►
Description
June 12, 2019 Jupyter Community Workshop talk by Russell Neches, Joint Genome Institute
A
A
A
A
Of
course
it's
got
to
go
so
you
know
it's
just
cutting
your
some
hobbits
or
sheep
or
whatever,
or
so
you
know
you
should
be
able
to
just
summon
an
undead
horde
and
let
loose
it's
an
embarrassingly
parallel
awful
ego
but
as
you're
spawning
all
the
zombies
that
are
gonna
go,
destroy,
New
Zealand
and
spare
you
of
people's
beautiful
vacation
photographs
forever.
You
notice,
you're
summoning
your
you're
summoning
circle
is
running
really
slow,
so
you
check
the
performance
statistics
song
on
the
head,
node
for
your
your
summoning
grid
and
well
I
like
well.
A
A
A
Thinking
you
just
have
to
communicate
with
it
with
you
are,
and
how
do
you
get
your
your
your
job
dispatch,
one
summoning
circle
to
the
2388
summoning
circles,
with
thirty
two
altars
each
that
you
have
available
in
this
great?
So
if
you
don't
tell
well,
Igor
does
things
in
a
very
specific
in
particular
way,
and
if
you
don't
talk
to
him
the
way
he
wants
to
be
spoken
to
well
he'll,
just
get
you
the
wrong
brain
or
just
no
bad
things
up.
This
was
white.
Dark,
dark,
wizards
are
always
so
gloomy
anyway.
A
A
Sorry,
so
user
experience
is
an
important
part
of
any
interactive
loop
right.
So
if
you
want
people
to
be
able
to
like
interact
with
your
program
or
their
program,
if
they're
running
it
on
your
infrastructure
eye,
friction
when
it
comes
to
doing
basic
things
like
taking
something
and
making
it
parallel,
she's
not
actually
basic,
it's
very
hard
and
significantly
damaged
the
usefulness
of
the
resource
for
the
intended
application
serving
research
also
a
part
well
code.
A
A
A
Little
Jupiter
extension
called
summon
Undead.
This
is
an
intentionally
silly
design.
Part
of
the
reason
why
I'm
here
is
I
would
like
to
solicit
some
feedback,
so
I
can
make
some
actually
useful
way.
This
works
is
you
make
a
little
array
of
dictionaries?
They
each
contain
variables
that
you
would
like
to
you'd
like
to
inject
into
a
code
cell
and.
A
Usually
one
passing
the
vector
of
run
arguments
you
asked
for
without,
but
the
names
of
the
outputs
one
go
back
one
one,
and
basically
this
is
specific
language
of
the
rule.
Seven.
So
at
some
point
that
we
do
you
want
to
collect
later
and
then
you
call
it
what
mode
cereal
cereal,
so
adhering
between
state
can
be
recreated
a
local
performance
for
level
here.
So.
A
A
Document
my
research
process,
mostly
I,
also
like
to
use
it
in
Haiti,
my
research
when
I
end,
but
the
most
part,
oh
well,
my
my
advisor
is
it
just
as
a
PhD
student
and
my
boss
was
a
postdoc,
alright,
show
them
Jupiter
notebooks
they
tend
to
glaze
over
like.
Why
do
you
have
26
notebooks
for
this
one
project?
Why
are
they
5050
swallowing.
A
Emily
research
process
is
total
chaos,
but
I'm
not
asking
for
a
solution
to
that.
I'm
actually
very
happy
with
this
I
love
the
chaos
I
want.
It
is
something
that
will
reduce
the
cognitive
load
of
using
the
resources
that
are
available
to
me,
so
the
heavy
lifting
the
particularly
computational
heavy
heavy
lifting
should
be
done
on
the
compute
nodes
HPC.
A
But
if
you
are
lazy
like
me,
you
will
probably
just
run
them
wherever
you
have
CPUs
available
and
if
there
are
enough
on
the
head
node
or
on
the
Jupiter
hub,
you
know
then
you'll
just
run
it
there,
probably
and
hope
nobody
tells
you
so
the
things
the
rapid
like
this
can
possibly
offer
is
a
nice
way
for
administrators
to
encourage
users
to
choose
extensible
set
of
settings
for
how
they
how
they
interact
where
and
how
they
run
their
jobs
on
on
the
HPC.
Without
having
to
be
you
know,
really
stirring
with
the
policy.