pm3d in gnuplot with binary data - binary

Question

I have some data files with content
a1 b1 c1 d1
a1 b2 c2 d2
...
[blank line]
a2 b1 c1 d1
a2 b2 c2 d2
...
I plot this with gnuplot using
splot 'file' u 1:2:3:4 w pm3d.
Now, I want to use a binary file. I created the file with Fortran using unformatted stream-access (direct or sequential access did not work directly). By using gnuplot with
splot 'file' binary format='%float%float%float%float' u 1:2:3
I get a normal 3D-plot. However, the pm3d-command does not work as I don't have the blank lines in the binary file. I get the error message:
>splot 'file' binary format='%float%float%float%float' u 1:2:3:4 w pm3d
Warning: Single isoline (scan) is not enough for a pm3d plot.
Hint: Missing blank lines in the data file? See 'help pm3d' and FAQ.
According to the demo script in http://gnuplot.sourceforge.net/demo/image2.html, I have to specify the record length (which I still don't understand right). However, using this script from the demo page and the command with pm3d obtains the same error message:
splot 'scatter2.bin' binary record=30:30:29:26 u 1:2:3 w pm3d
So how is it possible to plot this four dimensional data from a binary file correctly?
Edit: Thanks, mgilson. Now it works fine. Just for the record: My fortran code-snippet:
open(unit=83,file=fname,action='write',status='replace',access='stream',form='unformatted')
a= 0.d0
b= 0.d0
do i=1,200
do j=1,100
write(83)real(a),real(b),c(i,j),d(i,j)
b = b + db
end do
a = a + da
b = 0.d0
end do
close(83)
The gnuplot commands:
set pm3d map
set contour
set cntrparam levels 20
set cntrparam bspline
unset clabel
splot 'fname' binary record=(100,-1) format='%float' u 1:2:3:4 t 'd as pm3d-projection, c as contour'

Answer 1

Great question, and thanks for posting it. This is a corner of gnuplot I hadn't spent much time with before. First, I need to generate a little test data -- I used python, but you could use fortran just as easily:
Note that my input array (b) is just a 10x10 array. The first two "columns" in the datafile are just the index (i,j), but you could use anything.
>>> import numpy as np
>>> a = np.arange(10)
>>> b = a[None,:]+a[:,None]
>>> b
array([[ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9],
[ 1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
[ 2, 3, 4, 5, 6, 7, 8, 9, 10, 11],
[ 3, 4, 5, 6, 7, 8, 9, 10, 11, 12],
[ 4, 5, 6, 7, 8, 9, 10, 11, 12, 13],
[ 5, 6, 7, 8, 9, 10, 11, 12, 13, 14],
[ 6, 7, 8, 9, 10, 11, 12, 13, 14, 15],
[ 7, 8, 9, 10, 11, 12, 13, 14, 15, 16],
[ 8, 9, 10, 11, 12, 13, 14, 15, 16, 17],
[ 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]])
>>> with open('foo.dat','wb') as foo:
... for (i,j),dat in np.ndenumerate(b):
... s = struct.pack('4f',i,j,dat,dat)
... foo.write(s)
...
So here I just write 4-floating point values to the file for each data-point. Again, this is what you've already done using fortran. Now for plotting it:
splot 'foo.dat' binary record=(10,-1) format='%float' u 1:2:3:4 w pm3d
I believe that this specifies that each "scan" is a "record". Since I know that each scan will be 10 floats long, that becomes the first index in the record list. The -1 indicates that gnuplot should keep reading records until it finds the end of the file.