function get_noise, data1d a=convol(data1d, [-1.,2.,-1.]) n=n_elements(a) return, stdev(a[1:n-2])/sqrt(6.) end function niris_cogmag, wv, idata, vdata, wlc, wv1=wv1, wv2=wv2 ; ;+ ; NAME: ch_cogmag ; PURPOSE: determine the line-of-sight field from Stokes I and V profiles ; using the cener of gravity method ; Calling Sequence: ; ; b = ch_cogmag(wv, idata, vdata, wlc, line=line) ; INPUTS: ; wv wavelength (1D) in angstrom ; idata, vdata Stokes I and V data (2D) normalized by the continuum intensity ; idata(*,0) --> specral profiles at 0-th point ; idata(0,*) ---> spatial profiles at 0-th wavelength ; Optional output ; wlc center-of-gravity of wavelength ; OUTPUTS: ; b an array of line-of-sight field ; HISTORY ; 2007 June : first coded by J Chae ; 2007 July : dispersion "dwl" is replaced by wavelength array "wl" ; wavelength center is obtained as an optional output ;- ;common _MEsinglet, geff, lambda_rest ; Lande g-factor and rest wavelength in angstrom geff=3.0 lambda_rest=15648.5 ; nwv = n_elements(idata[*,0]) npoint =n_elements(idata[0,*]) wv1=fltarr(npoint) & wv2=fltarr(npoint) for k=0, npoint-1 do begin weight1=(1-(idata(*, k)+vdata(*,k))) weight2=(1-(idata(*,k)-vdata(*,k))) cutlevel1= (abs(min(weight1))) cutlevel2= (abs(min(weight2))) wv2[k] = total(weight1*wv)/total(weight1) wv1[k] = total(weight2*wv)/total(weight2) endfor coeff=4.67e-13*lambda_rest^2*geff B = (wv2-wv1)/(2*coeff) wlc = 0.5*(wv2+wv1) return, B end pro niris_MEsinglet, x, par, f ;common ch_MEsinglet, geff, lambda_rest ; Lande g-factor and rest wavelength in angstrom geff=3.0 lambda_rest=15648.5 ; B = par[0] ; magnetic field strength in Gauss theta = par[1] ; inclination in radian chi = par[2] ; azimuth in radian eta0 = par[3] ; ratio of line center opacity to the continuum one, fixed to 1.3 dlambdaD= par[4] ; Doppler width in angstrom, fixed to 0.15 a = par[5] ; damping parameter, fixed to 1 lambda0 = par[6] ; line center in A B0 = par[7] ; zeroth order of the source function B1 = par[8] ; first order of the source function lambda = x[0:n_elements(x)/4-1] v= (lambda-lambda0)/dlambdaD vb = geff*(4.67e-13*lambda_rest^2*B)/dlambdaD ; ch_voigt, a, v+vb, phib, psib ;, phibda, phibdu, psibda, psibdu ch_voigt, a, v, phip, psip ;, phipda, phipdu, psipda, psipdu ch_voigt, a, v-vb, phir, psir ;, phirda, phirdu, psirda, psirdu factor=1./sqrt(!pi) phib=phib*factor & psib=psib*factor phip=phip*factor & psip=psip*factor phir=phir*factor & psir=psir*factor st=sin(theta) st2 = st^2 ct = cos(theta) etaI = 1 + 0.5*eta0*(phip*st2+0.5*(phib+phir)*(1+ct^2)) etaQ = eta0*0.5*(phip-0.5*(phib+phir))*st2*cos(2*chi) etaU = eta0*0.5*(phip-0.5*(phib+phir))*st2*sin(2*chi) etaV = eta0*0.5*(phir - phib)*ct rhoQ = eta0*0.5*(psip -0.5*(psib+psir))*st2*cos(2*chi) rhoU = eta0*0.5*(psip -0.5*(psib+psir))*st2*sin(2*chi) rhoV = eta0*0.5*(psir - psib)*ct Delta = etaI^2*(etaI^2-etaQ^2-etaU^2-etaV^2+rhoQ^2+rhoU^2+rhoV^2) $ - (etaQ*rhoQ+etaU*rhoU+etaV*rhoV)^2 I = B0 + B1*etaI*(etaI^2+rhoQ^2+rhoU^2+rhoV^2)/Delta Q = -B1*(etaI^2*etaQ+etaI*(etaV*rhoU-etaU*rhoV) $ +rhoQ*(etaQ*rhoQ+etaU*rhoU+etaV*rhoV))/Delta U = -B1*(etaI^2*etaU+etaI*(etaQ*rhoV-etaV*rhoQ) $ +rhoU*(etaQ*rhoQ+etaU*rhoU+etaV*rhoV))/Delta V = -B1*(etaI^2*etaV+etaI*(etaU*rhoQ-etaQ*rhoU) $ +rhoV*(etaQ*rhoQ+etaU*rhoU+etaV*rhoV))/Delta f=[I, Q, U, V] end function niris_mefit, x, data, par, chisq=chisq, display=display n=n_elements(x)/4 weight=replicate(get_noise(data[0:n-1]), n) ;stop for k=1, 3 do weight=[weight, replicate(get_noise(data[n*k:n*k+n-1]), n)] weight=1./weight^2 ;stop parinfo = replicate({value:0., fixed:0, limited:[0,0], $ limits:[0.,0]}, 9) ; parinfo[fixed].fixed = 1 parinfo[[3,4,5,7,8]].limited[0] = 1 parinfo[[3,4,5,7,8]].limits[0] = 0.; [0., 0, 0.1, 0.1, ;TEMP parinfo[[3, 5]].limited[1]=[1,1] & parinfo[[3,5]].limits[1]=[20, 5];was, [1,1], [10,5] parinfo[0].limited=[1,1] parinfo[0].limits=[0,6000] parinfo[1].limited=[1,1] parinfo[1].limits=[0, !pi] parinfo[2].limited=[1,1] parinfo[2].limits=[0, !pi] parinfo[3].fixed=1 parinfo[4].fixed=1 parinfo[5].fixed=1 ;parinfo[3].limited[1]=1 & parinfo[3].limits[1]=5. parinfo[*].value = par ; [5.7D, 2.2, 500., 1.5, 2000.] if display then begin ;print, par niris_MEsinglet, x, par, f endif result=mpcurvefit(x, data, weight, par, sigma, function_name='niris_MEsinglet' , quiet=0, $ chisq=chisq, noderiv=1, itmax=100, tol=1.e-4, parinfo=parinfo) if keyword_set(display) then begin window, 2, xs=800, ys=800 if n_elements(result) gt 1 then begin for k=0, 3 do begin if k eq 0 then yr=[0.5, 1.2] else yr=[-0.1, 0.15] plot, x[0:n-1], data[n*k:n*(k+1)-1], $ psym=1, xst=1, yst=1, yr=yr, ytitle=(['I', 'Q', 'U', 'V']+'/I!d0!n')[k], xtitle='Wavelength (A)' oplot,x[0:n-1], f[n*k:n*(k+1)-1], linest=2, thick=2 oplot,x[0:n-1], result[n*k:n*(k+1)-1], linest=0, thick=2 endfor endif stop ;print, par, chisq/(n*4) ;print, 'blos=', par[0]*cos(par[1]), ', bt=', par[0]*sin(par[1]) endif return, result end pro init_par, x, data, par, display=display dwl_line=1.5;was, 1.5 dwl_core=0.15 btcoeff=5000. n=n_elements(x)/4 continuum =median((data[0:n-1])[where(abs(x[0:n-1]) ge dwl_line )] ) line=where(abs(x[0:n-1]) le dwl_line) blos=niris_cogmag((x[0:n-1])[line], (data[0:n-1])[line]/continuum, $ (data[3*n:4*n-1])[line]/continuum) core=where(abs(x[0:n-1]) le dwl_core ) ; or abs(x[0:n-1]-wlc[0]+0.4) le 0.1 ) icore=((data[0:n-1])[core]) qicore=((data[n:2*n-1])[core]) /icore uicore=((data[2*n: 3*n-1])[core])/icore bt=btcoeff*sqrt(sqrt(mean(qicore^2+uicore^2))) ;if keyword_set(display) then print, 'blos=', blos[0], 'bt=', bt[0] if blos gt 5000 then blos =5000. Bfield=sqrt(blos^2+bt^2) ; magnetic field strength in Gauss theta= atan(bt, blos) ; inclination in radian chi=atan(-uicore, -qicore) chi=0.5*median(chi+2*!pi*(chi lt 0.)) eta0 = 1.3 ; ratio of line center opacity to the continuum one dlambdaD= 0.15 ; Doppler width in angstrom adamp = 1. ; damping parameter lambda0 = 0. ; line center in A ; zeroth order of the source function ;///////////////// B1 = 4*(continuum-min(icore)) ; first order of source function B0=continuum-B1 par=[Bfield, theta, chi, eta0, dlambdaD,adamp, lambda0, B0, B1] ; 0 1 2 3 4 5 6 7 8 print, par end ;MAIN _______________________________________ function main_test, a, h display=1 platform='win' set_plot, platform sza=size(a) nx=sza(1) & ny=sza(2) & nw=sza(3) print, sza b=a lambda_rest=fxpar(h, 'refwv') dlambda=findgen(fxpar(h, 'NAXIS3'))*fxpar(h, 'INCWV')+fxpar(h, 'STARTWV') !p.multi=[0, 2,2] s=where(abs(dlambda) le 2.5, n) factor=median(b[*,*,40<(nw-1),0]) if 1 then begin bmag=fltarr(nx, ny) & inclination=bmag & azimuth=bmag chisqa=fltarr(nx, ny) inta=fltarr(nx,ny) parmap=fltarr(nx,ny,9) fit_array=fltarr(nx,ny,n_elements(s),4) endif t1=systime(/s) ;for xpos=0, nx-1 do for ypos=0, ny-1 do begin for xpos=484, 484 do for ypos=298, 298 do begin x=[dlambda[s], dlambda[s], dlambda[s], dlambda[s]] b1=b[xpos:xpos,ypos:ypos, s, *] cont=where(abs(x[0:n-1]) ge 1.5) data=reform(b1)/factor ; adjustment of non-zero bias in Q, U and V????? for k=1, 3 do data[k*n:(k+1)*n-1]=data[k*n:(k+1)*n-1]-median((data[k*n:(k+1)*n-1])[cont]) init_par, x, data, par, display=display result=niris_mefit( x, data, par, chisq=chisq, display=display) if (n_elements(result)-1) gt 0 then fit_array[xpos, ypos, *,*]=reform(result, n_elements(s), 4) if n_elements(par) eq 9 then begin bmag[xpos,ypos]=par[0] inclination[xpos, ypos]=par[1] azimuth[xpos,ypos]=par[2] chisqa[xpos, ypos]=chisq/(n*4) inta[xpos, ypos]=par[7]+par[8] parmap[xpos, ypos, *]= par endif if ypos eq 0 and xpos mod 50 eq 0 then print, 'Processed xpos of '+string(xpos) endfor t2=systime(/s) print, 'Computing time=', t2-t1, ' s' ;window, xs=nx*2, ys=ny*2 ;loadct, 13 ;device, decomposed=0 ; ;tv, bytscl(parmap[*,*,1], 0, !pi), 0 ;tv, bytscl(parmap[*,*,2], 0, !pi), 1 ;loadct, 3 ;tv, bytscl(parmap[*,*,0], 0, 3000), 2 ;loadct, 6 ;tv, bytscl(-parmap[*,*,6], -0.3, 0.3), 3 loadct, 0 stop return, parmap end