I'd like to plot a function and then place a histogram with counts based on the function in the same figure. The problem is that I can't seem to avoid gnuplot using the same number of samples to evaluate the function and the histogram bins.
I am attempting to make an illustrative example. Essentially I'd like to show an approximation of the analytic function's integral.
For example if I plot:
f(x) = (1/((x+.3)*(x+.3)) + 1)-(sin(x*5)+cos(x*10))*0.1
plot [0:2]  f(x) notitle lw 5, f(x)*1.5 with boxes
Of course I get an enormous number of histogram bins, when I'd really like to see a small handful (3-6). I realize that I'm trying to use gnuplot to produce a synthetic example when most people are probably making histograms using measured data, but I'd appreciate any advice.
Your problem begins earlier than you think. Function plots don't get binned into histograms by gnuplot. If you want to plot a histogram, you'll have to save the plotted data to a file (--> help table), then bin (--> help smooth frequency) and plot that.