►
From YouTube: SimPEG Meeting December 15th
Description
SimPEG weekly meeting from December 15th, 2021
A
B
Nice
to
see
you
guys
this
week,
so
it's
probably
our
last
meeting
for
the
month
of
december.
I
imagine
unless
people
want
to
get
together
next
wednesday.
B
At
some
point
I
know
it's
very
close
to
christmas,
but
if
you
guys
are
up
for
it,
I'm
I
can
host
something
I
will
be
up
in
the
mountains
down
in
denver.
So
I
will
see
we'll
see
if
anybody
wants
to
do
anything
next
week,
but
no
actual
plant
meeting
as
well
as
the
week
after
right
between
the
senior
years.
Obviously
so
this
will
probably
be
a
quick
meeting
today,
just
to
get
some
updates
from
where
people
are.
B
From
from
my
end,
we've
got,
I
got
the
the
cross
gradient
and
joint
total
variation.
Like
I
told
you
guys
last
week
I
got
this
pulled
in.
I
had
a
presentative
hoster
on
it
at
agu
is
pretty
pretty
quick,
but
I'm
still
having
thoughts
on
ways
that
we
can
improve
the
whole
joint
inversion
framework
itself.
B
What
we
do
now,
it's
it's
still
like
it
fits
it
works
it's
good
to
come
in,
but
I
feel
like
it
still
could
use
some
improvement
on
our
kind
of
systems
and
framework
and
make
it
a
little
bit
more
flexible
or.
B
A
little
bit
more
clear
on
the
steps
that
you
would
have
to
do
to
implement
your
own
jointed
version.
It's
it's
very
right
now.
It's
like
you
have
to
use
the
right
directives,
depending
on
which
kind
of
thing
you're
using,
but
it's
just
kind
of
a
little
bit
unclear
and
you
have
to
make
sure
you
do
everything
still
correct,
and
I
mean
in
the
correct
order
you
have
to
put
all
of
your
like
regularization
functions
in
the
correct
order.
All
of
your
data
misfits
in
the
correct
order.
B
So
hopefully
something
that's
a
little
bit
more
flexible
than
that
and
get
that
implemented,
but
the
only
way
we're
going
to
do.
That
is
if
people
start
using
it
and
tell
us
where
it's
not
working
for
them
and
where
they'd
see
it
proved
as
well,
because
otherwise,
it's
just
just
me
just
me,
looking
at
it
and
improving
it
or
trying
to
figure
out
what
I
think
would
work
better.
B
B
B
D
B
Well,
no
one
is
actively
developing
at
it,
but
you
are
more
than
welcome
to
he's
like
I
don't
know
about
that,
but
it
was
also
interesting
because
we
were
trying
to
get
him.
He
was
trying
to
get
simpek
installed
on
his
personal
laptop
and
it
wasn't
working
for
some
reason
and
it
turned
out
that
he
is
on
a
32-bit
system
and
it
was
just
causing,
like
obviously
we
don't
distribute
32-bit
binaries
at
all
anymore
and
it's
kind
of
like
we
can't
kind
of
kind
of
forge,
won't,
distribute
32-bit
binaries.
E
Yes,
I'll
go
yeah.
I
finally
finished
up
the
cluster
parallelization
using
our
custom
workers
for
the
using
the
implicit
formation
yeah.
It
works
pretty
good.
We
got
like
a
10
million
cell
model
running
on
it
right
now,
and
all
the
workers
have
about
350
gigs
on
them
each
and
the
masternode,
maybe
like
two
gigs
just
to
hold
with
stuff.
So
none
of
that
big
meaty
stuff
has
to
come
back
to
master
er,
sorry
to
the
main
and
yeah
it
runs
pretty
quick
too.
E
All
the
volume
averaging
seems
to
there
doesn't
seem
to
be
too
much
to
diminish
accuracy.
So
far,
so
yeah
working
pretty
good,
and
now
I
can
get
back
to
other
things,
simple
related
yeah.
I
don't
know,
maybe
that
that
pull
request
for
the
mt
there
was
just
that
one
or
the
first
code
test.
That
was
just
a
little
bit
under.
I
don't
understand.
What's
the
difference
between
the
two
code
codes.
B
One
is
your,
so
one
is
the
diff
coverage,
so
it
checks
how
how
covered
the
the
lines
you've
added.
F
B
E
Okay,
well,
I
think
it's
only
like
a
few
percent
under,
so
it
shouldn't
take
much
to
finish
that
off
and
yeah.
If
it's
not
in
the
in
this
year,
maybe
the
new
year,
hopefully
we'll
get
it
get
it
in.
B
E
Oh
okay,
I
must
have
misunderstood.
I
thought
you
said
you
did
push
that
stuff
already.
Okay,
that
makes
sense.
C
Not
too
much
I
mean
well
just
plugging
away
at
this
paper
draft
and
I
had
like
a
little
internal
usgs
presentation
and
summed
up
the
project
I've
been
working
on
yesterday,
so
I
was
doing
some
work
for
that
and
just
trying
to
get
some
odds
and
ends
tidied
up
before
the
holidays.
So
we
can
get
a
draft
of
this
paper
out
for
a
few
people
to
look
at
internally
before
we
kind
of
ship
it
off
to
external
people.
So
just
plugging
away.
E
Was
that
mtn
gravity
that
you
were
doing.
C
Hope,
yeah!
That's
that's
to
come
at
this
point.
It's
just
kind
of
some
inversions
of
the
graph
of
some
old
gravity
data
that
we
kind
of
compiled
and
we'll
kind
of
do
another
individual
kind
of
inversion
of
some
airborne
magnetic
data
over
the
area
and
then
hopefully
use
some
of
tebow's
pgi
inversion
stuff
to
do
a
joint
version
of
the
gravity,
the
magnetics
and
then
this
mt
data
set
that
we're
currently
collecting.
F
Oh
yeah
yeah,
I'm
trying
to
finish
up
some
stuff
for
before
the
other
day,
one
last
manuscript
and
archiving
things
well,
the
office
is
clean,
but
not
too
much
on
yeah,
not
too
much
on
the
on
the
on
the
coding
side
like
we
need
to
check
out
what
you
did
with
all
the
parallelization
dk.
That
would
be
interesting
to
see.
D
F
G
I'm
mostly
just
working
with
utem
stuff,
some
of
the
sponsors
are
they
have
this
really
large
loop,
utem
system.
I
think
la
montan
makes
it
they're
trying
to
I
guess,
image
some
some
stuff.
I
guess
the
raglan
deposit
was
the
place
they're
interested
in
it's
a
it's,
a
data
sort
of
a
the
data
that
they
collect
and
how
they
represent.
G
The
data
is
a
little
bit
new
to
me,
so
I've
been
doing
some
modeling
with
that
and
then
trying
to
make
sort
of
a
comprehensive
workflow
for
how
you
would
process
that
and
get
inversion
results
so
playing
around
with
that,
and
there's
also
been
something
on
my
radar
about
mobile
mt
and
some
of
the
data.
G
It
looks
like
there's
a
transfer
function
between
horizontal
components
of
the
electric
field
and
the
vertical
component
of
the
magnetic
field,
which
is
something
a
little
bit
new
to
me,
I'm
kind
of
wondering
if
that's
something
that
people
thought
of
implementing
in
our
our
mt
code
for
simpeg.
I
don't
know
how
often
that
that
comes
up,
but
that's
something
that
I'll
I'll
start
on
sometime
in
the
near
future.
So
I'm
interested
in
what
what
simpeg
could
do
with
that.
E
E
I
haven't,
really
pursued
the
mobile
mt
style
so
much,
but
we
we
did
want
to
implement
that
in
the
simpeg
stuff,
so
that
we
could
just
have
something
to
compare
against
indian
version.
So
yeah.
G
Yeah
that'd
be
great,
I
mean
it
sounds
like
90
of
the
work
is
done.
It's
just
okay,
let's
add
a
let's.
Let's
add
the
computation
that
goes
from
the
field,
the
solution
of
the
fields
to
that
datum,
and
then
some
derivatives.
A
Yeah,
mostly
agu
stuff,
so
gave
a
poster
earlier
this
week
on
the
casing
casing
work
I
can
drop
a
we
should
joe.
I
don't
know.
Are
you
able
to
share
the
poster
link
outside
of
yeah?
Like
can.
A
Okay,
mine's
just
a
pdf
but
I'll
drop,
a
link
in
some
notes.
So
if
you'd
like
to
take
a
look,
you're
welcome
to
and
then
gave
a
pre-recorded
talk,
which
is
part
of
a
session
that
brooke
minsley's,
organizing
and
that'll
be
live
I'll.
Do
a
live
version
tomorrow.
So
I
need
to
condense
it.
I
only
have
10
minutes
and
my
first
cut
of
it
was
half
an
hour
yesterday.
So
that's
not
gonna
work,
but
but
that's
fine.
It
should
be
a
good
session.
So
looking
forward
to
it.