program TEST_JETNET

	implicit	NONE

        Integer*4       hsize
        Parameter       (hsize = 1000000)
	Integer*4	max_inputs
	Parameter	(max_inputs = 30)

        Integer*4       hmemor (hsize)
        common  /pawc/  hmemor

c  common blocks for communication with JETNET

	Real*4		MAXI, MAXO
	Parameter	(MAXI=1000,MAXO=1000)	! dimension of in/out vectors

	Integer*4	MXNDJM
	Integer*4	MSTJN, MSTJM
	Real*4		PARJN, PARJM
	Real*4		OIN, OUT

	COMMON /JNDAT1/ MSTJN(40),PARJN(40),MSTJM(20),PARJM(20),
     &                OIN(MAXI),OUT(MAXO),MXNDJM
	SAVE /JNDAT1/

c  local variables

	Character*80	outfile		/ 'scweek:nets.his' /

	Integer*4	hidden		/ 5 /
	Integer*4	i
        Integer*4       icycle
	Integer*4	indim		/ 2 /
	Integer*4	j
	Integer*4	k
	Integer*4	layers		/ 3 /
	Integer*4	len
	Integer*4	NTRAIN		/ 100000 /
	Integer*4	NTEST		/ 5000 /
	Integer*4       NTOTIN, NTOT2N
	Integer*4	outlun
	Integer*4	seed		/ 12345 /

        Real*4          x, y
	Real*4		r
        Real*4          rvec(2)
        external        ranlux
        Real*4          sigma1, sigma2

c  begin

c  initialize HBOOK, open histogram file, book histogram

        call HLIMIT (hsize)
	outfile = 'test_jetnet.his'
	outlun = 20
        open (unit = outlun, file = outfile,
     1  access = 'direct', form = 'unformatted',
     1  status = 'unknown', recl = 1024)

	seed = 12345
	call RMARIN(seed, NTOTIN, NTOT2N)
        call HRFILE (outlun, 'histog', 'N')
	call HBOOK1 (1, 'output, H1', 100, 0., 1., 0.)
	call HBOOK1 (2, 'output, H2', 100, 0., 1., 0.)

        call HBOOK2 (101, 'x,y -- H0 (train)', 
     1    100, -4., 6., 100, -4., 6., 0.)
        call HBOOK2 (102, 'x,y -- H1 (train)', 
     1    100, -4., 6., 100, -4., 6., 0.)
        call HBOOK2 (103, 'x,y -- all (train)', 
     1    100, -4., 6., 100, -4., 6., 0.)

        call HBOOK2 (201, 'x,y -- H0 (test)', 
     1    100, -4., 6., 100, -4., 6., 0.)
        call HBOOK2 (202, 'x,y -- H1 (test)', 
     1    100, -4., 6., 100, -4., 6., 0.)

	write (*, *) 'number of input nodes'
	read (*, *) indim
	write (*, *) 'number of layers'
	read (*, *) layers
	write (*, *) 'number of hidden nodes'
	read (*, *) hidden

C...Set network architecture: MSTJN(1)-layered network with
C...MSTJN(11) hidden nodes, MSTJN(12) output nodes and
C...MSTJN(10) inputs.
	MSTJN(1)=layers		! number of layers
	MSTJN(10)=indim		! number of input nodes
	MSTJN(11)=hidden	! number of hidden nodes
	MSTJN(12)=1		! number of output nodes
C...Set sigmoid function:
	MSTJN(3)=1
C...Initial width of weights:
	PARJN(4)=0.5
C...Choose updating method 
	MSTJN(5)=0		! Normal Backprop
C...Initialize network:
	CALL JNINIT
C...Set suitable parameters for updating method
	PARJN(1)=2.0
	PARJN(2)=0.5
	PARJN(11)=0.999

c  generate NTRAIN vectors of input data for training with 50-50 types 1,2

	write (*, *) 'number of data sets for training'
	read (*, *) NTRAIN
	do i = 1, NTRAIN

c  generate input and assign values of the input vector 

          call RANMAR (r, 1)
	  if ( r .le. 0.5 ) then
	    call GENERATE_XY (1, x, y)
	    OIN(1) = x
	    OIN(2) = y
	    OUT(1) = 0.
            call HF2 (101, OIN(1), OIN(2), 1.)
	  else
	    call GENERATE_XY (2, x, y)
	    OIN(1) = x
	    OIN(2) = y
	    OUT(1) = 1.
            call HF2 (102, OIN(1), OIN(2), 1.)
	  endif
          call HF2 (103, OIN(1), OIN(2), 1.)

c  train the net

	  call JNTRAL

	end do		! i = 1, NTRAIN

c  Now generate data according to either types 1 or two and histogram
c  output of net.

	write (*, *) 'number of data sets for testing'
	read (*, *) NTEST

	do j = 1, 2
	  do i = 1, NTEST

	    call GENERATE_XY (j, x, y)
	    OIN(1) = x
	    OIN(2) = y

	    call JNTEST
	    call HF1 (j, OUT(1), 1.)
	    call HF2 (200+j, oin(1), oin(2), 1.)
	    call HF2 (203, oin(1), oin(2), 1.)

	  end do
	end do

c  store histogram and close

        call HROUT (0, icycle, ' ')
        call HREND ('histog')
        close (outlun)	

	stop
	END