►
From YouTube: SimPEG meeting Nov 11, 2020
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
We're
now
live
excellent.
Who
wants
to
start
us
off?
I
see
tebow's
name
at
the
top
tebow.
Would
you
like
to
start
us
off.
B
B
So
I
think
what
I'm
gonna
prioritize,
especially
today
and
maybe
a
bit
of
tomorrow,
is
actually
the
smaller
pr
and
so
that
we
can
merge
them
in
and
we
can
then
imagine
him
and
that
will
be
and
and
that
that
will
be
done
for
the
bigger
pgi,
pr
and
yeah
so
yeah
and
on
the
pgi
si
it's
it's
more.
B
Like
I
say
it's
like
functionalities
and
tests
are
there
it's
and
I've
started
writing
the
documentation,
but
it's
gonna
take
a
little
bit
and
maybe,
in
the
later
on,
I
will
ask
you
also
maybe
what's
the
best
way
to
acknowledge
what
we
modify
from
the
scikit-learn
package.
Okay,
there
yeah
like
a
number
of
functions
from
them
that
I
take
and
rewrite
so.
B
Do
we
put
that
in
every
description?
I've
started
doing
that,
but
I've
also
like
put
like
a
headers
inside
the
utils
files
to
acknowledge
it
as
a
whole.
So
there
might
be
some
things
to
to
to
to
think
about
that.
That's
that's
it
for
me.
A
Excellent
thanks
tebo,
so
next
to
tebow
I
see
xi'an.
Would
you
be
willing
to
go
next.
C
Okay,
yes
sure
yeah,
so
in
our
last
meeting
dr
sun
mentioned
about
sharing
our
synthetic
model
scenarios
and
also
the
data.
So
I'm
working
on
that.
Currently
I'm
working
on
locally
on
my
jupiter
notebook
to
have
all
the
plots
ready
and
once
it's
finished
I
will
upload
it
to
the
github.
D
Yeah,
I
don't
have
much
to
report,
had
a
meeting
with
shaolin
last
week
talking
about
the
cross
gradient
invert
code,
so
we
went
over
the
code
one
more
time
and
we
what
we're
going
to
do
next
is
shalom.
Have
you
have
you
already
done
the
pr
or
not
yet
pr?
I
already
submitted
apr.
Okay,
okay,
one
thing
that
we
talked
about
last
week
is
to.
We
realized
that
we
need
to
do
a
few
more
tests
which
we
already
did
last
year,
but
it
was
not
organized.
D
It
was
not
documented
to
make
sure
that
what
we
our
code
is
doing,
what
it's
supposed
to
do,
it's
calculating
the
cross-section
everywhere
properly
and
it's
it
can
calculate
the
grid
enhancing
property.
We
did
all
the
tests
last
year
when
we
first
started,
but
we
didn't.
D
Summit,
we
didn't
summarize
that
in
writing,
so
we
want
to
do
that
start
over
to
do
redesign
all
the
tests
and
do
that.
That
will
also
be
as
far
as
I
understand
so
that
will
also
help
you
guys
to
to
make
sure
that
the
code
is
working
properly
before
it's
get
merged
right.
D
So
that's
what
we're
going
to
do
next,
just
basically
designed
a
few
small
synthetic
tests
yeah
to
to
make
sure
that,
because
it's
a
little
it's
a
it's
a
simple
idea,
but
when
it
comes
to
computation
it's
very
complicated,
so
yeah!
That's
what
we're
going
to
do
next.
E
Yes,
as
doxon
just
mentioned
recently,
I'm
working
on
the
test
part
for
the
pr
I
submitted
last
week
and
another
thing
that
I
keep
keep
my
eyes
on
the
t-balls
pr
as
well.
Every
time
when
he
make
a
comment,
I
will
take
a
close
look
at
his
code
and
the
comments
and
trying
to
learn
how
to
play
this
game
well.
Well.
A
F
Just
really
quickly
I'll
go
over
things.
A
few
things
I'd
like
to
talk
about
later
after
everyone
gets
through
things,
but
I've
been
basically
finalizing
the
reviews
of
a
bunch
of
these
pr's
and
getting
them
all
in
the
master.
Still
I
was
helping
took
time
to
this
weekend.
I
was
thinking
about
the
mt
simulation
stuff
that
I
was
helping
lindsay
and
john
with
on
the
getting
the
fields
working
and
getting
it
working
simultaneously
with
the
frequency
domain,
em
bass
class.
I
think
we're
getting
we're.
I
think
I
hoped
I'll.
F
Let
them
talk
a
little
bit
more
about
it.
Maybe
we
can
discuss
it
later.
Yeah
finalizing
this
pr
for
the
dcip
thing
that.
G
F
There's
things
that
I
would
I
had
commented
on
that
I
wasn't
quite
sure
of
just
yet
so
I'd
like
to
finalize
those
later
after
everyone
gets
through
things,
as
well
as
what
doug
had
brought
up
on
his
pull
request
comments
about
the
name
of
this
fictitious
source
thing.
So
if
we
can
go
over
that
after
everyone's
gone
through
their
things,
that
would
be
good.
A
Thanks
joe
doug,
would
you
like
to
go
next.
H
Yeah
I
could
and
it'll
probably
lead
into
what
devin
has
has
been
doing
so
devin's
been
doing
some.
You
know
comparisons
of
simpag
and
ubc
coats,
and
so
I
think
you
know
he'll
he'll
talk
about
them
and
they're,
actually
looking
really
good,
but
I
was
I
was
going
through
his
poll
request
and
just
on
the
basis
of
that,
I
was
trying
to
kind
of
really
get
grab
hold
of
the
details
of
the
mathematics
that
was
going
in
and
the
details
of
the
numerics
that
are
being
coded.
H
Just
to
kind
of
you
know
throw
off-
and
you
know
say
this
but
yeah
in
the
end
of
the
day,
kind
of
understanding
those
details
are
are
important
and
then
just
now,
when
we're
talking
about
the
the
code
with
xiao
long
and
from
houston
and
and
judge
group-
and
he
was
talking
about
the
cross
gradients-
and
you
know
he
said
well,
you
know
in
principle
this
is
kind
of
straightforward,
but
in
practice
you
kind
of
need
to
take
care
of
a
lot
of
details,
and
I
think
the
same
is
true
with
emt.
H
You
know
the
discretizations
and
where
we're
putting
stuff
and
things
like
that,
and
we
do
have
quite
a
bit-
I
mean
you
had
like
pixels
in
their
neighbors
and
you've
got
a
couple
of
things
with
respect
to
you
know
em
and
you
know
hj
and
eb
fields,
and
you
know
so.
We
got
a
lot
of
stuff
there.
But
I
found
myself
yesterday
after
kind
of
interacting
with
devin
and
looking
through,
you
know
the
code
that
he
had
it's
just
like.
H
H
But
anyway,
I'm
just
wondering
if
that's
worthwhile
discussing-
and
I
I
think
we're
able
to
do
something
in
that
regard.
It
would
really
be
a
benefit
and
then
people
would
you
know,
have
a
really
good
kind
of
understanding
of
mathematically.
What's
going
on
numerically,
what's
going
on
and
then
see
the
python
code
and
it's
like
wow,
I
got
it.
I
Yeah
there's
there
is
a
place
for
that
when
we
were
first
making
the
simpeg
website
the
theory
and
the
api,
and
everything
was
all
kind
of
smashed
together
and
right.
Now,
it's
all
living
in
a
place
on
the
website
under
the
heading
packages
and
yeah.
When
I
was
outlining
sort
of
what
I
wanted
to
update
for
this
simpeg
website,
it
seemed
like
a
really
good
place
to
yeah
to
put
a
place
where
we
actually
go
and
do
that.
I
don't
think
we
need
to
necessarily
derive
a
lot
of
things
from
scratch.
I
In
some
ways,
that's
kind
of
what
em
geosci
is
for,
but
we
could
start
out
with
any
one
of
our
methods
and
say:
okay,
here
is
the
formulation
of
maxwell's
equations
for
this
particular
problem.
If
we
want
to
do
you
know
an
eb
formulation
and
then
we
end
up
with
okay.
This
is
how
the
discretization
works,
and
this
is
where
everything
is
living.
I
think
that's
extremely
useful.
H
That
that
actually
sounds
that
sounds
good.
How
advanced
is,
I
must
admit,
I
haven't
really
looked
at
that,
let's
say
for
the
dc,
so
if
we
what's
there
for
the
dc
and
then
for
the
latest
pull
request
that
you've
been
working
on
in
the
comparison
of
the
ubc
codes
and
simpek
codes.
How
much
of
that
background
information
to
understand
the
codes
would
actually
be
there
at
this
point.
I
Well,
I
have
I
mean
I've.
I
comment.
I've
been
commenting
stuff
and
trying
to
add.
You
know
doc
strings
as
I've
been
developing
it,
but
I
haven't
put
really
that
level
of
detail
into
the
formulation.
So
I
feel
if
you
were
reading
through
the
code,
the
comments
would
be
able
to
tell
you
what
you're
doing,
but
in
terms
of
this
nice
packaged
sort
of
thing,
that's
kind
of
more
focusing
on
okay.
I
How
is
the
math
working
I
haven't
developed
that
specifically,
but
it
does
basically
exist
on
the
website
right
now,
so
we
just
we
just
have
this
underdeveloped
catch-all
area
where
we're
talking
about
the
packages
that
simpeg
has,
and
I
mean
I
I
could
go
and
try
and
clean
clean
that
up
and
and
turn
it
into
that
theory
section.
If
we
wanted
to.
A
So
if
you
don't
mind,
I'm
gonna
hit
pause
on
this
conversation
and
we'll
keep
going
around
and
then
let's
circle
back,
because
I
think
that
there's
there's
a
few
things
to
unpack
here.
So
if
folks
do
have
ideas
or
if
you
have
a
thought
right
now
that
you
want
to
make
sure
you
keep
in
mind,
just
add
it
to
the
notes
under
under
doug's
name,
which
is
on
the
second
page
in
the
quick
reports
and
then
we'll
circle
back
and
make
sure
that
your
your
notes
and
thoughts
are
are
captured.
A
Okay,
doug
was
there
anything
else
you
wanted
to
to
sort
of
add
to
that
point.
While
it's
top
of
mind
for
everyone?
Oh,
I
I
think
that's
fine!
Okay,
do
you
have
any
other
other
okay
sounds
good
devin.
I
Sure
so
yeah
the
pull
request
for
potential
fields
is
merged.
So
we
have
a
much
more
stable
formulation
for
the
sensitivities,
so
that's
great
and
then
it
also
led
to
me
creating
an
I
o
utilities
section
under
utilities.
So
there's
we
want.
It
would
be
nice
to
be
able
to
just
easily
import
and
export
all
ubc
formatted
data
files
for
all
of
our
codes
and
I'm
basically
building
that
that
library
right
now
so
we
have
all
the
ones
for
our
potential
fields
codes.
I
So
it's
going
to
be
really
easy
to
just
import
and
export
ubc
formatted
data
and
then
yeah
I
mean
I
did
these
code
comparisons,
the
ubc
versus
the
simpeg
a
whole
bunch
of
examples,
and
I
think
we
get
really
quite
good
agreement.
So
I'm
pretty
happy
with
what
simpeg
is
doing
and
I
I
think
it's
quite
accurate
and
it's
it's
fast
and
I'm
really
happy
with
that.
So
I
have
those
just
yeah.
I
Yeah,
I
would
say
so
I
mean
I
just
needed
to
go
and
really
look
where,
where
things
were
being
where,
where
things
were
being
defined
on
the
mesh
and
then
the
big
thing
was
that
method
of
fictitious
sources.
I
Okay,
so
we
don't
have
to
get
into
crazy
details
of
it.
But
I
added
a
new
way
of
of
computing.
The
sources,
and
I
have
ways
of
we've
compared
all
of
the
things
that
simpeg
and
that
ubc
code
is
doing,
and
we
have
those
validations
in
notebooks
and
there's
markdown,
and
it
explains
everything
so
after
this
vip
pull
request
goes
through.
One
of
the
things
we
might
think
about
is
is
where
those
validations
are
going
to
live
so
that
everyone
else
has
access
to
them.
J
Nice,
nice-
and
it
is,
it,
is
like
sort
of
complete
like
you
can
use
both
for
like
a
cc
problem
or
cc
simulation
and
nodal
simulation.
I
Well,
I'm
doing
it
with
the
octree
mesh.
So
as
of
right
now
the
boundary
conditions
on
the
tree
mesh
wouldn't
allow
me
to
do
the
a
cell,
centered
formulation
without
you
know
creating
kind
of
air
cells.
I
So
if
you,
if
you
want
to
do
something
where
the
topography
is
flat-
and
you
have
all
your
your
electrodes
on
the
surface,
you
need
to
use
the
nodal
formulation
if
you're
doing
a
tree
mesh,
but
I
mean
we
have
a
formulation
that
works
very
well,
and
that
would
be
the
one
that
I
would
suggest
using
anyway.
I
So
okay
yeah,
so
it
sounds
like
right
now.
I've
been
working
on
the
doc
strings
for
the
discretize
website
and
maybe
I
could
think
about
doing
this
background
theory
type
stuff,
but
yeah.
I
have
my
hands
on
lots
of
things.
That's
about
it.
H
I
I
A
All
right,
thanks
devin,
I
had
one
question
for
you
just
on
the
I
o
utils,
if
you
started
an
issue
that
maybe
sketches
that
out
before
starting
a
pull
request
just
so
that
we
can
maybe
have
a
bit
of
a
conversation
around
the
structure
and
if
there's
things
that
we
want
to
sort
of
think
through
before
before
implementing.
I
Well,
it's
I
mean
it's,
it's
part
of
the
current
set
of
pull
requests,
they're,
just
they're
utilities
that
I
needed
to
develop
in
order
to
do
these
validations
to
be
able
to
do
the
work
that
I
was
doing.
Okay,
so
they're
kind
of
being
developed
as
as
I'm
working
on
other
stuff,
but
we've
I
mean:
we've
had
several
meetings
over
the
past
couple
of
years
about
doing
something
similar
and
it
never
moved
forward
and
it's
something
that
we
need.
So
I
guess
I'm
maybe
more
likely
to
try
and
just
give
it
to
the
people.
A
Okay,
joe,
do
you
have
comments
or
thoughts
on,
because
so
one
thing
in
general
I
would
say,
is
you
know
if
we
are
going
to
be
trying
to
sort
of
it's
helpful
to
sketch
out
an
issue
first
to
just
kind
of
communicate
that
you're
going
to
be
working
on
something
and
then
also
to
give
to
give
folks
a
chance
to
chime
in
joe
yeah?
I
don't
know
if
you
have
other.
F
Yeah
I
mean
like
I
I
don't
mind
them
like
having
a
bunch
of
iou
pills
like
people
can
just
go
like.
I,
like
I
kind
of
like
the
idea
of
having
that
as
a
module,
because
a
separate,
like
you
know,
folder
module,
because
then
we
could
go
in
and
add
makes
a
little
bit
a
place.
It
makes
it
as
a
place
that
can
be
like
oh
hey,
if
you
want
to
add
a
read
in
for
this
file,
put
it
in
this
module
instead
of
randomly
somewhere
else.
F
So
I
kind
of
like
that
idea,
but
in
general
yeah
I
would
have
preferred
that
these,
like
there's
two
parts
of
this
pr
that
I
think,
should
have
been
separated
in
the
first.
But.
A
All
right,
thanks
dear.
K
Yes,
last
week
was
mainly
this
automatic
reading
that
is
now
in
emg
3d,
which
is
now
within
the
consortium.
It's
now
easier
for
the
for
the
others
to
with
the
command
line
interface
throw
their
model
they
have,
they
have
it
usually
from
gravity,
so
it
has
very
different
dimensions
than
my
cscm
mesh
and
it
does
all
the
meshing
as
sensible
as
possible
and
gives
it
back
on
the
same
scale.
K
L
John
actually
yeah
I've
been
helping
joe
and
lindsay
there
with
the
frequency
domain
dependency
for
the
natural
source.
I
started
the
pr
for
the
part
one
there
and
yeah
it's
going
really
well
pretty
much
by
defining
having
the
source
defined
that
it
has
the
multiple
polarizations
or
does
it
just
have
number
sources
or
sorry
I
guess
it.
It
just
hasn't
now
another
parameter
that
just
tells
it
if
there's
multiple
polarizations
and
yeah.
L
I
took
joe
pushed
in
last
night
and
I
train-
or
I
put
it
all
into
the
other
receivers
tested
that
out
so
the
tipper
and
the
complex
resistivity
or
I
guess
the
resistivity
and
phase
now
all
work
and
pretty
much.
The
simulation
class
is
super
lightweight
now
yeah,
I
guess
I'll
start
cleaning
up
here
going
the
next
couple
days
and
yeah,
maybe
I'll
point
out
some
things
that
we
can
improve
or
if
there's
anything
else,
that
we
can
look
at
but
yeah
that
pr
should
be
ready.
H
L
Yeah,
I
know
it's
it's
all
the
forward.
Modeling
stuff
is
we
just
have
like
you
said
I
just
have
to
modify
the
me.
Sigma
derives
just
to
catch
the
right
sizes
and
yeah.
The
inversion
should
be
working.
That's
just
as
far
as
I
got
last
night
is
to
yeah
just
having
to
modify
that,
and
I
got
a
little
tired,
okay,
yeah.
No,
that's
all
good
and
yeah.
I've
just
been
dabbling.
L
Still
in
the
tiling
side
for
the
dc
yeah,
I
found
out
exactly
where
it's
breaking
and
I'm
just
gonna
see
what
I
can
do,
but
that's
about
it.
L
I
it
really
shouldn't
be
that
long
like
as
soon
as
we
get
it
for
the
dc.
I
really
just
have
to
you
just
have
to
make
the
tool
that
divides
it
up
the
simulations
of
the
frequencies.
So
I
don't
know
I'd
like
to
say
it
would
be
like
a
week's
worth
of
work.
Maybe
nothing
too
much.
Yeah.
I
Excellent
soggy,
I
had
one
more
question
for
for
them.
I'm
assuming
you
guys
are
doing
this
all
for
tensor
mesh
for
now.
Is
it
in
your
immediate
plans
to
get
it
working
for
tree.
F
F
A
Excellent
thanks
john
soggy.
J
Have
nothing
really
particular
irrelevance
here,
I
kind
of
maybe
I
was
a
little
bit
frustrated.
Getting
like
a
kid
comments
from
jenny.
I
think
she's
italian
girl,
like
she's,
trying
to
use
both
sim
pack
and
sim
peggy
m1d
and
the
structure
is
very
different.
She's
like
that
she's
using
the
older
version,
so
in
the
problem
class.
Yes,
I'm
just
keep
getting
kind
of
questions
from
her.
It's
not
matching,
it's
not.
It
doesn't
work
and
I'm
trying
to
kind
of
explain
bits
and
pieces.
J
Yes
like
it's
actually
because
of
the
lack
of
document
good
documentation.
I
guess,
and
also
it's
not
just
the
documentation,
but
I
wasn't
sure,
what's
the
best
way,
she's
like
feels
like
there's
like
quite
a
bit
of
knowledge
gap
between
users
and
developers-
and
I
don't
know
yeah
I
wasn't
like.
I
was
just
kept
frustrated
because,
like
I
just
keep
answering
her
questions
and
explain
like
as
much
as
I
can
and
then
what
keeps
coming
back
it
doesn't
work,
doesn't
work
and
it's
not
matching
yeah.
I
was,
I
wasn't
yeah.
J
I
wasn't
just
sure
how
like
a
what
is
the
best
approach
to
really
make
it
useful
for
people
do
not
necessarily
have
a
good
understanding
behind
about
the
physics
and
the
problem
and
data
anyway,
that's
that's
where
I
felt
and
a
little
bit
upside.
I
was
working
with
them
a
little
bit
about
the
quest
data,
so
there
is
a
big
ip
factor
in
airborne
data,
and
this
is
something
that
I
haven't
done
like
I've
worked
on
small
data
set,
but
this
is
actually
pretty
big
data
set
and
don't
already
work
on.
J
There
was
a
talk
in
the
sdg
actually
in
the
machine
learning
session.
Who
talked
about
the
negative,
so
I
was
excited.
Oh
okay,
we
can
actually
revisit
that
so
we're
going
to
like
we're
in
the
process
of
revisiting
and
see
if
what
correlation
we
can
find
from
airborne
ip
and
the
geochemical
data
sets
so
anyway,
that's
where
am
I.
A
Very
cool:
well,
we
can
maybe
circle
back,
I
think,
soggy
your
comment
or
about
sort
of
users
and
documentation.
I
think
ties
quite
nicely
into
doug's
question
about
where
we
should
be
documenting
things
that
are
more
on
sort
of
that
intersection
of
you
know,
theory
and
practice.
So
I
think
that
that
all
feeds
in
quite
nicely.
J
Yeah,
like
just
one
comment,
was
my
feeling:
if
you
got
kind
of
tight
connection
like
let's
say
we
can
set
up
call
and
just
like
have
a
casual
talk,
then
it's
sort
of
fine
but
like
jenny,
is
an
example.
Okay,
we're
not
going
to
have
a
very
casual
conversation
with
her,
but
like
basically
we're
interacting
with
coats
and
some
males
or
whatever,
like
in
a
documented
media
yeah.
I
think
it
would
mean
some
sort
of
effective
mechanism,
even
if
even
just
with
that
we
can
interact.
A
Yeah
yeah:
that's
a
really
good
point
I'll,
give
a
couple
updates
and
then
we
can
maybe
circle
back
to
some
of
these.
These
discussion
points
so
a
piece
of
good
news
is:
we
did
get
the
iris
funding
for
developing
a
processing
code
for
for
mt,
so
really
excited
about
that.
A
We'll
have
some
more
details
and
we
should
be
getting
in
touch
with
the
iris
folks,
hopefully,
either
later
this
week
or
early
next,
to
kind
of
get
a
get
a
sense
of
getting
things
up
and
running,
and
but
they
were
quite
positive
on
sort
of
our
community
approach,
and
so
I'm
very
excited
about
that.
A
Still
funding
so
the
the
goal
is
to
actually
go
in
and
basically
develop
an
open
source
version
of
gary
egbert's
mt
processing
code.
Well,
there's
two
aspects:
there's
that
and
then
there's
also
some
work
on
mth5,
which
is
meant
to
be
an
archival
and
storage
format
for
mt
for
processed
process
data.
A
A
A
I
mean
it
could
do
a
small
slice
of
sort
of
some
internship
work,
but
yeah
it'll
be
mostly
for
development,
so
this
is
with
we
put
it
in
with
carl
kapler
and
so
he'll
he'll
be
doing
a
lot
of
the
development
work
and
he's
got
a
lot
of
mt
experience
and
he's
currently
in
california,
so
a
cool
chance
to
to
onboard
somebody
new
to
the
community
too
other
items
yeah.
I
guess
joe
mentioned
this
already.
A
I
added
some
notes
on
that
discretize
pr
with
respect
to
naming,
so
we
can.
We
can
circle
back
to
that
and
then
the
other
big
item
that
I'd
appreciate
folks
taking
a
chance
to
to
read
through
is
I
started
to
sketch
out
a
document
just
to
kind
of
start
a
conversation
about
what
we
want
to
have
in
place
for
simpeg
governance.
A
This
is
the
goal.
My
goal
here
is
to
have
this
be
a
very
conversational
document.
So
please
feel
free
to
add
your
own
notes,
thoughts
comments.
It's
meant
to
be
a
starting
place
for
for
a
discussion,
and
so
we
can
sort
of
have
that
document.
Get
your
get
your
thoughts
down
and
maybe
either
next
week
or
in
two
weeks
time.
We
can
maybe
have
a
bit
more
of
a
focused
meeting,
just
kind
of
thinking
through
some
of
some
of
the
governance
items.
A
What
kind
of
questions
we
need
to
be
able
to
answer
and
what
we
want
to
have
in
place
and
to
feel
free
to
ping
me
if
there's
anything,
you
want
to
chat
about
one-on-one
with
this
or
anything
like
that.
I'm
very
happy
to.
I
think
this
is
an
important
conversation
for
us
for
us
to
have
and
yeah
getting
getting
your
thoughts
down
would
be
much
appreciated.
A
So
with
that,
let's
maybe
circle
back
to
some
discussion
items
so
there's
three
things
is
joe
brought
up
two
with
respect
to
mostly
naming
with
pull
requests,
and
then
we
have
the
discussion
with
respect
to
sort
of
the
simulation
forward.
Modeling
capturing
this,
this
theory
process
sort
of
interaction.
F
Just
a
few
things
so
generally
like
I
brought
up
in
here,
is
that
the
as
long
as
like
I
don't
have
anything
against
any
changing
anything.
That's
not
a
keyword
argument.
It
doesn't
really
matter
it's.
You
know
the
user
will
still
call
it
with
the
same
thing.
It
doesn't
matter,
so
we
can
change
them
whether
or
not
and
that's
fine.
The
other
thing
is
I'd
like
to
talk
a
little
bit
more
about
this
vne
property
and
the
pull
request
they
taught
lindsey
mentioned,
naming
it
as
shape.
F
I
I've
kind
of
pushed
back
and
naming
it
as
shape
because
it's
not
actually
a
shape.
It's
like
a
it's
a
list.
It's
like
a
list
of
the
number
of
edges
in
each
direction
like
along
each
component
so
like.
If
it's
going
to
be
named
anything,
I
think
it
should
be
closer
to
that,
and
even
if
it's
like
explicit,
what
did
I
put
on
there.
F
F
A
The
other
one
that
I
had
in
mind
was
the
grid
is:
do
we
want
to
call
things
so
right
now,
we've
got
like
grid
cc
grid,
and,
and
so
when
I
was
looking
at
the
code,
I
found
actually
sort
of
the
grid
cell
centers.
Well,
I
don't
know
if
we
want
the
grid.
Do
we
just
want
to
call
it
cell
centers
faces
x,
faces
y
faces,
z,
yeah,.
F
F
Right,
I
I
so
the
two
arguments
like
four
leaving
it
as
a
grid
is
that
okay,
well,
we
kind
of
have
this
bias
in
our
head.
That
yeah
grid
does
imply
some
sort
of
structure
to
it,
but
then
there's
also,
you
know
unstructured
grids
unstructured
these
other
things.
So
it's
like
okay,
that's
kind
of
an
argument
against
that
and
then
having
them
all
as
grids
allows
you
to
do
tab,
completion
to
see
what
lists
of
locations
there
are
on
on
when
you're
writing
the
code,
that's
kind
of
the
only
arguments
for
keeping
it.
J
F
J
Yeah,
I
think,
as
long
as
like
there's
a
little
bit
of
buffering
time
that
I
could,
like
oh
okay,
this
was
degree
cc
so
like
you're
gonna
kind
of
leave
like
at
this,
I'm
not
sure
what
it's
called,
but
that
I
can
still
use
degrees
cc
right.
Then
it's
a
certain
point,
we'll
we'll
duplicate
that
yeah.
To
do
that,
I
think
I've
really
opened
that.
F
I
think
the
only
the
only
kind
of
issue
that
might
cause
is
that
one
of
the
old,
I
think
one
of
the
old
like
if
you
wanted
the
edge
lengths
previously
you
would
be
like
oh
meshed,
up
edges
or
something
to
get
the
edge
lengths,
which
didn't
really
quite
make
sense.
So
we
might
have
a
little
bit
of
issue
with
that,
but
I
don't
I
don't
know.
I
have
to
look
because
we
have
like
edges,
x,
edges,
y
edges,
z,
yeah,.
J
F
My
my
thing
was
that
I,
like
I,
don't
mind
changing
that
like
I
just
should
just
be
like
edge
lengths
face
areas.
You
know
that's
what
I
they
are
going
to
be
and
that's
what
it
says
in
there.
I'm
just
saying
it'll
cause
a
little
bit
of
clash
for
renaming
something,
as
you
know,
edges
that
now
does
something
completely
different
than
what
it
used
to
just
a
little
bit
of
a
name
clashing
there.
Yeah.
A
A
Cc
a
lot,
that's
one
that
I
think
everyone's
touched
is
because
you're
building
building
a
model
but
yeah,
I
wouldn't
be
too
overly
concerned.
What
actually
prompted
me
with
sort
of
thinking,
revisiting
sort
of
getting
rid
of
grid
is
actually
looking
at
a
lot
of
the
code
like
when
we
do
it
internally.
You
end
up
just
saying
nodes
equals
like
mesh
dot
grid
nodes,
and
so
like
that.
A
That
naming
difference
that
that's
sort
of
what
prompted
me
to
to
really
think
that,
maybe
we
should
we
should
just
get
rid
of
the
grid
and
just
call
it
cell
centers,
like
I'm.
F
A
Okay,
well,
if
anyone
doesn't
want
to
disagree
on
camera,
you
have
time
to
chime
in
on
the
github
pull
request,
but
otherwise.
Well
we'll
take
the
nod
says
we
have
consensus.
A
Cool
joe
and
doug,
this
one
comes
to
doug's
point
about
the
names
for
fictitious
source
doug.
Would
you
like
to
maybe
give
just
like
a
quick
summary
of
what
you
had
mentioned
there
and
we
can
open
that
up.
H
I
guess
it
starts
with
actor
having
gone
through
the
the
code
that
I
didn't
really
understand
what
was
going
on,
and
I
didn't
really
know
where
these
fictitious
sources
are
coming
from
and
how
they're,
how
they're
being
used
so
come
back
to
that,
but
just
the
name
itself
for
dc
resistivity
a
fictitious
source
is
often
historically
like
an
image
source
or
something
like
that.
You
put
a
current
in
the
ground
and
then
you've
got
an
image
above
and
so
then
that
naming
is
it
will
be
confusing
to
to
to
people.
H
H
The
fact
that
you're
working
in
a
you
know
a
slightly
different
domain,
so
the
the
whole
business
of
there's
a
naming
thing
with
fictitious
sources,
but
there's
also
just
a
an
implementation
part
that
really
worries
me
with
respect
to
when
you've
got
situations
with
topography,
because
I
don't
see
how
you
can
have
you
could
just
shove
in
an
analytic
source
for
a
half
space
and
have
it
you
know,
have
it
applicable
in
all
kinds
of
situations.
I
brought
up
one
where
th
this
is
sort
of
like
a
sanity
check.
H
Is
this
going
to
work
in
a
case
where
you've
got
quarter
spaces,
but
the
other
one
that
I
think
could
be
could
be
useful
not
only
for
the
dc
resistivity
but
might
actually
be
sort
of
a
generic
thing.
We'd
use
in
all
of
our
em,
and
that
is
the
case
where
we've
got
topography
and
just
represented
by,
I
know
sort
of
like
a
trapezoid,
so
we've
got
something
you
can
imagine
a
flat
surface
and
then
you've
got
a
mountain.
H
That's
just
made
out
of
you
know
two
sloping
lines
and
it's
got
a
you
know
a
variable
flat
surface
and
we
had
something
like
that.
Then
we
could
just
kind
of
use
that,
as
as
a
bit
of
a
check
to
see
when
things
start
to
break
down
and
then
to
decide
whether
we
want
to
have
these
app,
you
know
these
fictitious
sources
still
be.
What
put
you
know
what
devin
is
programmed
in,
or
maybe
we
want
to
think
about
something
something
else.
I
J
Choice,
I
think
what
roman
did
was
more
like
a
sophisticated,
because
that's
how
the
ubc
code
works,
the
the
object
code,
not
the
tensor
code,
but
actually
he
used
this.
This
imaging
imaging
principle
to
derive
like,
like
he
take
account
those
topography
effects.
So
I
think
that's
how
that
the
ubc
dc
auction
code
works
yeah.
I
I
talked
to
a
woman
and
then
I
implemented
the
same
thing
that
the
dcip
octree
does.
J
Okay,
so
then
I
think
it's
interesting
well,
but
it's
still
worthwhile
to
test
when,
like
we
got
typography.
J
I
F
To
to
doug's
other
point,
though,
this
was
kind
of
one
of
the
reasons
I
brought
it
before
about
having
a
more
general
form
of
this
kind
of
right
hand,
side
solution
to
where
okay,
I
okay
yeah,
I
like
the
name
corrective
source
better,
because
that's
actually
what
this
is
doing.
If
you
look
at
some
of
the
papers
that
the
like
links
there,
that's
this
is
the
same
formulation
as
that
kind
of
corrective
source.
F
So
what
that
I
mean
we
should
I
I
would
like
I
like
the
idea
of
just
kind
of
being
able
to
give
it
have
the
source
and
be
able
to
give
it
a
just
ask
for
a
solution
back
from
it
for
a
primary
field.
It
doesn't
matter
how
it
does
it
just
be
able
to
give
it
a
function
that
given
a
source
location
it'll,
give
you
back
analytic
solution
or
give
you
back
a
solution.
It
doesn't
have
to
be
analytic
kind
of
makes.
You
think
that
we
could.
F
We
could
mix
maybe
integral
techniques,
to
account
for
topography
in
some
sense
in
some
situations,
because
that's
a
little
bit
easier
to
account
for
those
and
that
kind
of
thing
or
analytic
sense
or
we
could.
You
know
as
you
guys,
as
the
other
said,
mixing
fine
grain
actually
solutions
to
this
in
those
simulations.
That's
kind
of.
I
would
like
to
move
that
forward
with
that
kind
of
thing.
A
Yeah
and
I
think
to
to
doug's
point
well
correct
me
if
I'm
misinterpreting,
but
I
think,
there's
an
important
question
here
too
of
when
do.
We
include
something
in
simpeg
when
it
is
something
like
this.
That
is
not
necessarily
like
we're
not
at
the
stage
where
a
user
can
pick
this
up
and
sort
of
use
it
responsibly,
necessarily
right,
as
we
haven't
actually
vetted
if
it's
working
with
topography,
what
what
sorts
of
like
edge
cases
we
need
to
be
careful
with.
A
So
I
think
it's
it's
an
important
question
as
to
like.
When
do
we
actually
want
to
include
something
in
simpeg,
and
if
we
do
includes
this
like?
Should
we
be
throwing
warnings
right
now
saying
that
this
is
this
is
experimental
or
should
we
hold
off
until
we
sort
of
think
through
a
bit
of
a
more
general
implementation
that
could
account
for
topography?
A
H
To
be
really
certain
in
our
own
minds,
what
we're
actually
implementing
and
testing
it-
and
you
know
getting
some
notes
on
it,
so
that
people
can
understand
it
and
then
then
release
it.
Otherwise
people
can
potentially
misuse
it
or
get
confused
and
that's
ultimately
going
to
be
a
bigger
sink
for
our
time.
I
Yeah
I
like
the
idea
of
the
of
the
warnings,
because
I
guess
I
don't
like
the
situation
where,
like
we're
developing
something
and
it's
it's
getting
close
to
the
finish
line
and
then
because
it's
not
perfect,
we
don't
want
to
release
it
and
there's
all
these
these
tools
and
there's
all
these
things
that
people
could
be
using.
But
then
we
we
hold
off
because
we're
being
very
careful-
and
I
I
like
the
idea
of
putting
some
kind
of
warning
that
says.
Okay,
you
have
extreme
topography.
I
This
particular
formulation
might
have
large
associated
errors,
like
you,
you're-
probably
not
implementing
this
properly,
because
it
allows
us
to
really
like
regularly
release
improved
tools
to
the
community.
But
we
do
want
to
make
sure
that
whatever
it
is,
we
release
is,
is
working
properly
and
that
people
are
implementing
it
properly.
It's
a
very
good
point.
F
I
would
say
like
we
could
just
implement
the
warning
on
the
class
as
a
whole,
just
being
like
hey
this
may
change
in
the
future.
It's
currently
experimental
it
has
not
been
tested.
You
know
under
topography
effects
or
variable
topography,
just
things
like
that.
Like
I,
I
would
be
okay
with
putting
that
kind
of
warning
on
it.
F
So
it's
transparent
and
saying
that
hey
this
might
this
will
likely
change
in
the
future
cause.
That
is
something
that
I
mean
other
other
code
bases.
Do
that
kind
of
thing
too,
where
they
put
in
mornings
future
mornings
mornings
that,
oh,
this
you
know
code
is
going
to
change,
don't
use
it
like
this.
A
So
I
that's
one
potential
solution,
but
I
think
this
is
more
than
just
a
this
code
might
change
it's
that
we
haven't
verified
like
where
this
code
works,
and
I
think
that
those
are
two
distinct
things
and
it's
important
to
sort
of
like.
If,
if
we
include
this
in
the
simpeg
package,
I
mean
there,
there
is
sort
of
a
like
at
this
point,
a
responsibility
that
it's
working
and
so
it's
something
to
sort
of
think
through
yeah.
I
Well,
I
can
do
that.
I
think,
sometime
in
the
next
few
days,
we
have
an
analytics
solution
for
a
hemispherical
bite
getting
taken
out
of
the
surface.
I
believe,
although
I
think
that
I
don't
think
you
can
put
the
sources
and
receivers
in
that
bite,
but
I'll
look
around
I'll,
try
and
find
an
analytic
solution
where
there
is
some
kind
of
strange,
large
topography
and
I'll
throw
the
whole
suite
of
of
formulations.
H
And
that
particular
formulation
could
either
be
a
hemispherical
dimple
or
a
hemispherical
lump.
Is
that
correct.
I
Yeah,
the
only
thing
I'm
not
sure
of
is,
if
I
can
put
the
electrodes
inside
it,
and
I
don't
think
I
can,
I
think
you
put
them
on
either
side
of
the
depression
or
the
lump.
A
Excellent
cool,
okay,
we'll
look
forward
to
sort
of
to
seeing
that
and
if
folks
have
other
thoughts
on
this
feel
free
to
chime
in
on
that
full
request,
and
we
can
keep
keep
iterating
there.
F
I
think
at
this
point
then
I'll
probably
do
a
release
of
sempeg
for
now,
without
that
in
it
just,
I
think
the
pr
should
be
coming
a
lot
quicker
now
that
we
have
the
little
bug
worked
out.
That
was
causing
me
so
many
issues,
so
pierre
should
be
coming
in
quicker.
A
A
Where
do
we
perhaps
want
to
put
resources
on
computation
on,
and
I
think
this
also
speaks
to
soggy's
point
of
like
what
sorts
of
tutorials
do?
We
perhaps
need
to
get
folks
up
and
running
so
that
we
can
point
people
to
resources
rather
than
one
person
having
to
answer
answer
emails.
H
H
You
know
which
formulation
is
being
used
and
how
things
are
being
you
know
evaluated
and
then
make
a
complete
correspondence
with
the
you
know
with
the
code
so
that
you
know,
if
there's
a
particular
formula.
That's
that's
used
that
is
readable
in
in
python
or
there's
minus
signs
or
something
that's
coming
in
because
of
this.
But
you
kind
of
understand
where
that's
coming
from.
A
Yeah,
so
I
actually,
I
think,
there's
okay,
because
I
think
there's
actually
two
sort
of
levels
of
resources,
potentially
here
and
dieter
I'll.
Let
you
chime
in
with
your
comment
too
and
devin
here,
but
I
think
so.
What
I
potentially
see
is:
yes,
there's
the
there's,
the
theory
section
or
the
packages
section
that
devin
pointed
out
in
the
documentation,
but
in
some
sense
is
it
I
don't
know
how
in
depth.
A
We
want
to
go
into
theory
in
the
documentation
or
if
we
want
to
start
to
think
about
a
collection
of
more
lightweight
pages
or
like
a
blog
or
a
book
kind
of
thing.
That
includes
tutorial
information
and
sort
of
descriptions
and
a
bit
more
depth
with
respect
to
some
of
these,
some
of
these
concepts,
theater
I'll.
Let
you
speak
to
to
your
comment
here
too.
K
It
wasn't
really
just
something
to
keep
in
mind
because
doug
said
you
said
by
python:
it's
nice,
you
can
just
read
it,
but
it
can
be
very
hard
to
follow
it,
and
I
think
it's
not
actually
a
python
thing.
So,
but
it's
a
thing
from
the
objective
programming
paradigm
that
everything
is
inherited
and
has
a
state
and
and
particularly,
if
you're,
not
often
using
at
the
code.
This
can
make
it
very
difficult
to
follow
it,
which
yeah
just
yeah.
I
don't
think
it's
a
python
characteristic.
C
H
Dealer
within
your
consortium,
like
what
sort
of
kind
of
communication
problems
have
you
had,
because
you're
working
with
other
you
know
with
other
groups
and
you're
sort
of
you
know,
they're
using
part
of
your
codes
and
packages
and
you're
using
part
of
them
and
at
what
level
have
the
communications
been
challenging?.
K
K
A
K
Which
makes
it
sorry,
which
makes
it
very-
I
mean
it's
not
open
source,
that
framework
right.
My
code
is,
but
all
the
rest
not,
and
it
makes
it
completely
useless
for
anyone
else.
So
we
talk
here
at
zimper,
go
after
completely
different
beasts,
because
it
it's
it's
used
by
many.
K
I
I
But
it
sounds
like
this
is
sort
of
answering
a
lot
of
questions
about
where
to
put
stuff
and
we
all
seemed
like
we
were
pretty
okay
with
the
structure,
some
of
the
stuff
like
analytics,
we
decided
to
put
in
the
jioana,
but
there
is
a
place
for
theory,
and
it
does
separate
this.
I
This
theory
section
which
we
all
seem
to
kind
of
want
that
separate
from
what
you
would
be
putting
in
the
doc
strings
of
the
api
and
and
if
it's
a
priority,
then
I'm
totally
willing
to
to
go
for
it.
I
think
I
was
maybe
diverted
to
spend
some
time
on
the
discretize
website,
but
if
this
is
what
people
really
need
right
now,
I'm
willing
to
go
work
on
it.
I
F
I
Yeah
yeah,
sorry
about
that,
but
yeah
I
could.
I
could
spend
time
and
energy
on
on
that.
If
it's,
if
it's
a
high
priority
item,
no
problem.
J
Ahead
was
curious
about
where
to
put
like,
for
instance,
what
doug
asked
for
was
very
gory
details,
yeah
how
coach
is
really
implementing?
What
is
the
background
theory,
so
you
need
like
quite
a
bit
of
space
to
put
that
in,
and
I
was
curious
about
whether
we
want
to
put
that
into
specific
documentation
or
like
a.
We
have
a
separate
like
like
a
kind
of
ubc
style
manual
style
document
like
we
put,
you
dump
everything
into
the
documentation,
all
right
yeah.
J
I
I
don't
think
it
has
to
be
like
crazy,
gory
details,
but
it
would
be
nice
to
just
start
off
with
the
you
know,
the
pde
or
or
whatever,
where
we're
starting
whatever
maxwell's
equations,
how
we've
manipulated
it
for
a
particular
problem
and
then
just
show
a
reasonable
amount
of
steps
to
get
to
the
final
discretized
system,
and
just
say
you
know
what
kind
of
a
boundary
condition
was
implemented
and
then
what
does
that
sort
of
mean
for
how
you're
going
to
set
up
your
particular
simulation,
like
just
making
that
connection,
I
think,
is
maybe
the
the
biggest
thing
we
want
to
put
in
there.
H
I
would
kind
of
agree
with
that.
I
think
that
you
know
if
we
just
have
enough,
at
least
for
me
on
this
particular
example,
just
just
enough,
so
that
we
could
actually
start
at
square
one
and
either
have
references
or
young
little
cubicle,
diagrams
fields
and
fluxes
and
where
things
are
and
what
the
equations
are
and
then
how
the
steps
are
going
through
yeah,
a
very
kind
of
summary,
with
respect
to
yeah
what
the
process
is
and
what
the
equations
are.
Yeah.
E
H
I
I
I
I've
regularly
pointed
people
at
theory,
sections
for
ubc
manuals,
because
then
you
are
like
you
understand
what
the
formulation
is
and
how
they're
solving
it.
So
I
I
think
that's
like
a
starting
template.
We
might
cater
it
a
bit
more
to
the
simpeg
audience,
but
there
is
something
that
exists.
That
is,
is
being
used
right
now
and
somewhat
successfully.
H
H
D
Yeah,
I
just
want
to
follow
up
on
what
the
devil
and
doc
mentioned.
I
think
yeah
doug
is
exactly
right,
so
we
we
have
equations
but
there's
a
big
gap
between
equations
and
implementation.
So
what
I
maybe
I'm
out
of
my
mind,
maybe
the
tools
are
not
existing
yet,
but
what
have
what
I'm
envision,
ideally,
is
that
we'll
have
a
like
a
web
page.
It's
super
long
and
on
the
very
left
hand
side.
D
You
have
the
four
equations
and
then
you
go
to
the
right,
the
next
level
on
right
next
to
the
equations,
and
then
you
you
get
into
the
discretizations
and
then
you
get
to
into
the
next
level
to
the
right.
And
then
you
got
the
the
equation
that
you
actually
use
to
code
and
then
the
last
level
be
the
codes
and
when
the
user
click
one
encryption
and
all
the
codes
that
are
used
to
implement
this
particular
question
will
be
highlighted
and
when
you,
so
I
don't
know
if
that
the
tools
exist
or
not.
D
But
that
would
be
really
really
helpful.
You.
You
will
go
from
macro
equation
for
equations
all
the
way
to
the
exact
details
of
how
things
are
coded
right,
and
I
think
we
have
the
pieces
probably
out
there
most
of
most
of
the
pcs.
It's
just
a
way
of
how
to
integrate
them
together
into
a
user-friendly.
D
I
don't
know
a
browser
or
web
page,
something
like
that.
Something
is
missing
there
to
to
to
understand
the
code
where
you
have
to
be
really
familiar
with
the
with
the
the
maxwell,
the
em
theory,
and
also
you
need
to
be
really
good
at
the
numerical
solutions.
How
how
things
are
discretized
finite
volume,
final
element,
then
you
need
to
be
also
familiar
with
syn
with
python
coding,
which
is
quite
challenging
for
for
non-core
developers.
Like
me,
when
I
read
the
code,
I
know
python
is
pretty
proud
of
its
readability.
J
We
need
to
have
some
sort
of
tangible
plan,
so
it'll
be
great
jojo.
If
you
could
do
that,
I
think
the
current
sort
of
status
that
we
face
it's
bit
far
to
do
that.
So
I
think
we
could
probably
make.
D
D
D
That
would
require
somewhere
else
to
link
my
equations
to
your
code
right
when
someone
clicks
a
click,
the
that
probably
like,
let's
say,
100
questions,
anyone
click
one
equation
and
then
that
degree
will
be
linked
to
all
the
codes
in
simpac
that
is
used
to
implement
the
code
last
step.
I
definitely
cannot
do
because
I
didn't
do
the
code.
There
would
be
someone
like
devin,
who
are
familiar
with
all
from
theory
to
this
transition
to
equation
to
codes
to
do
that.
H
I
I'm
kind
of
with
soggy
a
little
bit
on
trying
to
maybe
do
this
incrementally.
H
Kind
of
got
like
the
bare
bones
stuff
of
like
okay:
this
is
this
is
how
we
get
from
from
a
to
z,
yeah
yeah,
and
then
that
would
form
then
the
basis
for
now.
We've
got
all
these
pieces,
so
here's
now
we
could
enact
your
vision
of
you
know
just
clicking
on
something
and
seeing
where
it
goes.
So
I
think
I
think
it's
a
that's
a
great
vision,
but
yeah
we
might
want
to
have
something.
That's
kind
of
intermediate
at
this
point.
Oh.
F
J
J
Is
there
a
good
way
to
sort
of
reference
where
the
api
like
say,
here's
like
we
were
solving
a
x
equal
to
b
and
here's
a
and
then
that
like
this,
is
where
the
code
is
to
generate
the
a
can.
We
like
refer
easily
okay,
I
think
that's
probably
then
sort
of
incremental
step.
We
can
make
basically
describe
the
pde
discretize
and
point
to
where
the
api
is.
So,
if
really
interesting,
gory
details,
they
can
take
a
look
at
the
code.
That's.
I
True
yeah
because
the
website's
going
to
auto
generate
the
api
and
then
it
will
auto
generate
the
hyperlinks
and
then,
whenever
we
make
this
this
theory
section,
I
think
those
will
be
created
after
the
api
is
auto
generated.
So
you
should
be
able
to
go,
find
it.
A
So
to
kind
of
make
progress
on
this,
you
know
we're
a
touch
over
time,
I'll,
throw
it
a
suggestion,
and
we
can
see
if
this
this
makes
sense,
because
I
think
that,
like
we,
we
want
to
figure
out.
I
guess
where
different
pieces
of
information
end
up
fitting.
I
think
that's
like
the
content
generation
is
is,
of
course,
a
challenge,
but
right
now
I
think
the
the
thing
we're
facing
is
sort
of
where
do
bits
and
pieces
of
information
sit.
A
Would
it
be
helpful
to
just
like
start
with
a
question
like,
like
you,
had
a
couple
questions
sort
of
thought
about
with
and
jaja
as
well
with
respect
to?
Basically
how
do
you
go
from
this
equation
to
to
the
implementation
and,
and
then
from
a
user
perspective?
How
do
you
sort
of
responsibly
use
that
implementation?
What
are
things
you
need
to
be
aware
of,
like
boundary
conditions
or
and
such
would
it
be
helpful
to
sort
of
just
like
start
with
a
single
question
and
then
outline
kind
of
step
by
step?
A
What
you
would
like
to
see
to
address
that
question
and
then
we
can
figure
out
where
those
pieces
of
information
kind
of
fit?
Does
that
sound?
Like
a
concrete?
That
sounds
like.
H
That
might
be
a
good
step
forward
kind
of
concrete
with
respect
to
yeah,
and
then
we've
got
the
dc,
so
we've
already
really
focused
in
on
something.
So
then
it's
a
matter
of
you
know
of
kind
of
parsing
out.
Basically
what
that
pull
request
is
at
this
point-
and
you
know
formulating
a
couple
of
specific
questions
that
are
connected
with
that
and
then
see.
Okay,
what's
required
to
answer
it
and
where
should
everything
live?
Can
we
just
have
links
to
specific
papers?
Do
we
actually
have
you
know
another
document
that
is
providing
missing
details?
H
You
just
try
to
sort
of
sort
that
out
as
we're
trying
to
get
something
yeah
I'd
be
happy
to
try
to
do
that.
Might
ask
for
some
assistance,
but
that
sounds
good.
A
That
sounds
good
and
jaja
if
you
have
time
to
also
even
just
document
a
single
example
of
just
like
here's,
some
pieces
of
information
that
I
feel
are
missing.
That
I
would
like
to
see
doesn't
need
to
be
anything
sort
of
in
depth,
but
I
think,
having
a
couple
perspectives
on
that
would
be
would
be
really
valuable.
I
What
we
want
to
do
with
this,
this
theory
section-
I
guess
I'd
like
to
put
something
down
as
like
a
goal
for
the
for
either
next
week's
meeting
or
or
the
week
after
something
that
really
ensures
it's
going
to
move
in
a
forward
direction.
A
So
I
think
we're
sort
of
in
the
think,
twice
act
once
phase
on
this,
and
so
maybe
one
thing
that
would
be
helpful
is
if
you
want
to
sort
of
put
down
a
couple
thoughts
on
maybe
a
bit
of
a
template.
What
you
have
in
mind
for
dc,
for
example,
and
just
kind
of
get
some
thoughts
down
on
this
google
doc.
And
then
we
can
sort
of
take,
take
that
and
see
if
it
matches
with
doug's
questions
and
jojo's
questions.
A
And
if
there's
a
trajectory
through
that
or
if
we
need
to
kind
of
rethink,
rethink
based
on
based
on
the
questions
that
are
that
are
being
asked.
A
No
quiet
all
right:
well,
thanks
everyone
we'll
call
it
a
wrap
and
let
you
go
have
go,
have
lunch
and
we'll
see
y'all
next
week.