►
From YouTube: SimPEG Meeting Aug 28
Description
Overview of JupyterCon 2018
B
The
conference
itself
was
because
his
self
was
like
there
two
days,
a
conference,
and
then
there
were
two
days
of
tutorials
the
workshops
before
so
I
went
to
one
day
of
workshops.
First,
when
I
did
was
on
deploying
urology
karate,
which
is
really
mean.
We
actually
worked
through
like
step
by
step,
how
to
set
up
on
Google
Cloud.
D
B
B
B
B
B
So
it's
there's,
there's
two
styles
of
tutorials
that
they've
done
I'm
just
gonna
show
my
screen:
hey,
Joe,
okay,
we're
just
catching
up
on
shooter,
gone
and
interesting
things
with
that.
So
there's
two
sets
of
tutorials
that
the
g4
teams
developed
and
like
how
to
deploy
your
of
Jupiter
hub.
There's
zero.
Did
you
clear
hub,
and
so
this
is
meant
to
be
deployed
on
a
cloud
and
they're
using
this
package
called
kubernetes.
B
It
basically
lets
you
sort
of
set
up
and
partition
resources.
So,
for
example,
if
I
set
up
a
Jupiter
hell
before
the
same
pain
team,
and
we
wanted
to
make
sure
that
everybody
has
at
least
I
can
go
Brown
and
we're
going
to
cap
it
at
eight
gigs
of
RAM,
so
that
our
Google
can
keep
this.
You
can
configure
all
of
those
kinds
of
things
who
is
the
keeper
Nettie's,
so
it
basically
started.
B
B
And
so
this
has
basically
everything
you
may
need.
If
you
want
to
set
up
a
Google,
Cloud
or
other
I
guess
they
got
them
for
most
as
the
cloud
providers,
all
the
big
club
providers,
and
then
you
can
also
set
this
up
when
you're
all
in
the
hardware,
but
that's
something
I'm
not
familiar
with
how
to
do.
You
have
to
have
kubernetes
and
all
of
that
installed
I
mean
that's
what
this
is.
B
The
other
one
that
they
have
info
on
is
called
the
littlest
your
home,
and
it's
just
for
smaller
deployments,
and
it's
basically
meant
to
be
like
via
a
single
server,
so
the
zero
to
Jupiter
hub
is
for,
if
you're
on
a
cloud
and
you've
got
access
to
resources.
This
is
for
a
single
server
like
smaller,
which
is
bear
cub.
B
So
if
you're
not
doing
I
mean
if
we
just
had
six
users
and
nobody's
doing
anything,
super
large,
the
master,
but
a
good
place
to
go
so
that
way,
if
all
fell
apart
done
with
the
UVC
stuff
for
for
350
like
this,
actually
is
not
a
hard
option.
It
should
definitely
be
get
backup,
like
ideally
you're
on
the
UVC
stuff,
and
somebody
else
is
doing
maintenance
for
you.
But
it's
good
yeah
good
to
know
that
that's
out
there
yeah.
B
B
E
B
E
B
A
B
B
Baby
there
was
I
think
at
some
point.
Most
of
the
talks
will
be
posted,
I,
don't
know
last
year
they
were
really
delayed
in
getting
them
up,
so
I'll
try
and
keep
an
eye
out
for
them,
and,
let
me
know
perfect:
there
are
a
few
other
just
like
interesting
things
that
I
thought
of
this
impact
community
in
general,
just
with
packages
coming
up
or
with
different
projects,
and
things
like
that,
just
trying
to
figure
out
where
to
start.
B
What
it
lets
you
bring
in
these
lip
maps
right
into
the
network,
which
is
pretty
slick
and
you
can
add
layers
and
basically
any
like
public
dataset.
They
thought
get
in
your
faces,
so
you
stall
I
know
you
were
doing
some
work
getting
stuff
reading
to
Google
and
I
actually
thinks
it's
effectively.
What.
E
B
B
Mean
like.
B
Yeah
images,
so
here
is
there
dots,
so
I
think.
B
So
that's
that's
one
that
was
pulled
from
NASA,
but
you
can
definitely
switch
to
the
Atlanta
you
for
sure.
And
then
you
can
add
your
markers
and
things
like
that
and
polygons
and
layers.
Okay
and
the
markers
was
actually
really
cool
bottom
of
the
polygons
and
layers
is
there
you
can
treat
them
as
widgets,
so
here
they
set
draggable
as
false.
D
G
B
B
B
C
So,
let's
see
back
to
your
other
things,
so
if
you're
doing
something,
that's
completely
mundane
but
does
come
up
well,
let's
suppose
that
we
wanted
to
book
a
place.
One
go
through
air
B&B
and
also
use
Google.
Maps
just
have
them
kind
of
located,
so
you
could,
you
know,
have
them
talk
to
each
other
and
you
could
maneuver
them
a
bit
and
would
this
help.
E
B
Necessarily
case
potentially
like
in
that
case,
just
using
Google
Maps
is
just
as
easy,
but
here
I
think
if
you
have
like
any
sort
of
datas
like
what
Dom
was
saying
if
he
wants
to
click
and
draw
a
polygon,
for
example,
on
them
up
and
then
get
these
points
programmatically
all
of
the
is
stored
as
a
widget,
so
he
can
access
them
in
the
same
notebook.
Okay,.
B
E
B
D
B
Yeah
children,
love
Jewry
life
is
now
like.
Actually
pretty
ready
for
for
use,
I
was
playing
around
with
it
while
there,
because
there
were
a
lot
of
things
before
that.
Well,
they
just
hadn't
mature
again,
like
figure
it
out
little
once
now,
it's
actually
pretty
stable
for
use,
and
so
actually,
if
you're
on
any
notebook.
So
here
let
me
see:
that's
we'll
come
back
to
this,
but
if
you
just
launch
a
notebook
normally
it
goes
to
tree.
So
your
URL.
B
And
I
need
to
build
the
leaflet
poking
around
that,
but
then
you
can
start
playing
around
with
jupiter
lab
one
of
the
things
that
a
lot
of
people,
something
just
extremely
beneficial
right
away,
is
the
file
system
of
spider.
So
then
I
can
drag-and-drop
move
things
around
so
rather
than
having
to
rename
that
a
book
to
include
this
or
whatever.
So
that's
those
are
simple
things
you
can
see.
B
So
that's
that's
helpful
were
just
like
you
know:
octane
you're,
simply
markdown
files,
there's
also
some
really
cool
things
where
you
can
run.
You
can
attach
a
kernel
to
your
markdown
and
run
your
code,
samples
to
make
sure
that
if
you
like,
created
an
example,
what
it
runs
I,
don't
remember
maybe
that
offhand
but
there's
there's
dogs
on
them.
D
B
Is
like
basically
a
lot
of
an
extension
to
Google
Earth,
but
here
they
actually
have
you
can
access
all
the
NASA
data
sets.
They
have
been
recommended
like
getting
that
API
right
with
them
so
that
you
can
access
any
type
of
data.
The
shows
are
using
it
in
Jupiter
and
right
now
to
get
access
to
the
API
and
you
do
need
to
request
access
filling
in
the
form,
but
I'm
pretty
sure
anybody
here
who
says
they're
doing
it
graduate
degree.
D
B
B
C
B
So
he's
got
things
like
quicker
applause,
but
one
of
the
biggest
and
I.
Don't
think
this
example
would
be
here
yet,
but
he's
now
gone
following
a
rendering
of
multiple
objects
in
there,
which
is
fantastic
so
that
if
you
want
to
have
multiple
datasets
or
like
services,
plus
your
inversion
results,
and
all
of
that
that
is
actually
possible.
B
B
C
B
So
that's
pretty
slick
and
I
think
he's
very
responsive
to
you.
So
if
we
started
looking
around
and
making
beautiful,
and
especially
if
those
things
that
we
can
jump
in
and
help
out
like
slicing,
that's
actually
pretty
easy
and
he'd
be
quite
receptive
to
that
and
he
should
be
there's
associates
here.
He's
got
some
good
tutorials
that
actually
sort
of
give
you
an
idea
of
how,
following
rendering
is
done
because
there's
things
like.
B
Sorry
so
I
say
surfaces
are
actually
very
expensive
to
compute,
but
there's
ways
to
sort
of
hack
around
that
just
by
setting
reasonable
ranges
and
opacity
on
things
that
gets.
You
look
close
enough
to
your
tree
isosurface.
So
those
are
things
that
you
can
play
around
with,
but
ISO
surfacing
is
actually
yeah.
E
A
B
B
Because
there
was
a
demo
that
he
did
of
like
structure
of
a
brain,
and
so
he
had
each
color
differently
and
you
could
see.
Maybe
that
was
actually
like
a
volume
and
he's
done
some
clever
things
too,
so
that
you
can
effectively
free
sample
as
its
Scrolls
as
you
zoom
it.
So,
for
example,
if
we
have
a
very
large
event,
you
don't
actually
want
to
render
the
entire
thing,
so
you
can
define
based
on
which
scale
you're
on
how
to
draw
the
image.
So,
like
he's,
he
knows
who
these
days
you.
B
Actually,
one
of
the
things
that
we
could
maybe
do
is
start
like
a
scratch
pad
repo
in
simpe
and
just
put
notebooks
of
like
things
that
you've
tried.
That
are
interesting.
So
if
somebody
starts
poking
around
with
voluminous
like
here's,
a
decent
example
that
uses
simple,
they
were
just
given
a
place
to
put
those
kinds
of
things
that
be
useful.
B
E
C
E
B
C
E
E
B
Ngo
Delphi
data
org.
If
you
go
there
right
now,
you'll
be
asked
to
sign
up,
but
because
I've
already
signed
up,
I
can
log
in
so
they
deployed
each
of
your
hub
for
Geoscience,
specific
use
cases.
So
like
what
comes
predefined
in
the
images,
we've
got
a
lot
of
speed,
standard,
Geoscience
packages
located
and
so
they've
got
a
few
examples,
and
this
one
is
actually
super
cool
because
they're
eating
gas
and
x-ray
so
desk
does
parallelization
and
x-ray
is
like
smart
storage
of
structured
data
and
so.
B
B
B
B
B
B
B
B
What
I
mentioned
in
kubernetes
before
it
is
basically
that
that
virtual
interface
creating
a
cluster
on
any
hardware,
so
here
we've
created
a
kubernetes
cluster
with
20
workers
and
you
can
go
in
and
change
the
adaptive
and
manual
scaling.
What's
one
of
the
really
cool
things
here,
it's
been
with
tasks
which
doing
all
the
paralyzation
and
such
you
get
a
dashboards
that
they've
created
that
actually
just
tell
you
what's
going
on
in
your
notebook,
so
here
I'm,
just
gonna
rearrange
some
things,
so
we
can
see
it
all.
B
So
the
next
thing
we're
going
to
do
is
take
time
series
means
over
this
data
set.
So
this
is
going
to
be
a
computationally
intensive
step
just
because
we're
going
through
a
very
large
data
style,
and
so
here
as
soon
as
you
start
done,
we
then
actually
start
getting
an
idea
of
what's
going
on
so
zetta
array
is
actually
we're
accessing
we're
paying
our
database.
It's
the
Czar
mean
chunk.
B
We
are
chunking
the
data
and
then
here
we're
aggregating
their
teeth
from
the
average,
and
so
you
can
see
things
happening
in
parallel
as
its
operating
so
and
so
you
can
have
a
look
at
what
the
workers
are
doing.
I,
don't
turn
it
like
I
haven't.
Actually,
if
you
ask
me
what
these
things
actually
are
telling
you
I,
don't
try
any
other
tasks,
we
saw
that
and
then
you
can
look
at
the
system.
So
we
can
see
what
our
CPU
memory
use
and,
with
all
of
those
kinds
of
things.
B
E
B
B
Yeah
so
I
think
that
just
reading
a
few
things
that
they
are
up
to
will
be
quite
valuable
for
us.
I
posted
a
link
on
slack
earlier
just
on
data
types
in
the
cloud
system
of
the
things
that
Ryan
Abernathy
who's,
one
of
the
core
developers
of
Penn
Gio
I
spent
some
time
talking
about
and
spent
time
talking
about.
Andy's
keynote
was
data
tapes
excel
that
back
to
the
students,
easy
plot.
B
C
B
Being
turn
around
and
just
not
familiar
at,
what's
going
on
so
there's
a
couple
of
file
formats
that
I
think
are
worth
looking
into.
This
isn't
like
a
huge
concern
for
us
yet
because
our
data
sets
are
still
small
compared
to
what
like
NASA
deals
with,
but
I
think
it's
good
to
stay.
Aware
of
so
because.
B
Of
the
ones
that
I
think
is
quite
relevant
to
us
potentially
in
the
future
is
this
is
our
and
that
is
sort
of
things
that
kind
of
market
themselves
as
the
cloud
equivalent
of
HCI
5
or
a
cloud
alternative.
Maybe-
and
so
that's
talked
a
bit
about
in
here
and
then,
if
you,
google
it,
you
can
get
a
bit
of
an
idea
of
just
like
how
it
operates
and
it's
interoperable
of
nominal
ancestor.
E
A
C
C
B
E
B
G
F
B
B
G
B
G
G
E
E
B
Is
where
I
think
to
we
need
to
make
better
use
of
slack
and
there's
potentially
that
we
could
create
a
new
channel
or
something
like
that
if
you
want
but
I
think
that
like
because
he
was
your
point,
the
fact
that
it
was
raining,
but
it
sort
of
fell
off
the
radar
like
you
need
an
avenue
to
push
and
say
this
thing
he's
done.
Can
we
get
it
in
because
it's
often
like
right
now,
there's
900
requests
on
simple,
yellow.
B
G
G
B
But
I
think
they
don't
hesitate
to
post
on
slack.
Hey.
Can
somebody
help
review
this
because
it
doesn't
have
to
be
like
it's
nothing
myself
or
wrong
yeah.
If
somebody
else
like
one
of
the
things
that's
becoming
cool
on
I'm,
on
the
interact
slack
and
basically
anytime,
we
knew
pull
request
is
generated.
They
just
throw
a
message
case.
Somebody
lender
review.
This
I
not
then
hopefully
starts
moving.
E
B
G
B
F
B
And
then
here,
if
you
don't
mind,
try
and
be
just
a
little
more
descriptive
here,
just
cuz
this,
like
whatever
is
in
your
main
comment
thread,
is
what
we
pulled
for
the
release
notes.
Oh
yeah
could
be
that
effort
me
that's
how
skating
and
then
I
think
there
were
a
couple
on
district
ties
and
I
meant
to
get
in
touch
with
you
Mike.
This
is
the
one
that
you
did
and
I
was
just
wondering
if.
B
B
Here's
a
couple
of
here.all
posts
on
slack
and
maybe
payments
if
somebody's
willing
to
review
both
of
both
of
these
with
this
one,
what
was
I
doing
this
I
actually
think
that
we
should
rethink,
but
for
now
is
perhaps
okay
just
to
allow
the
line
coloring
thickness
to
be
changed
on
the
clock.
I
think
the
viewers
actually
need
to
be
re-written
as
more
object
based
things,
but
that's.