Saturday, August 13, 2016

Simultaneously Solving in Integers a Nonlinear System of 5500 Diophantine Equations in 5500 Unknowns

Jsun Yui Wong

The following computer program seeks to solve the following nonlinear system of 5500 Diophantine equations:

            5500
x(i)  +   sigma    x(j)  -  (5500+1)   =  0, for i = 1, 2, 3,..., 5499,
            j=1

5500
pi   x(j)     -1      = 0.
j=1

This present system is based on Brown's almost linear function in La Cruz, Marinez, and Raydan [2, p. 25]; http://www.ime.unicamp.br/~martinez/lmrreport.pdf.

One notes the starting vectors of line 42, which is  42 A(J44) = -5 + FIX(RND * 11).

0 REM DEFDBL A-Z
2 DEFINT I, J, X

3 DIM B(9999), N(9999), A(9999), H(9999), L(9999), U(9999), X(9999), D(9999), P(9999), PS(9999), J(9999)

12 FOR JJJJ = -32000 TO 32000

    15 RANDOMIZE JJJJ

    16 M = -1D+37

    41 FOR J44 = 1 TO 5500
        42 A(J44) = -5 + FIX(RND * 11)


    43 NEXT J44
    128 FOR I = 1 TO 30000


        129 FOR KKQQ = 1 TO 5500
            130 X(KKQQ) = A(KKQQ)
        131 NEXT KKQQ
        133 FOR IPP = 1 TO (1 + FIX(RND * 3))

            181 J = 1 + FIX(RND * 5500)

            183 REM R = (1 - RND * 2) * A(J)

            187 IF RND < .5 THEN X(J) = A(J) - 1 ELSE X(J) = A(J) + 1


            189 REM X(J) = A(J) + (RND ^ 3) * R


        192 NEXT IPP

        251 SU = 0
        254 FOR J44 = 1 TO 5499


            258 SU = SU + X(J44)


        266 NEXT J44

        311 X(5500) = -X(1) - SU + (5500 + 1)


        351 PR = 1
        353 FOR J45 = 1 TO 5500


            355 PR = PR * X(J45)

        359 NEXT J45


        422 FOR J41 = 2 TO 5499


            439 P(J41) = -ABS(X(J41) + SU + X(5500) - (5500 + 1))


        427 NEXT J41


        441 P(5500) = -ABS(PR - 1)

        451 FOR J77 = 2 TO 5500


            452 IF P(J77) < 0 THEN P(J77) = P(J77) ELSE P(J77) = 0

        454 NEXT J77


        577 SP = 0

        578 FOR J99 = 2 TO 5500


            579 SP = SP + P(J99)
        580 NEXT J99
        595 P = SP

        1111 IF P <= M THEN 1670
        1452 M = P
        1454 FOR KLX = 1 TO 5500

            1455 A(KLX) = X(KLX)
        1456 NEXT KLX
        1557 GOTO 128
    1670 NEXT I

    1889 REM IF M < -99 THEN 1999


    1947 PRINT A(1), A(2), A(3), A(4), A(5497), A(5498), A(5499), A(5500), M, JJJJ
1999 NEXT JJJJ

This computer program was run with qb64v1000-win [6]. Copied by hand from the screen, the computer program’s complete output through JJJJ=-31985 is shown below:

0   0   0   0   0
0   0   5504   -4   -32000

-1   -1   -1   -1   -1
-1   -1   11000   -2   -31999

0   0   0   0   0
0   0   5501   -3   -31998

0   0   0   0   0
0   0   5503   -3   -31997

0   0   0   0  0
0   0   5500   -2   -31996

-1   -1   -1   -1   -1
-1   -1   11000   -2   -31995

-2   -2   -2   -2    -2
-2   -2     16498     -4    -31994

0   0   0   0   0
0   0   5501   -1   -31993

-1   -1   -1   -1   -1
-1   -1   11000   -2   -31992

0   0   0   0   0
0   0   5501   -1   -31991

1   1   1   1   1
1   1   7   -7   -31990

2   2   2   2   2
2   2   -5496   -3   -31989

1   1   1   1   1
1   1   1   0   -31988

0   0   0   0   0
0   0   5499   -3   -31987

1   1   1   1   1
1   1   1   0   -31986

1   1   1   1   1
1   1   1   0   -31985

Above there is no rounding by hand; it is just straight copying by hand from the screen.

Above at JJJJ=-31988, at JJJJ= -31986, and at JJJJ=-31985, M=0.  Of the 5500 unknowns, only the eight A’s of line 1947 are shown above.

On a personal computer with a Pentium Dual-Core CPU E5200 @2.50GHz, 2.50 GHz, 960 MB of RAM and with qb64v1000-win [6], the wall-clock time for obtaining the output through JJJJ= -31985 was 100 minutes.

Acknowledgment

I would like to acknowledge the encouragement of Roberta Clark and Tom Clark.

References

[1] Jean-Marie de Konnick, Armel Mercier, 1001 Problems in Classical Number Theory.  American Mathematical Society, Providence, Rhode Island, 2007.

[2] William La Cruz, Jose Mario Martinez, Marcos Raydan, Spectral residual method without gradient information for solving large-scale nonlinear systems of equations: Theory and experiments. Technical Report RT-04-08, July 2004.
http://www.ime.unicamp.br/~martinez/lmrreport.pdf.

[3] William La Cruz, Jose Mario Martinez, Marcos Raydan, Spectral residual method without gradient information for solving large-scale nonlinear systems of equations, Mathematics of Computation, vol. 75, no. 255, pp.1429-1448, 2006.

[4] Microsoft Corp. BASIC, second edition (May 1982), Version 1.10. Boca Raton, Florida: IBM Corp., Personal Computer, P. O. Box 1328-C, Boca Raton, Florida 33432, 1981.

[5] O. Perez, I. Amaya, R. Correa (2013), Numerical Solution of Certain Exponential and Non-linear Diophantine Systems of Equations by Using a Discrete Particles Swarm Optimization Algorithm. Applied Mathematics and Computation, Volume 225, 1 December 2013, Pages 737-746.

[6] Wikipedia, QB64, https://en.wikipedia.org/wiki/QB64.

[7] Jsun Yui Wong (2013, November 11). Solving Nonlinear Systems of Equations with the Domino Method, Second Edition. http://myblogsubstance.typepad.com/substance/2013/11/solving-nonlinear-systems-of-equations-with-the-domino-method-second-edition.html.


No comments:

Post a Comment