No subject

he stream lines produced with these filters take into account the sign and =
direction of the vector. =0A=0AI need to use a version where only the direc=
tion of the vector and not the sign is taken into account. That is, our vec=
tors are headless / non-polar. Thus for these two (linear / parallel / anti=
-parallel) cases, the streamline would be the same for my data:=0A=0A---> <=
---=0A=0A---> --->=0A=0AObviously with the filter as is, the first one stop=
s at the mid-point but the second is continuous.=0A=0AOur data is not unidi=
rectional (the orientation of the field varies with space) and so I cannot =
just make sure that one of the components (say the z) of the vector in the =
data set is always positive as this always leads to breaks in the streamlin=
es at some point.=0A=0AThanks in advance if anyone has a tip how to deal wi=
th this.=0A=0AMartin=0A=0A=0A