►
From YouTube: SimPEG meeting March 18
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
Get
on
to
more
normal
things
in
life.
Well,
thanks
everyone
for
for
taking
time!
I
know
life
is
crazy
in
some
places
right
now.
So
it's
good
to
see
you
all
in
see
everyone's
doing.
Okay,
one
quick
update
I
did
hear
both
from
Max
and
from
Ren
briefly
today.
So
Ren
is
on
a
flight
to
Taipei
and
then
to
Toronto
from
from
there.
So
hopefully
we'll
hear
tomorrow
that
she's
back
in
back
in
Canada,
so
yeah,
but
an
update
on
the
GW
B
set
of
things.
A
B
Iii
locket
and
I
think
Doug
mentioned
one
other,
so
they
seem
like
they're
able
to
sort
of
do
everything
and
teach
others,
and
so
I've
made
the
administrators
of
the
Facebook
group
and
I
got
I
posted
an
update
for
them
to
update
their
simpang
and
to
download
the
latest
notebooks.
So
now
that
that's
finalized
and
I
was
able
to
get
that
to
Ren
before
before
they
left
so
I
think
they'll
be
able
to
do
this.
B
Yeah
and
I
also
have
just
been
trying
out
the
that
functionality
and
I
like
the
new
notebooks.
The
tree
mash
seems
to
work
quite
well,
and
it
runs
very,
very
fast.
So
all
the
upgrades
that
you've
been
doing
lets
me,
you
know,
work
with,
say
three
to
six
thousand
cells
and
get
a
lot
out
of
it.
So
that's
been
quite
nice.
B
That's
were
the
update
on
gwb
and
and
yeah
looking
at
adding
some
more
of
the
tutorials,
so
we've
got
frequency
domain
a.m.
and
time
domain
tutorials
just
for
forward
modelling
that
are
gonna,
get
added,
I've
got
a
SIL
mesh
and
a
octree
mesh
for
each,
and
they
they
run
quite
quickly,
and
one
of
the
things
for
the
time
domain
that
I
focused
on
was
just
a
very
good
explanation
about
time.
Stepping
and
defining
waveforms
yeah
member.
Somebody
mentioned
that
this
was
a
troublesome
thing
for
certain
people,
so
I'm
trying
to
finalize
those
and
one
quick.
A
B
B
A
This
one,
though
I
think,
is
important.
I've
seen
a
number
of
folks
trip
up
here.
Is
that
so
and
I
don't
think
it
should
take
up
to
too
much
right
is
that
for
frequency
to
mean
we
need
to
be
worried
about
skin
deaths
for
time
domain?
It's
a
diffusion,
distance
and
usually
rule
of
thumb
is
you
know,
factor
of
two
yeah.
C
A
B
Okay,
that's
that
works
so
yeah
I'm,
basically
just
trying
to
finalize
those
and
then
build
the
website
locally,
and
you
know
just
start
accumulating
some
more
of
those
examples.
I
was
kind
of
hoping
for
suggestions
on
how
I'd
like
to
approach
getting
an
inversion
example
for
frequency-domain,
a
TM
and
time-domain
e/m.
That's
not
going
to
be
computationally
impossible.
A
Yes,
and
we
we
tried
then
played
around
with
this
and
I'd,
be
super
curious
to
hear
what
other
people
think
I
would
in
some
senses
be
tempted
to
do
one
like
so
much
one
that
is
going
to
be
pretty
pretty
quick
and
then
one
3d,
one,
that's
maybe
pretty
minimal,
but
still
gonna
potentially
have
to
chunk
away
a
bit
like
I.
Think
if
you
can
shoot
for
under
ten
minutes,
that's
tolerable
for
for
a
3d
inversion
and
I
mean
keeping
it
like
simple.
A
So
that's
no
block
in
a
half
space
or
a
layered
space
or
whatever,
with
a
couple
of
frequencies,
I
mean
the
there's
one
example
that
we
do
a
loop
loop,
e/m
system.
That's
a
couple
of
frequencies
it'll
be
easier.
Actually,
if
you
do
so
for
the
3d
case,
if
you
want
to
do
like
a
single
large
loop
just
with
multiple
data
inside
that,
because
if
we
just
do
a
single,
a
single
transmitter,
it's
less
salts.
So
that's
a
couple
things
I,
don't
know
if
other
people
soggy,
if
you
have
thoughts
on.
C
That's
that's
what
I
would
do
and
then
try
you
Sox
rematch,
so
that'll
save
some
time
yeah,
so
try
use
frequency
domain,
that's
probably
better.
At
the
at
the
moment,
time
domain
has
been
slow,
yeah.
B
A
C
B
I
mean
we
can
always
just
make
it
clear
that
you
know
this
is
just
showing
you
like
the
steps
and
what
you
would
get
out,
but
don't
don't
be
interpreting
the
model
that
comes
out
of
this.
You
just
use
4000
cells
or
something
something
really
easy.
You
invert
it.
You
get
something
it
wants
to
put
something
kind
of
in
the
right
place,
and
you
just
say
this
is
you
need
to
discretize
more,
but
this
is
how
you
would
do
it.
This
is
a
script
for
doing
it
and.
B
D
C
E
E
C
E
D
D
C
D
D
C
D
Shocked
as
I
just
installed,
my
new
machine
and
I
didn't
do
anything
okay,
so
the
people
are
actually
worth,
but
no,
but
it
does.
The
thing
is
that
you
want
to.
You,
want
to
kinda
install
something
all
right,
so
we'll
compile
this
C++
and
then,
after
this
off,
all
you
need
to
do
is
just
grab
the
files
in
Peggy,
midn,
impactive
and
then
you're
ready
to
roll
okay
under.
E
A
E
E
C
No
it's
so
it's
it's
a
long
story
short!
So
we
when
we
when
I,
develop
the
1d
code
and
somebody
in
USGS
was
interested
using
it
and
there
was
this
slow,
so
I
think
they
he,
but
he
kind
of
translated
some
of
the
by
Python
code
into
Fortran
code,
and
then
there
was
a
made
up
quite
a
bit
of
speed
up.
So
that's
actually
the
core
part
of
the
the
code,
so
yeah
it'll
be
great
to
rewrite
that
in
either
C
or
whatever
language
that
works.
C
C
C
B
C
So
yeah
I
just
wrote
like
a
simple
proposal:
oh
I
went
through
a
3d,
inversions
and
yeah
large
computation
power,
so
I
got
thousand
correct,
like
a
thousand
dollar
credit,
so
I'm
looking
forward
to
play
with
it
so
I
think
they
got
some
free
and
your
200
all
in
clinic
that
you
can
just
get
it,
but
there's
a
lot
of
limitations.
So
you
can
use
like
three
cores
or
something
like
that.
So
I
mean
that
and
it's
not
that
useful,
but
then
we'll
see
I
could
do
with
the
thousand
dollars
I.
C
Got
an
email
from
one
of
Rachel's
students
in
the
University
of
Calgary,
so
they
are
developing
some
of
the
fluid
and
coupled
flow
code,
and
he
seems
interested
in
using
simpang
to
develop
so
I
had
a
quick
chat
with
him
and
explaining
just
like
how
to
set
the
boundary
conditions
so
yeah.
It
might
be
a
like
interesting
Avenue.
C
So
we
kind
of
attacked
you
savage,
some
of
the
jus
fiscal
problem
and
what
falafel
problem,
but
having
a
little
bit
like
different
problems
and
us
impact
yeah
Albion
in
Avenue
and
I
think
that
flutes
problem,
although
there
are
a
lot
of
like
a
good
commercials
or
open
source
code,
might
be
nice
to
have
some
some
in
in
simpang.
Some
are
compatible
with
like
a
most
whoa
and
yes,
that's
sort
of
where
I'm
going
and.
C
A
Feel
free
to
invite
them
to
the
the
meetings
you'd
be
great.
C
E
D
E
D
D
That's
the
idea,
so
so,
basically
we're
kind
of
trying
to
read
them.
This
is
just
a
prototype.
You
know
to
to
be
able
to
to
replace
what
jiff
tools
with
would.
Basically
do
you
just
be
able
to
cut
a
knack
to
the
to
the
analyst
project,
so
I
have
like
frequency
n
times.
I
mean
data
submitted
over
like
some
geology
here
and
I
used.
I
use
synthetic
for
the
Ford
summation,
so
that's
also
pretty
cool.
I.
D
E
C
D
Maybe,
with
what
Devon
is
preparing,
it
will
enlighten
me
on
the
how
to
how
to
right
now,
but
time
domain
is
very
super
slow
way
compared
to
the
frequency
yeah,
it's
very
slow
anyway,
and
then
yeah
look
at
this
man.
Those
are
stitch,
twenty
inversion,
so
just
showing
your
own
profile
and
it's
pretty
impressive.
This
is
a
compact
conductor,
II
and
look
how
well
we're
nailing.
Oh
nice,
even
even
at
depth.
You
know
like
we're
even
like
picking
up
like
there
bit
here
so
and-
and
this
is
fast
right-
this
dish
Wendy
all
those
stations.
D
This
is
like
ten
thousand
stations.
It
takes
like
five
minutes
to
run
so
I.
Think
at
this
point,
I'm
not
even
gonna,
bother
going
to
3d
like
I.
Think
if
you,
if
you're
answering
these,
are
set
well
and
you're,
not
trying
to
push
it
too
hard
for,
like
nearing
their
edges,
the
wendy
code
is
really
good,
pretty
strong
code.
It's
like
you,
gotta,
be
proud
of
that
one.
So.
D
Darking
will
happen
if
you
start
to
review
on
a
fifth
hard,
like
the
the
shoulder
here:
alright,
okay,
otherwise,
like
when
you're
right
on
top
like
this.
Is
this
almost
so
Wendy
almost
a
windy
experiment?
Here,
it's
almost
away
in
the
experiment.
Yes,
it
really
really
depends
on
you
how
much
you're
pushing
you're
watching
your
data
well,.
D
D
C
C
C
D
C
A
C
C
It
yeah,
so
I
think
there's
a
bit
of
benefit
by
doing
so
cuz
like
then,
you
can
hard
code
time
stepping,
and
so
the
user
doesn't
really
need
to
worry
about
that.
So
you
make
sure,
like
you
sphere,
pretty
fine
time
discretization
and
then
it's
actually
fine,
because
you're
using
the
tiling.
Your
simulation
is
like
your
small
problem
like
a
single
problem
is
very
small.
Yes,
so
it's
like
having
many
time
steps
or
mini
factorization
doesn't
really
matter.
Yeah.
A
A
A
Because
that
would
be
really
nice
to
think
through
I
think,
as
we
look
at
like
sort
of
higher
level,
wrappers
or
sort
of
what
the
what
actually
we
want
users
to
be
interacting
with
is,
if,
like
we
can
use
your
ideas
there
to
basically
give
like,
if
you're
doing,
an
airborne,
iam
inversion,
and
you
want
to
use
a
3d
codes
like
this
is
this
is
the
set
up
you
should
be
doing
and
have
it
be
a
bit
more
up?
Can
you
need
it
cuz,
yeah,
right
now,
it
for
sure
is
like
much
easier
to
just.
C
E
Yeah
I
mean
just
in
the
in
the
midst
of
trying
to
get
everything
settled
and
getting
myself
yeah.
Let's
get
up
here
and
make
sure
everything's
going
well
I've
been
working
on
your
pipeline
stuff,
just
mostly
working
with
just
my
own
or
just
Forks,
and
pegged
my
own
repo
and
it's
kind
of
playing
around
with
it
inside
there
just
to
see
it
testing
and
see
stuff.
E
It
seems
like
it
build
thing,
builds
things
a
lot
quicker
and
then
Travis
does
the
tests
run
quicker,
generally
I'm,
just
trying
to
come
up
with
a
decent
way
to
do
a
bunch
of
tests
and
you're
out
there
stages
and
see
if
there's
losing
that's
done
with
that,
most
I
just
been
playing
around
with
it.
I've
got
I
can
generally
reproduce
the
functionality
that
we
have
with
Travis
right
now
on
there.
So
that's
good
and
it.
E
Multiple
systems
is
good
like
for
synthetic.
We
don't
necessarily
need
it.
Basically,
it
with
like
the.
What
would
we
do
with
sim
package
testing
on
multiple
versions,
just
yeah
with
disk
retires,
yeah
we'd
like
to
test
it
on
all
three
systems
that
compiles
properly
yeah,
see
other
than
also
well
before.
All
this
kind
of
stuff
happening,
I
was
chatting
with
Andres
with
insky
about
their
field
camp
I
like
trying
to
push
them
if
their
data
open
that
they
collect
this
year,
but
I
don't
know,
what's
gonna
happen
with
that
now
right
well,.
E
That's
like
at
this
point
like
they
only
have
10
students,
so
they
were
not
gonna
able
to
collect
that
much
data
concerning
moments
prepared
what
they
normally
collected.
A
field
camps
I'm
like
well,
you
know:
maybe
this
year
would
be
a
good
year
to
get
into
that
and
making
it
all
open
source
because
it
technically
should
be
open
and
anybody
can
request
it
if
they
want
it
they're,
just
not
a
good
source.
For
that
to
happen
is.
E
I
mean
I
think
they
could
I
mean,
there's
not.
Obviously
it's
a
I
think
it
would
be
a
little
bit
of
Licensing
thinking
should
I
know
you
got
some
of
the
dated
licenses
before
with
GWB
stuff,
and
it
was
definitely
one
of
those
things.
That's
like
okay.
Well,
you
should,
if
you're,
using
the
dating
site,
that
I
came
from
field
camp
and
know.
E
A
Because
what
we
couldn't
think
about
and
I
don't
know
if
this
is
a
terrible
idea
or
not,
but
it's
with
the
GOC
site.
It's
basically
just
to
have
a
data.
Yes,
I
got
XYZ
and
then
have
some
sort
of
template
for
people
to
put
data
somewhere,
and
we
don't
even
necessarily
have
to
be
the
ones
posting
the
data
like
we
could
ask
them.
If
it's
small,
you
put
it
on
github,
if
it's
large,
we'll
figure
something
else
out.
A
E
A
A
E
Just
in
general,
because
they
you
collect
a
lot
of
different
data
over
the
same
area.
Usually
it's
like
gravity,
magnetics,
frequency-domain,
e/m
surveys,
time-domain
IAM,
surveys,
electrical
resistivity,
stuff,
IP,
sorry,
his
nick,
like
they
usually
clicked
everything
over
the
same
area,
so
it'd
be
kind
of
nice
to
have
it
all
in
one
spot
to
go
play
around
with
it.
E
B
E
E
E
They're
just
another
random
thoughts
that
I
had
I,
not
a
sent
a
message
to
Lindsey
about
this,
but
I
just
want
to
get
other
people's
opinions.
So
what
what
what
they
do?
Encanta
Forge
right
now
for
all
their
feedstocks
and
everything
for
people
to
like
modify
them
is
there
suggested
form
of
this
is
to
fork
whatever
feedstock.
It
is
to
your
own
repository,
make
changes
there
and
then
push
from
your
repository
to
the
main
feedstock.
What
that
does
is
it
only
builds
things
once
right?
E
E
A
One
thing
I
was
one
well
I'll
waste,
because
I
was
chatting
with
Joe
with
this
briefly
is
I'm
wondering
if
we
can
actually
configure
Travis
or
Azure
to
actually
not
do
that.
Dual
builds
or
potentially
only
builds
I
mean
I.
Think
the
simulation
refactor
right
now
is
a
bit
of
a
unique
case,
because
we
like
it's
a
ginormous
pull
request,
whereas
hopefully
down
the
future,
like
pull
request
will
again
be
a
bit
more
bite-sized.
C
C
Not
really
not
really
so
one
example
that
I
had
was
somebody
actually
flike
who
are
not
a
simple
upper
and
they
forked
and
they
made
a
pull
request
and
actually
I
wasn't
sure
how
to
merge
that
in
even
into
our
our
breasts.
So
what
I
had
to
do
is
I
merge
that
into
a
separate
branch
so
I
generate
like
because
we
cannot
merge
that
into
master
directly.
So
what
I
have
to
do
right
generated
like
a
small
French
man
March
that
in
and
then
I
merge
that
into
a
master
like
what.
C
E
C
E
E
I
know,
we've
merged
things
and
two
other
repositories,
master
branches,
and
it
didn't
have
to
go
through
that
whole
roundabout
ways.
That's
one
curious
Ivan,
dug
into
what
the
synthetic
settings
are
per
mastered.
Poles
like
if
it's
some
sort
of
setting
somewhere
that
has
to
be
pulled
in
from
its
own
repository.
A
Yes,
one
other
concern
that
I
have-
and
this
is
more
in
the
workflow
side
of
things
is
I,
know
right
now,
it's
not
uncommon
for
folks
to
actually
like
create
a
branch,
or
you
know,
tell
tell
a
user.
Basically,
we
want
you
to
be
using
the
branch
or,
let
me
just
push
push
the
update
and,
like
I
mean
so
we've
done
that
a
bit
with
the
simulation,
just
as
it's
kind
of
evolved
right.
E
Dependency
like
I,
just
you
know,
random
thoughts
in
my
head.
Just
thinking
about
things,
one
to
Chatham,
oh
I'm,
not
feeling
strongly
either
way
on
anything.
Surely
the
only
thing
is
I
really
is
like,
and
this
maybe
it's
a
thing
and
we
testing
to
certain
branches
or
we
limit
like
we
were
testing
the
branches,
certain
branches,
I'll,
pull
requests
or
something
else.
I
can
look
into
it.
Some
more
that
way
as
well,
but
yeah.
D
E
D
E
B
A
D
A
D
A
Okay,
so
I
pasted
a
link
into
the
notes
there
and
I
can
drop
it
into
slack
or
into
the
chat
here.
It's
probably
right
now,
I've
been
trying
to
just
do
a
bit
of
sketching
around
here.
We've
got
this
Jupiter
meets
the
earth
project
and
one
of
the
things
that
I
pitched
is
actually
doing
inversions
and
when,
in
the
proposal,
pitched
MT,
because
I
knew
that
that
was
an
area
where
we
wanted
to
invest
a
bit
of
effort.
But
now
I
think
we
lucked
out
that
that's
actually
pretty
timely.
A
It
sounds
like
for
DOM
and
Jon
and
Devin's
been
keeping
a
pulse
on
things.
So
I
just
created
this
dog
to
be
like
a
bit
of
brainstorming,
a
bit
of
planning
and
then
down
the
line.
We
can
sort
of.
You
know
divvy
up,
labor
and
figure
out
who
wants
to
focus
on
what
pieces
and
and
and
all
those
sorts
of
things.
But
this
is
basically
what
I'm
roughly
planning
to
pitch
to
the
group
here.
A
So
if
folks
have
other
feedback
or
other
things
that
you
want
to
prioritize
and
in
particular
also,
if
there's
other
things
that
you
would
like
to
see
in
the
connections
with
Jupiter
and
Jupiter
hug
Jupiter
lab
those
are
things
that
are
totally
in
scope
with
this.
With
this
grant
so
I
put
down
a
couple
of
ideas,
there's
one
Joe
Hammond
has
done
some
really
cool
work
with
basically
sitting
up
on
Redis
to
do
data
streaming
and
what
would
be
really
cool
with
this?
A
We
could
in
Pegg
actually
be
writing
out
the
steel
subject
or
writing
even
in
version
results
and
things
like
that
to
this
reddit
server
and
that
actually
lets
you
basically
interrogate
well,
the
computation
is
running
so
I
could
be
basically
watching
gradient
updates,
and
things
like
that,
which
actually
would
be
kind
of
cool,
is
to
be
able
to
interest
by
some
of
the
data
that
would
typically
lose
right.
You
don't
keep.
A
We
don't
keep
all
of
our
Delta,
but
to
actually
sort
of
watch
that,
as
the
inversion
is
running,
would
be
kind
of
fun,
and
so
there's
the
pieces
about
there
to
sort
of
think
about
those
types
of
things.
So,
if
you
have
any
ideas
just
like
as
you're
working
just
like
all,
this
would
be
incredibly
useful
or
wouldn't
it
be
cool.
If
I
could
actually
look
at
this
piece
of
information
throughout
the
inversion.
D
A
D
A
Yes,
so
in
this
case,
I
think
there
would
need
to
be
a
little
instrumentation,
but
hopefully
it's
actually
pretty
light
and
not
something
I
actually
see
how
this
is
functioning
with
with
Joe's
work,
but
I.
Imagine
that
that
would
be
something
that
like
we
could
create
basically
a
directive
for
me,
it's
basically
like
inversion
watcher
or
something
like
that.
That
can
then
just
watch
whatever
variables
you.
You
asked
it
to
yeah.
D
I
think
yeah
well
I
to
me
when,
were
you
know
most
interesting?
Is
that
once
we
get
to
those
big
yam
problem,
so
we
kind
of
want
to
know
you
know
what
are
we
waiting
after
I
know?
Is
it
doing
like
vectorization
as
they're
doing
which
frequencies
it's
stuck
on
that
kind
of
stuff?
You
know
like
a
feedback.
Maybe
dass
can
do
this
for
you,
but
then,
if
you
don't
run
and
ask
we're
not
gonna,
have
it
yeah,
just
something
it'll
be
interesting
to
have
yeah.
A
A
D
A
A
No
I
mean
feel
free
to
add
to
that
later
on,
and
then
we
can
maybe
even
have
a
bit
of
a
focused
empty
meeting
when,
when
things
start
ramping
up
yeah.
D
So
we're
meeting
everything
yet
as
I
told
you
next
Wednesday
here
right
so
well,
now
it's
gonna
be
a
web
conference
but
yeah
and
then
probably
starting
I'm
guessing
the
next
month.
Okay,.