►
From YouTube: SunPy Community Meeting (29/6)
Description
B
C
C
F
F
Ok
good
morning,
well,
I
implemented
some
plots
or
I've
wanted
some
parts
of
the
code
that
have
allowed
me
to
create
plots
that
I
put
on
my
blog
in
the
past
week.
That
are
the
past
few
days.
I've
been
repacking
a
bit
with
my
mentors
suggestions
such
as
I
pulled
out
the
file
or
the
function
that
reads:
good
dog,
jennex
instrument
files
outside
of
the
response
function
that
I'm
making
or
the
amount
a
lot
of
making,
and
so
that
will
be
a
separate.
F
And
I'm
just
making
sure
the
game
part
of
my
wavelength
response
is
correct
this
morning
and
other
than
that
I've
plan
to
move
on
to
the
anisa
tivity
part,
which
is
going
to
be
the
shanty
pi
starting
to
dive
into
that.
We
kind
of
discussed
that
last
week
a
little
bit
but
yeah
for
the
most
part.
So
far,
I've
just
been
refactoring
this
week,
I
haven't
changed
too
much
or
moved
forward.
As
far
as
that
is
concerned,
you.
B
G
F
C
C
C
F
That's
actually
something
I've
been
testing
out
this
week,
too
I
was
using
a
data
frame,
or
I
have
been
using
a
data
frame
that
I
both
the
information
for
this
function
into,
but
I'm
looking
into
a
strip
height
able
to
use
instead
I'm
not
as
familiar
with
loading
it
and
getting
it
at
one
index
or
one
element
in
there,
but
they
they
house
units.
So
it
may
be
a
better
structure,
proud
of
testing
that,
at
the
current
time,.
H
F
A
E
B
C
D
On
it,
I
need
to
add
an
extension
that
seems
to
have
done.
Fine,
no
questions,
so
it's
great
to
yeah
I.
So
can
people
see
my
screen?
Yes
right,
okay!
Well,
let's
first
will
restart
the
colonel
see
you
don't
have
all
my
previous
rubbish
so
for
those
who
need
reminding.
Obviously
this
is
the
time
series
class
which
is
meant
to
be
replacing
the
light
curve.
So
it's
kind
of
more
more
generic,
more
powerful.
D
It's
worth
me
saying
at
the
opposite:
there
are
going
to
be.
There
are
some
features
of
light
curve
so,
for
instance,
downloading
data
that
is
not
part
of
the
time
series
that
we
partay
unit
down.
So
it
is
we're
saying
that
has
been
cut
out
at
the
moment,
which
is
why
in
I
have
an
example
gallery
which
is
online
on
on
my?
Is
it
my
fork
or
branch
icon
for
the
terminology,
but
within
that
it
does
actually
include
a
reference
to
light
curve
just
to
download
some
sample
data?
D
So
please
ignore
that
fact
that
will
obviously
be
fine.
What's
you
need
out
of
that
anyway,
but
essentially,
if
I
start
running
for
ooh
some
of
the
functionality
that's
currently
available,
and
the
big
thing
is
I
sort
of
want
at
this
point,
we
would
start
playing
around
with
it
breaking
it
getting
into
the
debate
on
how
we
get
it
running.
So
the
primary
thing
is
that
there's
a
subclass
for
every
source.
D
We
have
the
same
kind
of
structures
with
map,
so
we
have
a
time
series
factory
which
will
take
a
series
of
different
inputs,
and
essentially
it
should
be
reasonably
versatile.
But
if
I
go
for
the
very
basic
which
is,
if
I
give
it
some
sample
data
for
the
eve,
then
of
course
I've
got
now
an
eve
time.
So
we
could
we
can
peek
at
the
moment.
This
is
just
stealing
the
code
directly,
so
you
can
peek
as
normal.
If
I
should
do.
I
D
D
Of
course,
that's
just
a
list
of
light
curves
each
which
can
be
accessed
and
and
a
figure
comes
over
into
the
other
screen.
But
so
there
is
your
figure,
but
you
can
also
you
can
do
it
or
see
more
files.
If
you
want,
you
can
also
concatenate
now
I'll
show
you
the
concatenation
function
in
a
second.
Essentially,
this
at
the
moment
has
just
can
return
a
single
time
series
which
I
see
you
can't.
D
D
Obviously,
there
are
potentially
times
when
that
might
happen,
but
generally
that
shouldn't
be
a
major
issue
in
the
sense
that
it's
unusually
you'll
have
two
files
with
the
same
time
stamp
for
the
same
type
of
data
and
yet
two
different
values,
but
that
is
a
so
they
can
catenate
function
which
is
independent.
Obviously,
this
just
calls
the
concatenate
function
because
it
set
to
concatenate.
True
the
concatenate
function
itself
is
a
little
bit
a
dark
horse,
so
am
I
being
yeah.
D
I
mean,
if
I
look
at,
might
as
well
show
you
seeing
as
we've
already
sort
of
mentioned,
that
the
meta
in
here
at
the
moment
I'll
give
you
a
bit
more
space
with
you,
so
for
only
two
concatenated
to
time
series.
It's
reasonably
easy.
You've
got
the
the
first
metadata
and
the
second
metadata
all
stacked
within
your
metadata,
but
because
it's
an
iterative
process
at
the
moment
you
only
concatenate
one
at
a
time.
So,
if
you're
doing
a
concatenation
of
three,
it
will
do
the
equivalent
of
freaking
catenation
duplicate
the
nation
commands.
G
D
Done
and
then
I
bring
up
the
matter
on
that
you
will
see
what
is
currently
a
bit
of
a
problem
Oh.
Now
what
we've
got
is
we've
got
again.
We
got
the
two
nested
meta
from
the
first
concatenation
all
nested
in
the
second
congratulation
as
the
first
position
and
then
obviously
there's
a
the
second
one
is
nested
there.
So
you
can
imagine
this
is
going
to
get
unwieldy
if
you
could
get
to
me
for
five
different
times,
but
that
is
something
that
we
definitely
need
to
consider.
D
Danny's,
Joe
and
I
have
been
discussing
it,
but
the
problem
is,
we
haven't,
found
sort
of
a
single
good
solution
and
slowly
but
surely
our
solutions
are
getting
more
more
complicated,
so
that
might
require
a
bit
more
talking
about
actual
usage
case
before
we
can
put
a
decent,
but
the
main
functionalities
that
you
can
concatenate.
You
can
actually
concatenate
I,
don't
demonstrate
at
the
moment,
because
it's
a
little
bit
less
reliable.
You
can
concatenate
not
only
the
same
data
source,
but
you
can
catenate
different
data
sources.
D
D
We
will
be
doing
that,
but
obviously
that
potentially
won't
work
for
everything,
because
not
every
file
type
is
apparently
obvious
what
data
source
it
is
and
then
equally
there's
we've
kind
of
assumed
that
you
will
only
ever
used
one
data
source
when
instantiating
the
construct,
the
factory,
so
it's
kind
of
a
limitation,
but
it's
just
because
otherwise
it
would
be
so
complicated
because,
unlike
map
we're
pretty
much
everything
so
Fitz
files-
and
you
know
what
source
of
data
you
have
with
this,
you
done.
You
have
to
explicitly
stated
potentially
at.
C
One
thing
they're
so
for
plotting
a
into
something,
so
you
can
plot
together
two
different
light:
curves
time
serious
instances
right.
Yes,
so
it
doesn't
really
matter.
So
if
I
want
to
plug
two
instruments,
I
can
just
I,
say
I,
think
map
I
can
do
a
map
and
then
do
I
a
composite
map.
Would
you
be
like
for
something
which
is
different
and
touristic
I
could
do
the
same
with
with
my
curves.
If
I
map
understand
right
right.
C
D
Yeah
I
do
remember
low,
and
this
is
where,
at
the
moment,
you
can
actually
create
a
manual
time
series
based
on
just
placing
in
your
data
and
I
think
where
I
actually
put
the
the
methodology
to
do
it,
but
you
can
obviously,
if
you
wanted
to
create
a
more
advanced
plot,
and
this
is
one
thing
where
concatenation
will
work
in
a
lot
of
ways,
but
at
some
point,
if
you're
gettin
a
ting
a
lot
of
different
data,
it's
actually
kind
of
more
useful
I
think
to
start
thinking.
Okay!
D
Well,
why
don't
I
create
a
new
generic
time
series
rather
than
trying
to
use
one
of
the
source?
You
know
it's
a
que,
canta
nating
goes
data
to
Eve
data
may
cause
more
problems
than
just
getting
up
and
saying:
okay,
well,
I
can
get
the
ghost
data.
I
can
get
the
eve
data.
What's
usually
time
series
like
and
then
extract
the
data
is
Panthers,
combine
those
and
then
create
an
entirely
new
time
series,
which
is
my
own
time
series
for
whatever
I'm
analyzing
so
but
that's
obviously
a
question
of
use
case.
D
So,
within
the
within
inject
itself,
you've
got
access
to
all
of
the
primary
things,
so
the
data
which
at
the
moment
it's
using
and
data
frames
of
Siam,
actually
loaded
library,
so
obviously
the
the
debate
about
whether
to
use
anything
asteroid
pie
unfortunate.
In
the
moment
there
is
no
suitable
data
structure
so,
and
the
beta
frames
is
what
we
got.
You've
obviously
also
got
the
so
metadata
which
again
can
get
confusing
if
you
start
going
as
amazing,
but
for
a
simple
perspective,
the
metadata
is
just
as
normal.
We
also
have
a
units
you
fell
now.
D
This
is
a
dictionary,
and
this
is
basically
the
the
way
that
we
currently
bolt
on
unit
support
to
to
the
time
series.
So
I'll
explain
that
a
bit
more
in
detail
later,
but
essentially
all
it
does,
is
it
matches
the
column
headers
to
an
astro
my
unit
to
type
and
also
you
could
change
it
to
spy
it's
a
dictionary
at
the
end
of
the
day.
D
So
again
we
have
access
to
more
specific
details,
but
most
of
these
haven't
been
populated.
This
is
actually
a
part
where
it's
actually
a
part
where
we
start
to
neat
deciding
use
case,
and
what
of
these
particular
values
might
be
useful,
so
I've
essentially
stolen
them,
mostly
from
the
map
class,
and
we
want
to
decide
whether
we
want
to
keep
and
get
rid
of
them
so
that
that's
all
a
part
of
the
debate,
the
ongoing
debate.
D
Obviously
it's
worth
saying
that
with
the
plots
you
can
pop
them
with
subplots
as
well,
so
it's
essentially
exactly
the
same
plotting
mechanism.
It's
taking
a
second
surrender,
exactly
the
same
plotting
mechanism.
Is
you
got
the
the
old
light
curve?
So
again
I
just
posted
that
code
over,
so
you
can
also
extract
individual
columns
if
you
want
and
then
obviously
from
that
again,
I
can
just
look
at
the
data.
D
D
I
D
It
so
okay,
so
truncation
is
one
of
the
big
features
again
missing
from
the
light
curve
series.
So
here
we
have
three
different
methods
of
truncation.
You've
got
a
truncation
via
simply
using
the
the
indexes,
so
in
this
case
it's
very
sly
singh,
except
that
it's
not
slicing.
I
think
I
think
this
is
just
yeah.
D
On
a
day
to
day
basis,
you
can
truncate
using
pics
string
using
strings
for
a
date
time
so
and
that
uses
the
Sun
pi
parser
now
to
pass
that,
and
essentially
what
it's
doing
is
it's
turning
those
into
a
date,
though,
into
a
time
range
at
some
point,
I'm
range
and
then
obviously
it
off
it
will
enable
you
to
truncate
it.
So
essentially
we
can
out
you
can
catenate
to
make
a
larger
dataset.
You
could
truncate
to
make
a
smaller
dataset.
D
D
So
did
you
do
it
right
so
say
we
have
the
original
tire
goes
day.
Two
there
and
say
we
want
to
zoom
into
a
particularly
active
time.
Then.
Obviously,
we
just
truncate
it
to
time
that
we
care
about
so
in
this
case
with
truncate
to
that
region
there,
and
you
can
see
it's
nice
or
more
legible
in
the
diagrams,
but
other
than
that
they're,
both
exactly
the
same
they're,
both
the
same
type
that
both
goes
time
series,
which
means
that
the
plotting
all
looks
reasonably
similar.
D
So
yeah,
that's
that's
obvio,
reasonably
useful
function,
one
of
the
good
things
is.
We
have
the
ability
to
down
sample
or
to
resample
full
stop.
Essentially
it
just
resample
the
data
using
the
internal
pandas
dataframe
and
whatever
those
functions
you
use.
So,
for
example,
this
is
exactly
the
same
data
I've
downsampled
it
quite
massively.
So
now
it's
much
smoother
already
look.
C
D
G
D
10
minutes
and
then
it's
saying
take
a
mean
of
those
10
minutes
and
you
can
use
you
can
use
mean
some
or
standard
deviation.
I,
don't
know
whether
there
are
other
methods,
that's
annoying.
The
the
I
didn't
notice,
the
Panthers.
We're
actually
lists
out
what
the
overall
methods
are.
But
yes,
so
when
you,
when
you're
resampling
in
this
case,
when
you,
when
you're
down
sampling
of
CES,
you
can
pick
every
whatever
time,
I
thinking
so
and
then
equally,
obviously,
you
got
the
exact
opposite,
which
is
you
could
up
a
sample?
D
And
so,
in
this
case
it's
taking
its
taking
that
same
downsampled,
which
is
every
10
minutes,
and
it's
upsampling
it
to
every
one
minute.
I,
don't
know
why
they
have
t
4
minute.
That's
just
a
notation
that
Panthers
Hughes
but
I've
tried
to
see
main
tell
you
that
it
is
it
there
is
that
no
so
T
explicitly
means
minutes.
H
means
our.
D
A
D
C
J
D
D
It's
actually
sent
to
the
pandas,
even
though
we've
written
it
with
the
notation
like
this
is
actually
their
old
style
of
notation
I.
Don't
know
why
they
changed
that
style.
I
think
it
might
be
specifically,
so
that
is
easier
for
people
to
add
their
own
functions,
essentially
like
this
you're
locked
into
whatever
they've
limited,
restricted
as
option.
J
D
D
Well,
I'll
have
a
look
at
those
worthwhile
thing
for
me
to
check
out
yes
at
the
moment.
I
don't
actually
know
beyond
using
the
standard
ones.
I
don't
actually
know,
but
it's
definitely
worthwhile
thing
to
see
whether
we
can
employ
because
it'd
be
useful.
I
seem
it
that
hard
there's
got
to
be
some
way.
I
mean
that
it'd
be
stupid
for
them
not
to
given
some
kind
of
functionality
for
but
yes
I.
D
So
in
this
case,
I,
don't
know
why
it
does
this.
It
seems
to
pad
it
out,
but
not
fully,
but
that
that's
just
what
you
get
from
a
nap
sampling
process.
So
we
turned
it
from
the
smooth
low
resolution
curve
to
a
slightly
more
jagged
higher
resolution
curve.
Again,
that's
just
using
the
Panthers
function,
so
there's
I
can't
take
any
credit
for
any
complex
coding.
I'm,
throwing
this
stuff
over.
D
We
thought
it'd
be
worthwhile
to
be
able
to
convert
from
a
time
series
into
various
different
data
forms
so,
for
instance,
have
you
can
also
get
the
data
frame
back
now?
If
you
do
dot
data
at
the
moment,
it
does
obviously
return
that
original
data
frame,
but
it's
kind
of
important
in
in
our
opinion
and
for
the
general
projects
that
we
try
to
be
agnostic
to
what
our
internal
storage
is.
D
So
essentially,
I
don't
really
generally
think
people
should
be
using
the
pandas
dataframe
itself
when
they
can
avoid
it,
because
it
makes
more
sense
them
to
use
our
functions.
Then,
if,
in
the
future
we
find
that
there's
a
if,
for
instance,
a
strip
I
creates
time
series
style,
storage
mechanism,
then
or
suitable
table
which
has
time
series
support,
then
we
can
hold
over
to
the
new
data
storage
without
users
having
their
code,
something
coming
very
often,
so.
D
We've
also
got
the
ability
to
output
to
an
asteroid
pie
table,
and
one
of
these
sort
of
useful
things,
I
think,
is
that
we
actually
have
units
porting
over.
So
in
this
case
it's
counts.
Admittedly,
one
of
the
things
is
I've.
I
believe
I've
got
the
units
correct,
so
the
units
are
currently
hard-coded
within
the
instrument
classes
themselves.
D
I
believe
I've
got
them
correct,
but
anybody
who
actually
uses
the
data
on
an
individual
instrument
would
probably
have
a
much
greater
understanding
as
to
because
I
only
tried
to
read
the
documentation
and
they
haven't
always
made
them
totally
clear
to
me,
but
yes,
so
that
that
supports
units
which
are
all
quite
fond
an
equally.
Obviously,
you
can
easily
output
to
an
array.
D
So
if
all
you
want
is
the
actual
number
data
we,
as
I
said,
you
can
create
a
time
series
from
scratch.
So
if
I,
the
generally,
obviously
you
can
from
Panthers.
Admittedly,
for
some
reason
from
Panthers
hasn't
worked
from
a
daydream,
doesn't
look
right
now.
It
was
working
yesterday,
so
I'm
going
to
see
what
the
issue
is,
but
you
can
create
it
from
an
array,
so
very
dr.
paper.
D
So
that's
just
created
it
from
an
array
and
it
should
be
able
to
peek
what
is
it
won't
be
any
impressive,
but
that
is
the
array
that
I
gave
it
of
two
columns.
Likewise,
you
can
create
it
from
a
table.
I.
Just
reply
table
and
once
again
we
can
be
with
one
lessee
so
again,
a
very
basic
example,
but
just
to
give
you
the
idea
that
you
can
actually
use
a
strip,
I
tables,
you
can
also
explicitly
if
you
obviously
know
what
units
you're
going
to
be
using
so
so
with
default.
D
Units
and
it
will
its
case
its
specialist
units
because
no
way,
that's
it
Jessie
from
table.
Okay.
Well,
that's
not!
Why
kick
me
right
now?
It
was
working
the
other
day,
so
I
live
code.
What
can
you
do?
Yes,
the
whole
point
is
it's
meant
to
brought
in
the
units
that
I've
thrown
it
so
I've,
given
it
just
an
ordered
dictionary
for
the
units
and
it's
meant
to
be
able
to
put
those
across
so
so
right.
Okay,
you
can
obviously
change
the
units
by
changing
the
value
within
the
dictionary
itself.
That's
reasonably
self-explanatory.
D
All
it
is
is
a
dictionary
one
of
the
important
things
which
is
going
to
be
harder
to
demonstrate.
Oh
no
because
I'm
demonstrating
from
the
eve
so
s
right,
Oh,
overwrite
the
column.
Ok,
you
can
overwrite
how
long
have
I,
actually
I
haven't
demonstrated
how
you
get
the
quantities
out
which
is
interesting.
They
should
have
done
so.
Let
me
get
a
time:
Sue's
got
Eve
dot
city
and
then,
let's
just
grab
some
column
from.
Even
if
I
can
remember
come
from
even
about
that,
so
that's
it.
So
you
can
extract
a
quantity
reasonably
easily.
D
I
D
That
would
just
repair
return,
the
elevators
to
return
but
pandas,
and
so
a
time
frame,
but
potentially
I
want
to
make
it
so
that
you
can
return
without
quantity
and
just
to
return
the
array,
but
primarily
we
can
return
the
quantity
itself
using
that
quantity
function,
I've
got
to
me,
I've
been
trying
to
put
it
so
you
can
update
an
entire
row
I'm,
including
using
quantities
for
values
within
the
road.
But
at
the
moment
that's
not
really
working
out
too
well.
D
Something
I
need
to
talk
to
Danny
Institute
about,
but
it
becomes
very
unreliable
because
obviously
each
column
will
have
its
own,
potentially
its
own
units
and
then
you've
got
to
convert.
So
it
means
you
gotta,
throw
in
a
list
of
quantities
and
you've
got
to
convert
each
of
those
men.
Also,
there
is
a
very
easy
way
to
access
an
individual
row.
I
have
found
I,
don't
seem
to
be
able
to
find
a
way
to
accident
individually
within
Anders,
because
their
indexes
aren't
always
using
the
same
forms.
Obviously
it's
fine.
D
D
Ok,
that's
odd
equals
you
to
quantity,
oh
cute
and
Valley
need
I.
Haven't
I've
changed
it
now
too
right
there.
We
are
so
if
we're
using
a
quantity
that
is
a
quantity
value
which
I've
collected
from
the
row.
I
can
then
modify
that
quantity
value,
and
so
now
it's
and
to
be
different
numbers
special,
so
and
then
finally,
I
can
obviously
throw
that
back
into
the
original
table.
I've
got
this
over.
I
equals
true.
D
It
defaults
to
true,
but
it's
worth
making
sure
I
highlight
that
and
essentially
then,
we've
just
updated
the
values
within
that
time
series
object
so
later.
I
don't
expect
you
to
necessarily
remember
what
the
original
values
were,
but
we
have
updated
if
I
could
figure
out
to
calm
the
views
we
uses
ACM
lat
so
that
column
has
been
updated
and
I
say
he
come.
It
would
be
easier
for
me
to
show
you
if
I
had
a
bigger
screen
resolution
on
here
at
won't.
D
By
doing
and
exactly
the
same
can
be
done,
you
can
throw
in
just
an
asteroid
pie
a
numpy
array,
so
yeah
you
can
put
in
quantities
you
can
put
in
a
raise
and
it
will
figure
the
rest
out
the
update
row
stuff
as
I
say.
That's,
that's
that's
causing
trouble,
but
essentially
that's
where
we're
up
to
the
main
reason
I
want
to
demonstrate.
It
was
to
give
you
guys
an
idea
of
the
sort
of
functionality
it's
already
in
there
again
get
people
give
people
the
opportunity
to
start
playing
around
with
it.
C
Cool
thanks
and
for
the
oh,
you
should
make
a
pull
request
with
that
branch.
Stay
sighs
doing,
and
you
know
that
you
are
going
to
be
updating,
so
we
can
see
there
where
it
is
and
the
other
thing
I
was
going
to
ask
in
the
video
which
you
are
the
operations
on
on
the
time
series
like
the
pandas
right
yeah
it
do
the
units
get
updated
with
depending
of
the
operation
you
do
so
so
so,
for
example,
when
you
have
a
flux
which
is
a
per
second
and
you
the
last
some.
C
D
D
D
C
That
should
be
something
like
the
function
that
person
will
introduce.
It
should
have
also
a
function
for
the
units
or
some
Jana,
no
III
don't
have
a
clue.
Hopefully,
I
just
come
to
my
mind
and
that's
something
that
people
could
be.
You
know
believing
blindly
on
the
units
and
then
get
something
like
that
which
could
be
wrong.
Yeah.
D
I'll
just
write
that
down
as
a
note,
but
yes
I
mean
that's,
that's
going
to
be
a
very
difficult
sort
of
nitty-gritty
detail
when
it
comes
to
use
because,
yes,
as
soon
as
you
start
resampling
or
anything
that
then
you're
going
to
have
you're
going
to
have
issues
with
units,
you'll
see
yeah,
obviously
with
your
what
you're
doing
with
the
data
and
that's
where
I
don't
know.
If
there's
an
overall
structure,
you
could
say
well
in
every
case,
when
you're
doing
this,
you
can
do
it
that
way.
Yeah.
C
D
That's
that's
where
it
doesn't
it
doesn't.
It
depends
on
what
your
congratulating,
if
you
could
get
to
mating
the
same
data,
for
so
say
the
classic
example
I'm.
Giving
here
is
if
I'm
Canton
ating
goes
data.
Okay,
so
over
here
I'll
go
at
my
screen
is
a
chair.
I
will
show
it
again
if
you
could
can
dating
codes.
Data
like
I,
do
on
this
line.
Here
then
not
open.
D
They
might
obviously
have
different
a
different
cadence,
but
they'll
still
be
covering
the
same
time
range,
and
this
is
why
it
gets
complicated,
because
there
are
two
different
kind
of
uses
case
for
concatenation
Eva,
homogeneous
data
in
homogeneous
data.
Obviously,
homogeneous
data
is
reasonably
easy
because,
yes,
you
then
got
up
as
a
worker.
I
can
keep
all
you
know
all
the
metadata,
but
give
it
to
time
dependence.
D
D
Generally,
if
you
wanted
to
compare
results
for
multiple
instruments,
I
mean
again
I
can't
say
for
sure,
in
the
sense
that
I
never
used
a
light
curve
or
time
series
in
this
case
for
actual
science.
But
it
seems
reasonably
normal
to
me
that
there
are
going
to
be
times
when
you
would
want
to
have
more
data
from
multiple
sources
together.
D
E
D
C
D
D
C
That's
why
I
was
asking
me
for
whether
you
could
plot
different,
like
different
objects
together,
because
you
may
have
say
a
x-ray
light
curve
and
I
particles
event,
and
you
may
want
to
blow
them
together,
but
they
are
different.
They
have
different
units
and
they
have
your
friend
and
time
killings
and
they
are
different
instruments
as
well.
So.
D
E
D
To
be
fair,
when
you
say
when
I
say
in
homogeneous,
I
mean
in
homogeneous
in
sauce
Eve,
for
example,
as
in
homogeneous
data
from
itself.
So
I
read
as
an
example
where
I
can't
say
off
the
top
of
my
head,
but
so
we've
got
quite
a
lot
of
different
values
with
Eve.
We've
got
a
number
which
are
fluxes
relating
to
things,
but
we've
also
things
which
accounts
particle
counts.
I
believe
so
so
so.
E
E
D
Not
necessarily
a
manipulation,
but,
for
example,
if
you
wanted
to
output
the
data
to
a
table,
then
it's
convenient
to
have
them
all
in
one
object
and
then
they're
all
in
one
object
or
output
as
a
table
without
having
to
map
around.
Obviously
you
could
get
up
and
say
well,
I
could
create
the
table
from
the
two
separate
sources
manually,
but
it
then
adds
another
layer
of
complexity
to
the
overall
operation.
A
I'm
still
not
complete,
and
then
you
think
about
it
more,
but
I'm
still
a
little
bit
I,
don't
know,
I'm,
not
convinced
that
having
multiple
instruments
in
the
same
time
series
is
something
that
we
necessarily
want
to
be
encouraging.
I
know
it's
something
that
pandas
can
handle
quite
well.
I
just
don't
know
ever
there's
enough
of
a
use
case
for
us
to
handle
it
anytime
soon
eat,
but.
J
I
would
say
things
that
might
be
relevant,
including
Manas.
Are
you
having
multiple
instruments
intensities,
but
other
information
about
instruments
states
in
the
time
series
like
this
might
be
a
natural
way
to
store
that
weather
comes
to
four
second
recipe:
time
series
for
the
attendant
here:
decimation
stabilize
those
are
more
technical
based
examples,
but
I
can
certainly
imagine
situations
where
we
have
other
time
a
native
contingent
on
us,
while
measurement
that
we
want,
and
we
want
to
keep
coupled
in
the
same
object.
Relations.
A
Two
extra
columns
right,
I,
know,
technically
so
is
extra
information
is
extra
instrumentation
put
that
come
through
a
lot
more
baggage
than
just
on
this
time
step.
We've
recorded
this
piece
of
data,
this
theta
theta,
and
there
were
these
properties.
It's
just
like
a
charter
management
on
other
properties.
J
Might
be
reported
and
transported
under
a
difference
scheme,
so
it
may
not
necessarily
be
like
easily
be
like
partition
to
exactly
the
same
sample
across
the
board
or
something
for
that.
The
lack
of
I
guess
if
flexibility
is
a
good
thing.
If
we
can't
imagine
exactly
what
this
will
need
to
hold
right.
A
D
Think
the
problem
is,
there's
a
potentially
options
for
using
it,
but
I
do
agree
that,
as
a
general
rule,
I
can
imagine
that
not
having
it
wouldn't
be
the
end
of
the
world.
It
would
take
time
series
much
less
functional,
and
so
maybe
it
is
worth
cutting
it
out
for
now.
Just
make
it
simpler,
more
more
easy
to
manage,
but
I,
say
I,
don't
mind,
that's
one
of
the
reasons
for
bringing
it
up,
but
it
is
a
big.