def altoptm(l,df,zs):
m= numpy.array([l[0],l[1]])
V= numpy.array([[l[2],l[4]],[l[4],l[3]]])
a= numpy.array([l[5],l[6]])
return numpy.sum((multiskewnormal(zs,m,V,a)-df)**2.)
V11= 0.2**2./2.
a0= 0.
a1= 3.
l= optimize.fmin(optm,[m0,m1,V00,V11,V01,a0,a1],(vs,))
#g= optimize.fmin(gaussoptm,[m0,m1,V00,V11,V01],(vs,))
altdf= grid.df.flatten()
altdf/= numpy.sum(altdf)*(grid.vRgrid[1]-grid.vRgrid[0])\
*(grid.vTgrid[1]-grid.vTgrid[0])
h= optimize.fmin(altoptm,[m0,m1,V00,V11,V01,a0,a1],(altdf,vzs,))
m= numpy.array([l[0],l[1]])
V= numpy.array([[l[2],l[4]],[l[4],l[3]]])
a= numpy.array([l[5],l[6]])
ma= numpy.array([h[0],h[1]])
Va= numpy.array([[h[2],h[4]],[h[4],h[3]]])
aa= numpy.array([h[5],h[6]])
X= multiskewnormal(zs,m,V,a)
X= numpy.reshape(X,(len(xs),len(ys)))
#Also cumulative for contouring
sortindx= numpy.argsort(X.flatten())[::-1]
cumul= numpy.cumsum(numpy.sort(X.flatten())[::-1])/numpy.sum(X.flatten())
cntrThis= numpy.zeros(numpy.prod(X.shape))
cntrThis[sortindx]= cumul
cntrThis= numpy.reshape(cntrThis,X.shape)
Z= multiskewnormal(zs,ma,Va,aa)
Z= numpy.reshape(Z,(len(xs),len(ys)))
#Also cumulative for contouring
sortindxa= numpy.argsort(Z.flatten())[::-1]
cumula= numpy.cumsum(numpy.sort(Z.flatten())[::-1])/numpy.sum(Z.flatten())
cntrThisa= numpy.zeros(numpy.prod(Z.shape))
cntrThisa[sortindxa]= cumula
cntrThisa= numpy.reshape(cntrThisa,Z.shape)