987.

....flux from the left in the first sphere cell is given by the interface condition: rirco(ri, O) Z) 3.2 Time integration After spatial discretization a stiff system of ODEs arises that has to be integrated in time. Just as in [1] we use an implicit time integrator, the public domain code DASSL[3, 4]. DASSL is a variable step variable order BDF method. When the order is restricted to one it results in the familiar Backward Euler method, which is the only known implicit method which combines positivity and mass conservation. In our experiments we also allowed higher order BDF schemes; ....

J.J. Dongarra and E. Grosse. Distribution of mathematical software via electronic mail. Commun. ACM, 30:403-407, 1987. (netlib@research.att.com).

....CG method and the MG method, so it is necessary that both methods are efficiently parallelized and implemented for achieving high performance on parallel machines. The CG method consists of matrix vector multiply, inner product and vector addition with multiplying by a scalar called daxpy in BLAS [36, 13]. The MG method consists of smoothing method, matrix vector multiply, restriction and prolongation. Vector addition has O(n) dataparallelism for n dimensional vector, and inner product has O(n= log n) data parallelism, thus parallel steps are O(n=p) and O(n=p log p) respectively with p ....

Dongarra, J. and E. Grosse, "Distribution of mathematical software via electronic mail," Communications of the ACM, vol. 30, pp. 403--407, 1987.

..... David Hough, MIS Maths,Coventry University, Coventry CV1 5FB, UK. d.hough coventry.ac.uk or na.dhough na net.ornl.gov) Distribution can then be arranged in some mutually convenient way. After an initial period of testing it is hoped that CONFPACK will also become available from netlib; see [2]. The request send index for conformal will keep you abreast of any additions to the netlib conformal mapping library . Naturally, we shall be pleased and interested to hear of your experiences, both good and bad, with CONFPACK. 1.2 Description of CONFPACK CONFPACK is a library of standard ....

J.J. Dongarra and E. Grosse. Distribution of mathematical software via electronic mail. Commun. ACM, 30:403--407, 1987.

....is part of the Higher Educational National Software Archive which includes HENSA Micros at the University of Lancaster. Both HENSA services are available free to all members of the UK Higher Education community. The HENSA Unix service grew out of an SERC funded pilot scheme to provide a UK netlib [1] server at Kent. The original netlib service at AT T allowed research workers to obtain high quality scientific software via electronic mail and the provision of a UK server was aimed at reducing the need for expensive accesses to the US server. This UK pilot server was expanded and diversified. ....

Jack Dongarra and Eric Grosse. Distribution of mathematical software via electronic mail. Commun. ACM, 30(5):403--407, May 1987.

....(see also below in the section on the future of the NRM2 routines) The original, published code for the single precision version, snrm2, with the introductory comments stripped out and with some minor corrections, is presented in Figure 2. This is the code currently available from netlib[18]. FORTRAN 66 TO FORTRAN 77 However harmful it may be considered [19] the use of explicit GOTO statements and labels was unavoidable in Fortran 66 due to the extremely limited control structures that were provided. For example, there was no IF ELSE IF ELSE block available; thus even a simple ....

J. J. Dongarra and E. Grosse, `Distribution of mathematical software via electronic mail', Commun. ACM, 30, (5), 403--407, (1987).

....suitable for student use is available in the handbook [2] and in the freely distributed notes [10] An electronic mail magazine MATLAB Digest has been created recently as a forum for questions, answers and tips. It is planned that Mfiles willbedistributedthroughtheNETLIBsoftware facility [3] (the collection [5] is already available from NETLIB) To be added to the MATLAB Digest electronic mailing list send an email to MATLABusers request mcs.anl.gov. MATLAB is widely used both for research and for teaching. In myown research in matrix computations I ....

J.J. Dongarra and E. Grosse, Distribution of mathematical software via electronic mail, Comm. ACM, 30 (1987), pp. 403-407.

....also be included as rows in the matrix. Similarly, an entire document is usually the text object of interest, but smaller, more topically coherent units of text (e.g. paragraphs, sections) could be represented as well. For example, LSI has been incorporated as a fuzzy search option in NETLIB [6] for retrieving algorithms, code descriptions, and short articles from the NA Digest electronic newsletter. Regardless of how the original descriptor object matrix is derived, a reduced dimensionapproximationcan be computed. The important idea in LSI is to go beyond the original descriptors to ....

J. J. DONGARRA AND E. GROSSE, Distribution of mathematical software via electronic mail, Communications of the ACM, 30 (1987), pp. 403--407.

..... Dynamic process control . Nonblocking collective operations . Fortran 90 and C language bindings . Graphics . Real time support For more information, see the MPI specific newsgroup at comp.parallel.mpi. The official version of the specification document can be obtained from netlib [3] by sending an email message to netlib www.netlib.org with the message: send mpireport. ps from mpi . A postscript file will be mailed back to you by the netlib server. The document may also be obtained via anonymous ftp from www.netlib.org mpi mpi report.ps. A hypertext version is available on ....

Dongarra, J., and Grosse, E. Distribution of mathematical software via electronic mail. Commun. ACM 30, 5 (July 1987), 403--407.

....square nonsymmetric matrices, rectangular matrices, and symmetric positive de nite matrices. Our test set is the entire University of Florida sparse matrix collection [7] which includes the Harwell Boeing test set [15, 16] the linear programming problems in Netlib at http: www.netlib.org [13], as well as many other matrices. We exclude complex matrices, nonsymmetric matrices for which only the pattern was provided, and unassembled nite element matrices. Some matrices include explicit zero entries in the description of their pattern. Since we ignore numerical cancellation, we included ....

J. J. Dongarra and E. Grosse. Distribution of mathematical software via electronic mail. Comm. ACM, 30:403-407, 1987. 26

....University of Southwestern Louisiana U.S.L. Box 4 1010 Lafayette, LA 70504 1010 U.S.A. Email: rbk usl.edu Phone: 318) 482 5270 April 19, 1995 1 INTLIB The FORTRAN 77 package INTLIB is ACM Transactions on Mathematical Software Algorithm 737 [6] and is thus available over NETLIB [3] and [2] either via FTP (ibid. or over the World Wide Web 1 . It consists of standard FORTRAN 77 subroutines and functions for the elementary operations, some of the Fortran standard functions, and certain logical and utility operations. INTLIB achieves portability with simulated directed ....

Grosse, E., and Dongarra, J. J., Distribution of Mathematical Software via Electronic Mail , SIGNUM Newsletter 20 (3), pp. 45--47, 1985.

....University of Southwestern Louisiana U.S.L. Box 4 1010 Lafayette, LA 70504 1010 U.S.A. Email: rbk usl.edu Phone: 318) 482 5270 April 19, 1995 1 INTLIB The FORTRAN 77 package INTLIB is ACM Transactions on Mathematical Software Algorithm 737 [6] and is thus available over NETLIB [3] and [2], either via FTP (ibid. or over the World Wide Web 1 . It consists of standard FORTRAN 77 subroutines and functions for the elementary operations, some of the Fortran standard functions, and certain logical and utility operations. INTLIB achieves portability with simulated directed rounding, ....

....directly. PROGRAM TESTINTLIB C Intervals are represented as double precision arrays C with two elements DOUBLE PRECISION X(2) F(2) DOUBLE PRECISION TMP2(2) TMP3(2) C Initialize machine constants and interval constants used in C the standard functions CALL SIMINI C The range over [1,2] will be computed X(1) 1D0 X(2) 2D0 C Round out in case the decimal to binary conversion is C not exact CALL RNDOUT(X, TRUE. TRUE. C Compute X 4 X 3 X CALL POWER(X,4,TMP2) CALL POWER(X,3,TMP3) CALL ADD(TMP2,TMP3,TMP2) CALL ADD(TMP2,X,F) WRITE(6, F(1) F(2) END ....

Dongarra, J. J., Grosse, E., Distribution of Mathematical Software Via Electronic Mail , Comm. ACM 30 (5), pp. 403--407, 1987.

....given in [6] If one would integrate such a sti system in time with an explicit scheme this would lead to time steps in the order of 10 7 , even when the solution is close to steady state. Therefore it is necessary to use an implicit time integrator, in our case the o the shelf solver DASSL[1, 3]. DASSL is a variable step variableorder BDF method. When the order is restricted to one it results in the familiar Backward Euler method, which is the only known implicit method which combines positivity and mass conservation. In our experiments we also allowed higher order BDF schemes; negative ....

J.J. Dongarra and E. Grosse. Distribution of mathematical software via electronic mail. Commun. ACM, 30:403-407, 1987. (netlib@research.att.com).

....criterion as input and slices the PDG before printing it on the screen. The tool also generates files which can be used to visualize the graphs. 6.4.3 Target Language The tool works on FORTRAN. The accepted FORTRAN subset is not explicitly defined. It can handle several source files from netlib [DG87]. Subroutine calls are assumed to be free of side effects and are not expected to modify their parameters. The verified subset can be found in the appendix. The parser and presumably the slicing tool can handle some parallel constructs but these are not used in this evaluation. 6.4.4 Handling The ....

Jack J. Dongarra and Eric Grosse. Distribution of mathematical software via electronic mail. Communications of the ACM, 30(5):403--407, May 1987.

....factorizations and estimating condition numbers. Dense and banded matrices are handled, but not general sparse matrices. In all areas, similar functionality is provided for real and complex matrices, in both single and double precision. LAPACK is in the public domain and available from netlib [DG87]. The original goal of the LAPACK project was to make the widely used EISPACK and LINPACK libraries run efficiently on shared memory vector and parallel processors. On these machines, LINPACK and EISPACK are inefficient because their memory access patterns disregard the multilayered memory ....

....code is written in a standard language, such as Fortran, and that the source code can be compiled on an arbitrary machine to produce a program that will run correctly. We call this the mail order software model of portability, since it reflects the model used by software servers such as netlib [DG87]. This notion of portability is quite demanding. It requires that all relevant properties of the computer s arithmetic and architecture be discovered at runtime within the confines of a Fortran code. For example, if it is important to know the overflow threshold for scaling purposes, it must be ....

[Article contains additional citation context not shown here]

Dongarra, J., Grosse, E., Distribution of mathematical software via electronic mail, Communications of the ACM 30, 1997, (See http://www.netlib.org/).

No context found.

J. Dongarra and E. Grosse. Distribution of mathematical software via electronic mail. Communications of the ACM, 30#5#:403#407, July 1987. URL http:##www.netlib.org#.

No context found.

Jack J. Dongarra and Eric Grosse. Distribution of mathematical software via electronic mail. Communications of the ACM, 30#5#:403#407, May 1987.

No context found.

Dongarra, J. J. and Grosse, E. 1987. Distribution of mathematical software via electronic mail. Commun. ACM 30, 5 (May), 403-407.

No context found.

J. Dongarra and E. Grosse. Distribution of mathematical software via electronic mail. CACM, 30(5):403--407, 1987.

No context found.

J. J. Dongarra and E. Grosse, "Distribution of mathematical software via electronic mail," Comm. ACM, vol. 30, pp. 403--440, 1987.

No context found.

J. J. Dongarra and E. Grosse, Distribution of mathematical software via electronic mail, Comm. ACM, 30 (1987), pp. 403--407.

No context found.

J. J. Dongarra and E. Grosse, Distribution of mathematical software via electronic mail, Comm. ACM, 30 (1987), pp. 403-407.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC