def test_galcencyl_to_vxvyvz():
vr,vp,vz= -17.,6.,35.
phi= numpy.arctan(4./3.)
vxyz= bovy_coords.galcencyl_to_vxvyvz(vr,vp,vz,phi,vsun=[-5.,10.,5.])
assert numpy.fabs(vxyz[0]-10.) < 10.**-10., 'galcenrect_to_vxvyvz conversion did not work as expected'
assert numpy.fabs(vxyz[1]+20.) < 10.**-10., 'galcenrect_to_vxvyvz conversion did not work as expected'
assert numpy.fabs(vxyz[2]-30.) < 10.**-10., 'galcenrect_to_vxvyvz conversion did not work as expected'
return None
def _fit_orbit_mlogl(new_vxvv, vxvv, vxvv_err, pot, radec, lb, customsky,
lb_to_customsky, pmllpmbb_to_customsky, tmockAA, ro, vo,
obs):
"""The log likelihood for fitting an orbit"""
#Use this _parse_args routine, which does forward and backward integration
iR, ivR, ivT, iz, ivz, iphi = tmockAA._parse_args(True, False, new_vxvv[0],
new_vxvv[1], new_vxvv[2],
new_vxvv[3], new_vxvv[4],
new_vxvv[5])
if radec or lb or customsky:
#Need to transform to (l,b), (ra,dec), or a custom set
#First transform to X,Y,Z,vX,vY,vZ (Galactic)
X, Y, Z = coords.galcencyl_to_XYZ(iR.flatten(),
iphi.flatten(),
iz.flatten(),
Xsun=obs[0] / ro,
Zsun=obs[2] / ro).T
vX,vY,vZ = coords.galcencyl_to_vxvyvz(ivR.flatten(),ivT.flatten(),
ivz.flatten(),iphi.flatten(),
vsun=nu.array(\
obs[3:6])/vo,Xsun=obs[0]/ro,Zsun=obs[2]/ro).T
bad_indx = (X == 0.) * (Y == 0.) * (Z == 0.)
if True in bad_indx: X[bad_indx] += ro / 10000.
lbdvrpmllpmbb = coords.rectgal_to_sphergal(X * ro,
Y * ro,
Z * ro,
vX * vo,
vY * vo,
vZ * vo,
degree=True)
if lb:
orb_vxvv = nu.array([
lbdvrpmllpmbb[:, 0], lbdvrpmllpmbb[:, 1], lbdvrpmllpmbb[:, 2],
lbdvrpmllpmbb[:, 4], lbdvrpmllpmbb[:, 5], lbdvrpmllpmbb[:, 3]
]).T
elif radec:
#Further transform to ra,dec,pmra,pmdec
radec = coords.lb_to_radec(lbdvrpmllpmbb[:, 0],
lbdvrpmllpmbb[:, 1],
degree=True,
epoch=None)
pmrapmdec = coords.pmllpmbb_to_pmrapmdec(lbdvrpmllpmbb[:, 4],
lbdvrpmllpmbb[:, 5],
lbdvrpmllpmbb[:, 0],
lbdvrpmllpmbb[:, 1],
degree=True,
epoch=None)
orb_vxvv = nu.array([
radec[:, 0], radec[:, 1], lbdvrpmllpmbb[:, 2], pmrapmdec[:, 0],
pmrapmdec[:, 1], lbdvrpmllpmbb[:, 3]
]).T
elif customsky:
#Further transform to ra,dec,pmra,pmdec
customradec = lb_to_customsky(lbdvrpmllpmbb[:, 0],
lbdvrpmllpmbb[:, 1],
degree=True)
custompmrapmdec = pmllpmbb_to_customsky(lbdvrpmllpmbb[:, 4],
lbdvrpmllpmbb[:, 5],
lbdvrpmllpmbb[:, 0],
lbdvrpmllpmbb[:, 1],
degree=True)
orb_vxvv = nu.array([
customradec[:, 0], customradec[:, 1], lbdvrpmllpmbb[:, 2],
custompmrapmdec[:, 0], custompmrapmdec[:, 1], lbdvrpmllpmbb[:,
3]
]).T
else:
#shape=(2tintJ-1,6)
orb_vxvv = nu.array([
iR.flatten(),
ivR.flatten(),
ivT.flatten(),
iz.flatten(),
ivz.flatten(),
iphi.flatten()
]).T
out = 0.
for ii in range(vxvv.shape[0]):
sub_vxvv = (orb_vxvv - vxvv[ii, :].flatten())**2.
#print(sub_vxvv[nu.argmin(nu.sum(sub_vxvv,axis=1))])
if not vxvv_err is None:
sub_vxvv /= vxvv_err[ii, :]**2.
else:
sub_vxvv /= 0.01**2.
out += logsumexp(-0.5 * nu.sum(sub_vxvv, axis=1))
return -out
def _fit_orbit_mlogl(new_vxvv,vxvv,vxvv_err,pot,radec,lb,tmockAA,
ro,vo,obs):
"""The log likelihood for fitting an orbit"""
#Use this _parse_args routine, which does forward and backward integration
iR,ivR,ivT,iz,ivz,iphi= tmockAA._parse_args(True,False,
new_vxvv[0],
new_vxvv[1],
new_vxvv[2],
new_vxvv[3],
new_vxvv[4],
new_vxvv[5])
if radec or lb:
#Need to transform to ra,dec
#First transform to X,Y,Z,vX,vY,vZ (Galactic)
X,Y,Z = coords.galcencyl_to_XYZ(iR.flatten(),iphi.flatten(),
iz.flatten(),
Xsun=obs[0]/ro,
Ysun=obs[1]/ro,
Zsun=obs[2]/ro)
vX,vY,vZ = coords.galcencyl_to_vxvyvz(ivR.flatten(),ivT.flatten(),
ivz.flatten(),iphi.flatten(),
vsun=nu.array(\
obs[3:6])/vo)
bad_indx= (X == 0.)*(Y == 0.)*(Z == 0.)
if True in bad_indx: X[bad_indx]+= ro/10000.
lbdvrpmllpmbb= coords.rectgal_to_sphergal(X*ro,Y*ro,Z*ro,
vX*vo,vY*vo,vZ*vo,
degree=True)
if lb:
orb_vxvv= nu.array([lbdvrpmllpmbb[:,0],
lbdvrpmllpmbb[:,1],
lbdvrpmllpmbb[:,2],
lbdvrpmllpmbb[:,4],
lbdvrpmllpmbb[:,5],
lbdvrpmllpmbb[:,3]]).T
else:
#Further transform to ra,dec,pmra,pmdec
radec= coords.lb_to_radec(lbdvrpmllpmbb[:,0],
lbdvrpmllpmbb[:,1],degree=True)
pmrapmdec= coords.pmllpmbb_to_pmrapmdec(lbdvrpmllpmbb[:,4],
lbdvrpmllpmbb[:,5],
lbdvrpmllpmbb[:,0],
lbdvrpmllpmbb[:,1],
degree=True)
orb_vxvv= nu.array([radec[:,0],radec[:,1],
lbdvrpmllpmbb[:,2],
pmrapmdec[:,0],pmrapmdec[:,1],
lbdvrpmllpmbb[:,3]]).T
else:
#shape=(2tintJ-1,6)
orb_vxvv= nu.array([iR.flatten(),ivR.flatten(),ivT.flatten(),
iz.flatten(),ivz.flatten(),iphi.flatten()]).T
out= 0.
for ii in range(vxvv.shape[0]):
sub_vxvv= (orb_vxvv-vxvv[ii,:].flatten())**2.
#print sub_vxvv[nu.argmin(nu.sum(sub_vxvv,axis=1))]
if not vxvv_err is None:
sub_vxvv/= vxvv_err[ii,:]**2.
else:
sub_vxvv/= 0.01**2.
out+= logsumexp(-0.5*nu.sum(sub_vxvv,axis=1))
return -out
def fakeDFData(binned,qdf,ii,params,fehs,afes,options,
rmin,rmax,
platelb,
grmin,grmax,
fehrange,
colordist,
fehdist,feh,sf,
mapfehs,mapafes,
ro=None,vo=None,
ndata=None,#If set, supersedes binned, only to be used w/ returnlist=True
returnlist=False): #last one useful for pixelFitDF normintstuff
if ro is None:
ro= get_ro(params,options)
if vo is None:
vo= get_vo(params,options,len(fehs))
thishr= qdf.estimate_hr(1.,z=0.125)*_REFR0*ro #qdf._hr*_REFR0*ro
thishz= qdf.estimate_hz(1.,z=0.125)*_REFR0*ro
if thishr < 0.: thishr= 10. #Probably close to flat
if thishz < 0.1: thishz= 0.2
thissr= qdf._sr*_REFV0*vo
thissz= qdf._sz*_REFV0*vo
thishsr= qdf._hsr*_REFR0*ro
thishsz= qdf._hsz*_REFR0*ro
if True:
if options.aAmethod.lower() == 'staeckel':
#Make everything 10% larger
thishr*= 1.2
thishz*= 1.2
thishsr*= 1.2
thishsz*= 1.2
thissr*= 2.
thissz*= 2.
else:
#Make everything 20% larger
thishr*= 1.2
thishz*= 1.2
thishsr*= 1.2
thishsz*= 1.2
thissr*= 2.
thissz*= 2.
#Find nearest mono-abundance bin that has a measurement
abindx= numpy.argmin((fehs[ii]-mapfehs)**2./0.01 \
+(afes[ii]-mapafes)**2./0.0025)
#Calculate the r-distribution for each plate
nrs= 1001
ngr, nfeh= 11, 11 #BOVY: INCREASE?
tgrs= numpy.linspace(grmin,grmax,ngr)
tfehs= numpy.linspace(fehrange[0]+0.00001,fehrange[1]-0.00001,nfeh)
#Calcuate FeH and gr distriutions
fehdists= numpy.zeros(nfeh)
for jj in range(nfeh): fehdists[jj]= fehdist(tfehs[jj])
fehdists= numpy.cumsum(fehdists)
fehdists/= fehdists[-1]
colordists= numpy.zeros(ngr)
for jj in range(ngr): colordists[jj]= colordist(tgrs[jj])
colordists= numpy.cumsum(colordists)
colordists/= colordists[-1]
rs= numpy.linspace(rmin,rmax,nrs)
rdists= numpy.zeros((len(sf.plates),nrs,ngr,nfeh))
#outlier model that we want to sample (not the one to aid in the sampling)
srhalo= _SRHALO/vo/_REFV0
sphihalo= _SPHIHALO/vo/_REFV0
szhalo= _SZHALO/vo/_REFV0
logoutfrac= numpy.log(get_outfrac(params,ii,options))
loghalodens= numpy.log(ro*outDens(1.,0.,None))
#Calculate surface(R=1.) for relative outlier normalization
logoutfrac+= numpy.log(qdf.surfacemass_z(1.,ngl=options.ngl))
if options.mcout:
fidoutfrac= get_outfrac(params,ii,options)
rdistsout= numpy.zeros((len(sf.plates),nrs,ngr,nfeh))
lagoutfrac= 0.15 #.0000000000000000000000001 #seems good
#Setup density model
use_real_dens= True
if use_real_dens:
#nrs, nzs= 101, 101
nrs, nzs= 64, 64
thisRmin, thisRmax= 4./_REFR0, 15./_REFR0
thiszmin, thiszmax= 0., .8
Rgrid= numpy.linspace(thisRmin,thisRmax,nrs)
zgrid= numpy.linspace(thiszmin,thiszmax,nzs)
surfgrid= numpy.empty((nrs,nzs))
for ll in range(nrs):
for jj in range(nzs):
sys.stdout.write('\r'+"Working on grid-point %i/%i" % (jj+ll*nzs+1,nzs*nrs))
sys.stdout.flush()
surfgrid[ll,jj]= qdf.density(Rgrid[ll],zgrid[jj],
nmc=options.nmcv,
ngl=options.ngl)
sys.stdout.write('\r'+_ERASESTR+'\r')
sys.stdout.flush()
if _SURFSUBTRACTEXPON:
Rs= numpy.tile(Rgrid,(nzs,1)).T
Zs= numpy.tile(zgrid,(nrs,1))
ehr= qdf.estimate_hr(1.,z=0.125)
# ehz= qdf.estimate_hz(1.,zmin=0.5,zmax=0.7)#Get large z behavior right
ehz= qdf.estimate_hz(1.,z=0.125)
surfInterp= interpolate.RectBivariateSpline(Rgrid,zgrid,
numpy.log(surfgrid)
+Rs/ehr+numpy.fabs(Zs)/ehz,
kx=3,ky=3,
s=0.)
# s=10.*float(nzs*nrs))
else:
surfInterp= interpolate.RectBivariateSpline(Rgrid,zgrid,
numpy.log(surfgrid),
kx=3,ky=3,
s=0.)
# s=10.*float(nzs*nrs))
if _SURFSUBTRACTEXPON:
compare_func= lambda x,y,du: numpy.exp(surfInterp.ev(x/ro/_REFR0,numpy.fabs(y)/ro/_REFR0)-x/ro/_REFR0/ehr-numpy.fabs(y)/ehz/ro/_REFR0)
else:
compare_func= lambda x,y,du: numpy.exp(surfInterp.ev(x/ro/_REFR0,numpy.fabs(y)/ro/_REFR0))
else:
compare_func= lambda x,y,z: fidDens(x,y,thishr,thishz,z)
for jj in range(len(sf.plates)):
p= sf.plates[jj]
sys.stdout.write('\r'+"Working on plate %i (%i/%i)" % (p,jj+1,len(sf.plates)))
sys.stdout.flush()
rdists[jj,:,:,:]= _predict_rdist_plate(rs,
compare_func,
None,rmin,rmax,
platelb[jj,0],platelb[jj,1],
grmin,grmax,
fehrange[0],fehrange[1],feh,
colordist,
fehdist,sf,sf.plates[jj],
dontmarginalizecolorfeh=True,
ngr=ngr,nfeh=nfeh)
if options.mcout:
rdistsout[jj,:,:,:]= _predict_rdist_plate(rs,
lambda x,y,z: outDens(x,y,z),
None,rmin,rmax,
platelb[jj,0],platelb[jj,1],
grmin,grmax,
fehrange[0],fehrange[1],feh,
colordist,
fehdist,sf,sf.plates[jj],
dontmarginalizecolorfeh=True,
ngr=ngr,nfeh=nfeh)
sys.stdout.write('\r'+_ERASESTR+'\r')
sys.stdout.flush()
numbers= numpy.sum(rdists,axis=3)
numbers= numpy.sum(numbers,axis=2)
numbers= numpy.sum(numbers,axis=1)
numbers= numpy.cumsum(numbers)
if options.mcout:
totfid= numbers[-1]
numbers/= numbers[-1]
rdists= numpy.cumsum(rdists,axis=1)
for ll in range(len(sf.plates)):
for jj in range(ngr):
for kk in range(nfeh):
rdists[ll,:,jj,kk]/= rdists[ll,-1,jj,kk]
if options.mcout:
numbersout= numpy.sum(rdistsout,axis=3)
numbersout= numpy.sum(numbersout,axis=2)
numbersout= numpy.sum(numbersout,axis=1)
numbersout= numpy.cumsum(numbersout)
totout= fidoutfrac*numbersout[-1]
totnumbers= totfid+totout
totfid/= totnumbers
totout/= totnumbers
if _DEBUG:
print totfid, totout
numbersout/= numbersout[-1]
rdistsout= numpy.cumsum(rdistsout,axis=1)
for ll in range(len(sf.plates)):
for jj in range(ngr):
for kk in range(nfeh):
rdistsout[ll,:,jj,kk]/= rdistsout[ll,-1,jj,kk]
#Now sample
thisout= []
newrs= []
newls= []
newbs= []
newplate= []
newgr= []
newfeh= []
newds= []
newzs= []
newvas= []
newRs= []
newphi= []
newvr= []
newvt= []
newvz= []
newlogratio= []
newfideval= []
newqdfeval= []
newpropeval= []
if ndata is None:
thisdata= binned(fehs[ii],afes[ii])
thisdataIndx= binned.callIndx(fehs[ii],afes[ii])
ndata= len(thisdata)
#First sample from spatial density
for ll in range(ndata):
#First sample a plate
ran= numpy.random.uniform()
kk= 0
while numbers[kk] < ran: kk+= 1
#Also sample a FeH and a color
ran= numpy.random.uniform()
ff= 0
while fehdists[ff] < ran: ff+= 1
ran= numpy.random.uniform()
cc= 0
while colordists[cc] < ran: cc+= 1
#plate==kk, feh=ff,color=cc; now sample from the rdist of this plate
ran= numpy.random.uniform()
jj= 0
if options.mcout and numpy.random.uniform() < totout: #outlier
while rdistsout[kk,jj,cc,ff] < ran: jj+= 1
thisoutlier= True
else:
while rdists[kk,jj,cc,ff] < ran: jj+= 1
thisoutlier= False
#r=jj
newrs.append(rs[jj])
newls.append(platelb[kk,0])
newbs.append(platelb[kk,1])
newplate.append(sf.plates[kk])
newgr.append(tgrs[cc])
newfeh.append(tfehs[ff])
dist= _ivezic_dist(tgrs[cc],rs[jj],tfehs[ff])
newds.append(dist)
#calculate R,z
XYZ= bovy_coords.lbd_to_XYZ(platelb[kk,0],platelb[kk,1],
dist,degree=True)
R= ((_REFR0-XYZ[0])**2.+XYZ[1]**2.)**(0.5)
newRs.append(R)
phi= numpy.arcsin(XYZ[1]/R)
if (_REFR0-XYZ[0]) < 0.:
phi= numpy.pi-phi
newphi.append(phi)
z= XYZ[2]+_ZSUN
newzs.append(z)
newrs= numpy.array(newrs)
newls= numpy.array(newls)
newbs= numpy.array(newbs)
newplate= numpy.array(newplate)
newgr= numpy.array(newgr)
newfeh= numpy.array(newfeh)
newds= numpy.array(newds)
newRs= numpy.array(newRs)
newzs= numpy.array(newzs)
newphi= numpy.array(newphi)
#Add mock velocities
newvr= numpy.empty_like(newrs)
newvt= numpy.empty_like(newrs)
newvz= numpy.empty_like(newrs)
use_sampleV= True
if use_sampleV:
for kk in range(ndata):
newv= qdf.sampleV(newRs[kk]/_REFR0,newzs[kk]/_REFR0,n=1)
newvr[kk]= newv[0,0]*_REFV0*vo
newvt[kk]= newv[0,1]*_REFV0*vo
newvz[kk]= newv[0,2]*_REFV0*vo
else:
accept_v= numpy.zeros(ndata,dtype='bool')
naccept= numpy.sum(accept_v)
sigz= thissz*numpy.exp(-(newRs-_REFR0)/thishsz)
sigr= thissr*numpy.exp(-(newRs-_REFR0)/thishsr)
va= numpy.empty_like(newrs)
sigphi= numpy.empty_like(newrs)
maxqdf= numpy.empty_like(newrs)
nvt= 101
tvt= numpy.linspace(0.1,1.2,nvt)
for kk in range(ndata):
#evaluate qdf for vt
pvt= qdf(newRs[kk]/ro/_REFR0+numpy.zeros(nvt),
numpy.zeros(nvt),
tvt,
newzs[kk]/ro/_REFR0+numpy.zeros(nvt),
numpy.zeros(nvt),log=True)
pvt_maxindx= numpy.argmax(pvt)
va[kk]= (1.-tvt[pvt_maxindx])*_REFV0*vo
if options.aAmethod.lower() == 'adiabaticgrid' and options.flatten >= 0.9:
maxqdf[kk]= pvt[pvt_maxindx]+numpy.log(250.)
elif options.aAmethod.lower() == 'adiabaticgrid' and options.flatten >= 0.8:
maxqdf[kk]= pvt[pvt_maxindx]+numpy.log(250.)
else:
maxqdf[kk]= pvt[pvt_maxindx]+numpy.log(40.)
sigphi[kk]= _REFV0*vo*4.*numpy.sqrt(numpy.sum(numpy.exp(pvt)*tvt**2.)/numpy.sum(numpy.exp(pvt))-(numpy.sum(numpy.exp(pvt)*tvt)/numpy.sum(numpy.exp(pvt)))**2.)
ntries= 0
ngtr1= 0
while naccept < ndata:
sys.stdout.write('\r %i %i %i \r' % (ntries,naccept,ndata))
sys.stdout.flush()
#print ntries, naccept, ndata
ntries+= 1
accept_v_comp= True-accept_v
prop_vr= numpy.random.normal(size=ndata)*sigr
prop_vt= numpy.random.normal(size=ndata)*sigphi+vo*_REFV0-va
prop_vz= numpy.random.normal(size=ndata)*sigz
qoverp= numpy.zeros(ndata)-numpy.finfo(numpy.dtype(numpy.float64)).max
qoverp[accept_v_comp]= (qdf(newRs[accept_v_comp]/ro/_REFR0,
prop_vr[accept_v_comp]/vo/_REFV0,
prop_vt[accept_v_comp]/vo/_REFV0,
newzs[accept_v_comp]/ro/_REFR0,
prop_vz[accept_v_comp]/vo/_REFV0,log=True)
-maxqdf[accept_v_comp] #normalize max to 1
-(-0.5*(prop_vr[accept_v_comp]**2./sigr[accept_v_comp]**2.+prop_vz[accept_v_comp]**2./sigz[accept_v_comp]**2.+(prop_vt[accept_v_comp]-_REFV0*vo+va[accept_v_comp])**2./sigphi[accept_v_comp]**2.)))
if numpy.any(qoverp > 0.):
ngtr1+= numpy.sum((qoverp > 0.))
if ngtr1 > 5:
qindx= (qoverp > 0.)
print naccept, ndata, newRs[qindx], newzs[qindx], prop_vr[qindx], va[qindx], sigphi[qindx], prop_vt[qindx], prop_vz[qindx], qoverp[qindx]
raise RuntimeError("max qoverp = %f > 1, but shouldn't be" % (numpy.exp(numpy.amax(qoverp))))
accept_these= numpy.log(numpy.random.uniform(size=ndata))
#print accept_these, (accept_these < qoverp)
accept_these= (accept_these < qoverp)
newvr[accept_these]= prop_vr[accept_these]
newvt[accept_these]= prop_vt[accept_these]
newvz[accept_these]= prop_vz[accept_these]
accept_v[accept_these]= True
naccept= numpy.sum(accept_v)
sys.stdout.write('\r'+_ERASESTR+'\r')
sys.stdout.flush()
"""
ntot= 0
nsamples= 0
itt= 0
fracsuccess= 0.
fraccomplete= 0.
while fraccomplete < 1.:
if itt == 0:
nthis= numpy.amax([ndata,_NMIN])
else:
nthis= int(numpy.ceil((1-fraccomplete)/fracsuccess*ndata))
itt+= 1
count= 0
while count < nthis:
count+= 1
sigz= thissz*numpy.exp(-(R-_REFR0)/thishsz)
sigr= thissr*numpy.exp(-(R-_REFR0)/thishsr)
sigphi= sigr #/numpy.sqrt(2.) #BOVY: FOR NOW
#Estimate asymmetric drift
va= sigr**2./2./_REFV0/vo\
*(-.5+R*(1./thishr+2./thishsr))+10.*numpy.fabs(z)
newvas.append(va)
if options.mcout and thisoutlier:
#Sample from outlier gaussian
newvz.append(numpy.random.normal()*_SZHALOFAKE*2.)
newvr.append(numpy.random.normal()*_SRHALOFAKE*2.)
newvt.append(numpy.random.normal()*_SPHIHALOFAKE*2.)
elif numpy.random.uniform() < lagoutfrac:
#Sample from lagging gaussian
newvz.append(numpy.random.normal()*_SZHALOFAKE)
newvr.append(numpy.random.normal()*_SRHALOFAKE)
newvt.append(numpy.random.normal()*_SPHIHALOFAKE*2.+_REFV0*vo/4.)
else:
#Sample from disk gaussian
newvz.append(numpy.random.normal()*sigz)
newvr.append(numpy.random.normal()*sigr)
newvt.append(numpy.random.normal()*sigphi+_REFV0*vo-va)
newlogratio= list(newlogratio)
fidlogeval= numpy.log(1.-lagoutfrac)\
-numpy.log(sigr)-numpy.log(sigphi)-numpy.log(sigz)-0.5*(newvr[-1]**2./sigr**2.+newvz[-1]**2./sigz**2.+(newvt[-1]-_REFV0*vo+va)**2./sigphi**2.)
lagoutlogeval= numpy.log(lagoutfrac)\
-numpy.log(_SRHALOFAKE)\
-numpy.log(_SPHIHALOFAKE*2.)\
-numpy.log(_SZHALOFAKE)\
-0.5*(newvr[-1]**2./_SRHALOFAKE**2.+newvz[-1]**2./_SZHALOFAKE**2.+(newvt[-1]-_REFV0*vo/4.)**2./_SPHIHALOFAKE**2./4.)
if use_real_dens:
fidlogeval+= numpy.log(compare_func(R,z,None)[0])
lagoutlogeval+= numpy.log(compare_func(R,z,None)[0])
else:
fidlogeval+= numpy.log(fidDens(R,z,thishr,thishz,None))
lagoutlogeval+= numpy.log(fidDens(R,z,thishr,thishz,None))
newfideval.append(fidlogeval)
if options.mcout:
fidoutlogeval= numpy.log(fidoutfrac)\
+numpy.log(outDens(R,z,None))\
-numpy.log(_SRHALOFAKE*2.)\
-numpy.log(_SPHIHALOFAKE*2.)\
-numpy.log(_SZHALOFAKE*2.)\
-0.5*(newvr[-1]**2./_SRHALOFAKE**2./4.+newvz[-1]**2./_SZHALOFAKE**2./4.+newvt[-1]**2./_SPHIHALOFAKE**2./4.)
newpropeval.append(logsumexp([fidoutlogeval,fidlogeval,
lagoutlogeval]))
else:
newpropeval.append(logsumexp([lagoutlogeval,fidlogeval]))
qdflogeval= qdf(R/ro/_REFR0,newvr[-1]/vo/_REFV0,newvt[-1]/vo/_REFV0,z/ro/_REFR0,newvz[-1]/vo/_REFV0,log=True)
if isinstance(qdflogeval,(list,numpy.ndarray)):
qdflogeval= qdflogeval[0]
if options.mcout:
outlogeval= logoutfrac+loghalodens\
-numpy.log(srhalo)-numpy.log(sphihalo)-numpy.log(szhalo)\
-0.5*((newvr[-1]/vo/_REFV0)**2./srhalo**2.+(newvz[-1]/vo/_REFV0)**2./szhalo**2.+(newvt[-1]/vo/_REFV0)**2./sphihalo**2.)\
-1.5*numpy.log(2.*numpy.pi)
newqdfeval.append(logsumexp([qdflogeval,outlogeval]))
else:
newqdfeval.append(qdflogeval)
newlogratio.append(qdflogeval
-newpropeval[-1])#logsumexp([fidlogeval,fidoutlogeval]))
newlogratio= numpy.array(newlogratio)
thisnewlogratio= copy.copy(newlogratio)
maxnewlogratio= numpy.amax(thisnewlogratio)
if False:
argsort_thisnewlogratio= numpy.argsort(thisnewlogratio)[::-1]
thisnewlogratio-= thisnewlogratio[argsort_thisnewlogratio[2]] #3rd largest
else:
thisnewlogratio-= numpy.amax(thisnewlogratio)
thisnewratio= numpy.exp(thisnewlogratio)
if len(thisnewratio.shape) > 1 and thisnewratio.shape[1] == 1:
thisnewratio= numpy.reshape(thisnewratio,(thisnewratio.shape[0]))
#Rejection sample
accept= numpy.random.uniform(size=len(thisnewratio))
accept= (accept < thisnewratio)
fraccomplete= float(numpy.sum(accept))/ndata
fracsuccess= float(numpy.sum(accept))/len(thisnewratio)
if _DEBUG:
print fraccomplete, fracsuccess, ndata
print numpy.histogram(thisnewratio,bins=16)
indx= numpy.argmax(thisnewratio)
print numpy.array(newvr)[indx], \
numpy.array(newvt)[indx], \
numpy.array(newvz)[indx], \
numpy.array(newrs)[indx], \
numpy.array(newds)[indx], \
numpy.array(newls)[indx], \
numpy.array(newbs)[indx], \
numpy.array(newfideval)[indx]
bovy_plot.bovy_print()
bovy_plot.bovy_plot(numpy.array(newvt),
numpy.exp(numpy.array(newqdfeval)),'b,',
xrange=[-300.,500.],yrange=[0.,1.])
bovy_plot.bovy_plot(newvt,
numpy.exp(numpy.array(newpropeval+maxnewlogratio)),
'g,',
overplot=True)
bovy_plot.bovy_plot(numpy.array(newvt),
numpy.exp(numpy.array(newlogratio-maxnewlogratio)),
'b,',
xrange=[-300.,500.],
# xrange=[0.,20.],
# xrange=[0.,3.],
# xrange=[6.,9.],
yrange=[0.001,1.],semilogy=True)
bovy_plot.bovy_end_print('/home/bovy/public_html/segue-local/test.png')
#Now collect the samples
newrs= numpy.array(newrs)[accept][0:ndata]
newls= numpy.array(newls)[accept][0:ndata]
newbs= numpy.array(newbs)[accept][0:ndata]
newplate= numpy.array(newplate)[accept][0:ndata]
newgr= numpy.array(newgr)[accept][0:ndata]
newfeh= numpy.array(newfeh)[accept][0:ndata]
newvr= numpy.array(newvr)[accept][0:ndata]
newvt= numpy.array(newvt)[accept][0:ndata]
newvz= numpy.array(newvz)[accept][0:ndata]
newphi= numpy.array(newphi)[accept][0:ndata]
newds= numpy.array(newds)[accept][0:ndata]
newqdfeval= numpy.array(newqdfeval)[accept][0:ndata]
"""
vx, vy, vz= bovy_coords.galcencyl_to_vxvyvz(newvr,newvt,newvz,newphi,
vsun=[_VRSUN,_VTSUN,_VZSUN])
vrpmllpmbb= bovy_coords.vxvyvz_to_vrpmllpmbb(vx,vy,vz,newls,newbs,newds,
XYZ=False,degree=True)
pmrapmdec= bovy_coords.pmllpmbb_to_pmrapmdec(vrpmllpmbb[:,1],
vrpmllpmbb[:,2],
newls,newbs,
degree=True)
#Dump everything for debugging the coordinate transformation
from galpy.util import save_pickles
save_pickles('dump.sav',
newds,newls,newbs,newphi,
newvr,newvt,newvz,
vx, vy, vz,
vrpmllpmbb,
pmrapmdec)
if returnlist:
out= []
for ii in range(ndata):
out.append([newrs[ii],
newgr[ii],
newfeh[ii],
newls[ii],
newbs[ii],
newplate[ii],
newds[ii],
False, #outlier?
vrpmllpmbb[ii,0],
vrpmllpmbb[ii,1],
vrpmllpmbb[ii,2]])#,
# newqdfeval[ii]])
return out
#Load into data
binned.data.feh[thisdataIndx]= newfeh
oldgr= thisdata.dered_g-thisdata.dered_r
oldr= thisdata.dered_r
if options.noerrs:
binned.data.dered_r[thisdataIndx]= newrs
else:
binned.data.dered_r[thisdataIndx]= newrs\
+numpy.random.normal(size=numpy.sum(thisdataIndx))\
*ivezic_dist_gr(oldgr,0., #g-r is all that counts
binned.data.feh[thisdataIndx],
dg=binned.data[thisdataIndx].g_err,
dr=binned.data[thisdataIndx].r_err,
dfeh=binned.data[thisdataIndx].feh_err,
return_error=True,
_returndmr=True)
binned.data.dered_r[(binned.data.dered_r >= rmax)]= rmax #tweak to make sure everything stays within the observed range
if False:
binned.data.dered_r[(binned.data.dered_r <= rmin)]= rmin
binned.data.dered_g[thisdataIndx]= oldgr+binned.data[thisdataIndx].dered_r
#Also change plate and l and b
binned.data.plate[thisdataIndx]= newplate
radec= bovy_coords.lb_to_radec(newls,newbs,degree=True)
binned.data.ra[thisdataIndx]= radec[:,0]
binned.data.dec[thisdataIndx]= radec[:,1]
binned.data.l[thisdataIndx]= newls
binned.data.b[thisdataIndx]= newbs
if options.noerrs:
binned.data.vr[thisdataIndx]= vrpmllpmbb[:,0]
binned.data.pmra[thisdataIndx]= pmrapmdec[:,0]
binned.data.pmdec[thisdataIndx]= pmrapmdec[:,1]
else:
binned.data.vr[thisdataIndx]= vrpmllpmbb[:,0]+numpy.random.normal(size=numpy.sum(thisdataIndx))*binned.data.vr_err[thisdataIndx]
binned.data.pmra[thisdataIndx]= pmrapmdec[:,0]+numpy.random.normal(size=numpy.sum(thisdataIndx))*binned.data.pmra_err[thisdataIndx]
binned.data.pmdec[thisdataIndx]= pmrapmdec[:,1]+numpy.random.normal(size=numpy.sum(thisdataIndx))*binned.data.pmdec_err[thisdataIndx]
return binned
