################################################################
# hscale(right length,nlabels,
# minlabel,maxlabel,[T][N])
# arg5=T: top axis, N: nolabels
define(`hscale',`[ define(`hs_nl',`ifelse(`$2',,5,`$2')')
A: line `$1' ; Start: A.start; End: A.end
td = 1/(hs_nl-1)*A.len/5
define(`hs_ud',`ifinstr(`$5',T,down,up)')
# Major tics
for i=1 to hs_nl-2 do { line hs_ud graphsize/20 thick 0.4 \
from i/(hs_nl-1) between A.start and A.end }
# Minor tics
for i=0 to hs_nl-2 do { Ts: i/(hs_nl-1) between A.start and A.end
for j=1 to 4 do { line thick 0.4 hs_ud graphsize/20/2 from Ts+(j*td,0) } }
# Labels
ifinstr(`$5',N,,`for_(0,eval(hs_nl-1),1,
`Ts: m4x/(hs_nl-1) between A.start and A.end
s_box("\small %g",`$3'+m4x/(hs_nl-1)*(`$4'-(`$3'))) \
at Ts ifelse(`$5',T,above,below)') ')
]')
# vscale(up length,nlabels,
# minlabel,maxlabel,[R][N] )
# arg5=R: right axis, N: nolabels
define(`vscale',`[ define(`vs_nl',`ifelse(`$2',,5,`$2')')
A: line `$1' ; Start: A.start; End: A.end
td = 1/(vs_nl-1)*A.len/5
define(`vs_lr',`ifinstr(`$5',R,left,right)')
# Major tics
for i=1 to vs_nl-2 do { line vs_lr graphsize/20 thick 0.4 \
from i/(vs_nl-1) between A.start and A.end }
# Minor tics
for i=0 to vs_nl-2 do { Ts: i/(vs_nl-1) between A.start and A.end
for j=1 to 4 do { line thick 0.4 vs_lr graphsize/20/2 from Ts+(0,j*td) } }
# Labels
ifinstr(`$5',N,,`for_(0,eval(vs_nl-1),1,
`Ts: m4x/(vs_nl-1) between A.start and A.end
s_box("\small %g",`$3'+m4x/(vs_nl-1)*(`$4'-(`$3'))) \
at Ts ifelse(`$5',R,ljust,rjust)')')
]')
# f_gauss(x,mean,stddev)
define f_gauss`{ std_gauss(($1-($2))/($3))/($3) }'
define std_gauss`{ expe(-($1)^2/2)/sqrt(twopi_) }'
################################################################
[
n = 2000 # number of samples
stddev = 0.25
graphsize = 2
B: box wid graphsize ht graphsize
# Gemerate n Gaussian random numbers
randn(z,n,0,stddev)
m = 0
for i=1 to n-1 by 2 do { m +=1; x[m] = z[i]; y[m] = z[i+1]
dot(at B+(x[m],y[m]))
}
hscale(right B.wid,5,-1,1) with .Start at B.sw
hscale(right B.wid,5,-1,1,TN) with .Start at B.nw
vscale(up B.ht,5,-1,1) with .Start at B.sw
vscale(up B.ht,5,-1,1,RN) with .Start at B.se
N: box invis wid B.wid ht B.ht with .sw at B.nw+(0,graphsize/10)
nb = 20
binmax = n/5*0.5
dx = N.wid/nb
histbins(x,m,-1,1,nb,hbin)
for i=0 to nb-1 do { line up hbin[i]/binmax * N.ht \
from N.sw+((0.5+i)*dx,0) thick dx/(1bp__) outlined "gray" }
box wid N.wid ht N.ht at N
# vscale(up N.ht,5,0,1) with .Start at N.sw
E: box invis wid B.wid ht B.ht with .sw at B.se+(graphsize/10,0)
nb = 20
binmax = n/5*0.5
dy = E.ht/nb
histbins(y,m,-1,1,nb,vbin)
for i=0 to nb-1 do { line right vbin[i]/binmax * E.wid \
from E.sw+(0,(0.5+i)*dy) thick dy/(1bp__) outlined "gray" }
box wid E.wid ht E.ht at E
# hscale(right E.wid,5,0,1) with .Start at E.sw
ng = 16
for i=0 to ng do { x = i/ng*N.wid
U: x,(f_gauss(x,N.wid/2,stddev)*m*dx)/binmax*N.ht
X[i]: N.sw+U
Y[i]: E.nw+(U.y,-U.x) }
fitcurve(X,ng)
fitcurve(Y,ng)
