►
From YouTube: SimPEG meeting Feb 27
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).
B
A
A
We
won't
be
strict
on
that
at
all,
and
what
I'll,
try
and
do
later
this
week
is
create
like
a
bit
of
a
template
that
you
can
do
support
and
build
from
so
they'll
be
a
set
up
about
hi
I'll
put
a
Travis
in
there,
and
some
dots
I
knew
contain
what
is
suitable
for
your
publication.
So
if
you
don't
want
to
build
Doc's
like
if
you're
just
you
know
deploying
three
notebooks
and
you
don't
need
dots,
that's
fine!
D
A
F
F
G
B
G
A
Yeah,
yes,
those
things
like
that
and
then
to
you
I
mean
if
you
have
the
environment
description
or
a
docker,
you
can
deploy
it
to
my
binding
as
well,
so
people
can
just
run
in
here
examples.
The
other
thing
I
would
recommend
is
using
fixture
so
because,
like
what
I'm
going
to
do
through
my
case
with
neighboring
point,
look
in
the
paper
to
the
repository
of
everything,
but
then
each
example
will
be
individually.
How
about
so
do
I
on
picture,
and
so
it
just
makes
it
a
little
easier
in
terms
of
like
referencing.
A
A
E
A
So
it's
up
to
you,
we're
not
gonna,
be
strict
on
that
at
all
I.
It
would
be
definitely
easier
for
people
like
to
actually
be
able
to
reproduce
your
stuff.
It's
like
for
me.
The
casing,
stuff
I,
might
not
have
all
of
my
stuff
merged
down.
Friends
time
we
initially
submitted
in
that
case,
I'll
provide
the
branch
information.
D
A
H
H
H
So
that
is
kind
of
happening
now,
and
dads
are
kind
of
just
wanted
to
go
through
that
quickly
kind
of
what
is
in
there
and
that
and
what
is
kind
of
in
the
pipeline's
in
terms
of
like
stuff
that
I've
already
written
but
just
needs
to
be
pushed
and
none
of
the
stuff
that
I
want
to
do.
You
know
I,
guess
the
longer-term
development.
H
H
H
Show
me
the
most
current
is
the
is
the
new
objective
functions.
I
mean
not
that
I
need
to
worry
about
that,
but
I
just
never
bothers
operating
from
masters,
because
I
was
just
running
my
stuff
and
I
was
like
either
it's
like
this
or
it's
like
a
week
off,
just
like
what
is
going
along
here
and
I.
Just
yeah
I
wasn't
committed
to
doing
that
until
just
recently
and
when
I
did
it
was
pretty
easy.
The
other
thing
was
the
I
mean
I,
guess
the
pull
request.
H
Initially
it's
from
when
FDM
and
TM
were
changed
like
a
year
and
a
half
ago
and
the
models
and
the
mappings.
Those
are
things
to
that
came
convinced
to
that.
So
it's
it's
just
kind
of,
because
I
was
doing
my
own
thing
and
I
branched
off
and
I
didn't
bother,
keeping
up
and
I
guess
to
with
like
the
NSA
and
FB
was
off
EMF.
That
was
of
tab.
That
was
master.
H
So
if,
if
something
came
in
it
like
took
a
while
and
then
I
just
yeah
like
I,
wasn't
keeping
track
of
everything
that
was
happening
and
making
sure
and
I
mean
also
just
learning
how
to
work
with
get
up
and
like
doing
all
these
things
so
yeah
it
just.
It
is
one
of
those
things
but
in
terms
of
I
guess.
H
H
A
I
H
H
Yeah
and
then
I
added
some
functions
and
I
mean
the
motivation
for
that
was
to
add
in
some
functions
to
that
actually
work
with
the
data
that
kind
of
give
you
some
sort
of
utility
to
do
things,
for
example,
to
import.
The
di
files
list,
unfortunately,
is
the
Center
for
empty
data
and
take
a
3d
survey
and
break
it
into
one
the
surveys
and
problems
or
want
these
surveys,
I
guess
and
data,
but
being
able
to
carry
the
data
like
the
uncertainty
and
floor
with
it
and
implementing
plotting
for
the
data
on
the
data
class.
H
That
can
plot
the
uncertainty
and
then
I
had
a
viewer
that
worked
with
rarest
kind
of
essentially
just
tables
but
they're
numb
pirates,
but
they're
in
the
table
form.
So
you
can
index
meant
by
by
the
name
by
the
column
name,
and
then
you
can
source
them
or
index
them
with
with
the
index,
so
I
could
sort
out
frequencies
and
receiver
types
and
stuff
like
that,
but
it
was
quite
cumbersome
and
I
had
to
make
some
assumptions
that
I
wasn't
quite
happy
with
so
I
rebound
that
viewer
to
actually
work
with
the
data.
Now.
J
A
As
long
as
you
inherit,
the
frequency
domain
data
you'll
be
able
to
index
your
data
right.
Okay,
oxidizes,
the
the
thing
right
now
that
we
to
the
lack
utilities
on
is
really
greedy
in
data.
So
if
it
needs
a
lot
of
work,
bringing
it
EDI
files,
we
have
some
stuff
to
bring
in
a
handful
of
others,
but
yeah
right
now.
But
that's
probably
one
thing
will
deceive
it.
There's
not
a
lot
of
work
on.
A
H
Yes,
I
mean
I:
I
have
really
rely
on
the
NSA,
yet
data
class
I
I
mean
I.
Think
that's
a
conversation
that
we
need
to
have,
but
I
think
it's
getting
too
big
for
his
own
good
for
the
purpose
that
it's
supposed
to
serve
in
within
sim
pack.
But
but
then
we
need
to
figure
out
how
we're
gonna
then
support
all
those
utilities
how
we
want
to
interact
with
them.
H
I
find,
especially
if
we
are
passing
the
data
as
the
thing
that
we
are
in
working
for
it
should
almost
only
care
about
that,
and
then
we
have
any
math
or
something
that
we
can
slap
on
to
it
when
we
want
to
view
it.
So,
for
example,
if
you
are
paralyzing
that
you
aren't
passing
all
your
plotting,
stuff
and
I
think
actually
something
that
I
wanted
to
bring
up
later
was.
A
H
H
So
I
can
show
you
guys
just
here
quickly
on
the
utils,
so
in
true
simply
passing
my
util
start
just
like
all
over
the
place,
so
I
had
the
EDI
file
of
the
util.
That
is
a
class
that
is
just
the
importer
and
currently
it's
still
relying
on
the
rack
array.
I
just
haven't
Doven
and
and
kind
of
addressed
that.
H
So
it's
probably
one
of
those
things.
I
had
I,
oh
yeah.
It
was
probably
this
I
needed
set
up
like
first
and
reason
and
I
couldn't
have
a
hint
sorry,
it's
it's.
Everything
needs
to
be
kind
of
like
brought
up
to
snuff,
with
a
lot
of
things
that
have
been
happening,
but
it
just
keeps
track
of
things
in
process
properties
and
has
location,
fakir
property.
That
is
the
mat
that
is
selectable,
and
then
it
has
a
list
of
station
fix.
H
So
so
currently,
in
order
for
it
to
be
updatable,
because
it
pops
up
figures
in
these,
you
need
to
run
the
matplotlib.
Ask
you
with
the
QT
back-end.
So
what
happens
in
the
notebook
when
you
run
it,
it
will
just
pop
up
a
window.
I
will
take
over
your
notebook
and
you'll.
Just
have
that
selection.
You
can
also
run
it
in
the
notebook
macro,
flip
notebook
interface,
but
then
I
haven't
figured
out
how
to
get
it
to
update
the
figures,
so
you
would
have
to
then
initiate
the
figure
into
different
blocks
and
I.
H
Don't
know
how
to
get
the
event
loop
to
talk
between
two
blocks
in
the
notebook
I'm,
absolutely
certain
you
can
and
but
then
again.
I
also
think
that,
like
Jupiter
lab
will
do
that
for
you,
so
I
can
see
Jupiter
lap
just
like
you
could
just
pop
up
this
math.
Lastly,
whatever
thing
and
then
this
one-
and
once
you
kick,
they
will
all
update
so
I
kind
of
as
much
as
I
love
math
lip
it
is.
You
are
able
to
do
a
lot
of
interactive
stuff,
but
it's
definitely
not
designed
for
that.
H
So
I've
been
using
bouquet
a
little
bit
and
like
it
is
just
literally
by
default.
It's
interactive
like
you
just
you,
don't
have
to
do
anything
unless
or
except
just
tell
it
like
like
these-
are
together
and
like
they
have
a
function
for
that,
so
so
kind
of
I
guess
the
next
kind
of
with
this
is
to
update
it
into
a
truly
interactive
platform,
but
then
just
to
show
you,
the
Plast
utils
then
are
like
a
much
bigger
thing
and
I
implemented.
H
It
is
a
little
convoluted,
but
so
I
have
what
I
call
two
base
plots.
There
are
kind
of
more
meant
to
be
standalone
things
and
are
updatable,
so
that
is
what
the
viewer
uses
the
personable
here.
It's
40
per
a
platoons
station
plot,
so
it
makes
that
makes
a
very
specific
view
on
the
data
and
has
a
setup
function
to
set
up
the
the
plot,
and
then
it
has
a
draw
function
that
you
can
then
call
to
redraw.
The
plot
do
you're
kind
of
saving
yourself,
the
pre
setting
off
of
the
figure.
H
Yeah,
so
I
did
that
for
the
paper
and
apparently
stivity
face
stationed
and
then
I
have
data
NECN
plot
methods
class
that
the
data
inherits,
which
are
then
kind
of
the
lower-level
plot
apparent
resistivity.
So
it
just
plots
an
apparent
search,
TD
curve
of
a
specific
location
and
a
specific
list
of
components.
So
it's
kind
of
trying
to,
though
this
is
inherited.
H
So
this
you
can
call
directly
from
the
data
and
if
we
look
at
the
apparent
crostini
drawing
function,
it
is
just
then
calling
the
data
and
I
implemented
it
here
this
that
way
that
it,
you
can
give
it
a
list
and
it
will
plot
the
first
object
in
the
list.
The
data
out
there
with
uncertainties,
but
the
all
the
other
ones
are
then
just
overlaid.
So
it's
kind
of
for
comparison.
A
F
H
H
F
E
F
H
H
J
H
H
The
sources
inherit
from
base
sources
stuff,
like
that.
The
problems
currently
are
aligned
very
well.
But
but
again,
there
are
some
things
that
we
decided
to
do
when
we,
when
I
started
with
the
NSE
M
to
do
that
kind
of
made
it
such
that
it
was
better
to
keep
it
completely
separate.
Even
though
it's
sharing
a
lot
and
the
biggest
fact
there
is
that
kind
of
I
require
two
sources,
so
I
can't
kept
track
of
that
internally,
but
now
I
think
it
would
be
a
much
better.
H
It's
a
much
better
implementation
to
implement
it
as
its
own
Shores,
but
it
just
takes
in
two
FDM
sources.
Currently
is
its
own
thing
and
it
kind
of
has
to
do
everything
and
on
its
own,
whereas
if
it's
just
inheriting
directly
from
FDM,
there's
just
much
more
flexibility
saying
with
the
receivers,
they
don't
really
need
to
be
under
cm
specific,
they
should
be
FDM
specific
or
they
should
just
live
in
lefty
admin
be
called
and
that
to
give
losses
to
just
implement
the
sources
as
FDM
instead
of
and
then
you
could
just
yeah.
H
Think
the
overhead
is
it's.
It's
gonna
be
less
than
what
it
is
now,
because
I
think
we
just
understand
the
the
whole
thing
a
lot
better
and
then
I
do
have
a
lot
of
things
that
I've
implemented
on
my
own
brands
that
aren't
necessarily
related
to
the
NECN,
but
I
just
wanted
that
from
my
work,
so
I
implemented,
locking
so
I
can
log
all
of
my
immersion.
H
So
I
can
guess
like
a
very
comprehensive
log
of
what
is
happening,
which
I
find
really
useful
for
debugging
and
can
keep
track
of
what
is
happening
at
any
given
moment.
I've
also
implemented
factorization
storage
for
the
FDM
problem,
so
essentially,
instead
of
factorizing,
your
a
matrix
every
single
time
in
to
store
it
for
its
frequency
and
I
mean
if
you
have
the
memory
to
store
it
in
memory.
There's
a
factor
of
approximately
five
speed-up
of
your
inversion,
which
is
quite
significant.
H
So
I
want
to
kind
of
get
that
in
same
with
the
some
of
the
data
data,
slash
receivers
in
natural
source,
they're
they're,
implemented
in
embarrassed
sim
pack
way,
which
makes
sense
when
you
only
have
one
receiver
per
source
kind
of
or
when
it's
kind
of
few
receivers
with
per
source
or
sorry.
You
have
the
same
receiver
object
in
multiple
locations,
then
it
works
really
well.
But
for
me,
I
have
at
least
eight
different
receivers
of
multiple
locations
that
interlace
because
of
the
impedance.
H
H
So
you
literally
doing
the
exact
same
math
twice
and
just
grabbing
half
of
it,
so
just
having
some
sort
of
utility
that
just
like,
oh
that's,
just
completely
redundant,
just
put
it
put
it
there
and
put
it
there
and
then
I
mean
I
can
think
of
other
ways
to
or
I
have
thought
of
other
ways
and
I'm
sure
we
can
come
up
with
other
ways
to
try
to
even
make
it
faster,
storing
it
like
lazy,
storing
it
but
whatever.
But
it's
just
when
I
look
at
my
time,
so
I'm
just
like
that.
H
That
should
not
be
taking
15
seconds
every
time
because
you
use
that
like
what,
like
six
seven
times
per
per
like
stack
or
not,
not
even
stack
but
like
just
it's
just
it's
painful
I.
Just
never
noticed
that
when
I
was
running
a
when
I
was
not
running
a
logger
and
B
when
I
was
like
factorizing,
seven
or
six
times
a
week,
which
takes
a
long
time
but
just
yeah.
H
It's
kind
of
where
I
see
quite
a
lot
of
this
moving
I
mean
the
code
works
I
think
fairly.
Well,
like
it,
it
solves
the
problem,
but
it's
deftly
not
optimized
for
some
of
those
things,
and
then
there
are
some
new
data
types
and
utils
data,
utils,
buing,
utils
and
stuff
like
that.
That
I
want
to
kind
of
implement
and
then
that's
kind
of
a
more
long-term
goals.
H
H
Until
very
recently,
there
are
people
snuggling,
but
it's
kind
of
hard
to
keep
track
and
and
I
mean
I'm
I'll,
be
the
first
I'm,
not
very
active
in
keeping
track
and
and
stuff,
but
but
I
think
having
some
examples
that
are
kind
of
different
levels.
Do
some
of
the
mundane
tasks
that
you
need
to
do,
I
mean
not
I
mean
it's
a
good
practice
for
everybody
to
have
to
write
their
own
EDI
reader.
G
H
Like
it's,
it
has
its
time
in
place,
but
then,
but
then
you
should
just
be
able
to
jump
onto
something
and
just
kind
of
use
it
and
like
dive
into
those
things
in
the
inversion
that
we
just
haven't.
I
mean
I,
have
wanted
to
investigate,
but
just
haven't
had
to
I
am
building
less
to
initially
to
be
able
to
do
those
things.
H
At
the
end,
I
mean
it
just
lowers
the
overhead
of
having
to
learn
like
all
of
these
things,
and
since
it
is
just
essentially
a
variation
of
the
FDM
problem,
it
should
be
that
there
are
other
things
that
I
would
like
to
talk
and
I.
Think
we
should
kind
of
think
about,
but
those
go
to
the
long-term
is
kind
of
kind
of
the
user
level
that
we
want
to
kind
of,
have
and
kind
of
how
we
want
to
deal
with
things.
H
I
mean
like
the
data
I
mean
I,
like
the
idea
of
the
data
to
be
just
like
very
slick.
It's
very
you
just
have
to
you.
Just
have
your
observed
object
and
it's
just
that
just
with
minimal
information
and
like
uncertainty
and
floors-
and
those
are
in
my
in
my
mind,
required
plotting
is
not,
but
you
will
require
it
at
a
different
level.
So
it's
just
kind
of
figuring
out
how
we
want
to
deal
with
that.
H
And
the
same
with
all
the
input
and
outputs
I
I
mean
it
ultimately
has
to
go
into
the
data,
but
it
I
kind
of
envision
it
as
another
API
like
some
sort
of
a
data,
parser
data
worker
or
whatever.
We
want
to
call
it
that
kind
of
deals
with
that,
and
it
just
has
a
property
that
is
theta.
So
you
just
make
that
property.
H
H
But
there's
kind
of
I
guess
ties
into
I
mean
just
kind
of
division:
I
guess
the
yeah
there's
but
of
course,
strategic
planning
off
its
impact
and
I
mean
and
just
kind
of
facilitating
those,
because
I
think
there
are
a
lot
of
users
that
would
like,
like
that
they
have
a
neat
the
iPhone
and
then
just
want
to
test
that
an
inversion
or
separate
the
iPads.
So
if
you
just
had
a
nice
interface,
they
can
kind
of
look
at
it,
but
then
it
just
once
you
start
doing
that.
H
You
open
up
the
kind
of
the
you
open
up
for
let
PhD
killing
implementation
tasks
that
are
so
much
more
fun.
But
just
don't
get
you
very
far
in
your
research
like
in
terms
of
tying
in
with
like
bouquet
and
like
all
these,
like
awesome
tools
where
it's
just
like
things
just
pop
up
everywhere
and
just
I
guess
just
it's
amazing
to
look
at
some
of
these
things.
H
Yeah
you
just
like
you,
but
but
then
again,
it's
just
I
mean
like
I
learned
that,
like
from
psych
I
and
come
from
the
community
like
that,
people
are
just
so
open
to
people
using
their
stuff.
So
if
you,
just
if
you
add
a
small
latch
for
people
to
latch
on
to
it's
just
a
lot
smaller
jumping
into
to
implement
some
of
the
thanks
for
your
problem
and
you
kind
of
get
the
example
of
like
oh,
it
is
actually
just
that
easy.
H
H
Live
in
simpler,
say,
like
simple,
should
definitely
not
inherit.
Natively
bouquet
I
almost
feel
it
shouldn't
do
that
with
matplotlib
either,
because
I
have
a
lot
of
issues
with
that.
Sometimes
I
don't
know
how
to
work
around
them
now,
but
if
you're
putting
it
on
a
cluster,
for
example,
it
costed
you
a
bit
of
grief,
I
need
to
know
how
to
work
around
it.
It
depends
on
what
version
of
cluster
software
you're
using
but,
for
example,
the
cluster.
H
H
Think
that
is
more
important
at
this
point,
but
how
what
we
do
with
it
later
on
we'll
do
later
on
animated.
But
but
for
me
like
some
of
that
stuff,
I
would
just
like
to
make
an
example:
how
to
set
up
simple
efficiently
on
your
cluster,
and
it
just
is
matha
blip
doc
to
use
anything
except
QT,
I,
think
more
or
less.
A
Thoughts
on
what
is
how
to
develop
dogs
and
his
reanalysis
or
anything
on
the
scripts,
which
is
fine,
then
you
can
actually
warm
up
the
message
so
that
it
like
sort
of
turns
into
a
book
style,
an
example,
and
so
that's
next
week,
potentially
is
actually
and
we'll
have
lives
tested.
One
of
the
things
that
you
and
I
talked
about.
This
may
be
actually
starting
and
it's
sort
of
repository,
but
actually
improving
some
very
high
standards.
A
D
H
What
mates
I
guess
examples
gallery
it's
kind
of
nice,
so
they
have
a
gallery.
It's
similar
to
matplotlib,
where
you
have
kind
of
these
nice
pictures,
you
can
click
on
them
and
they'll
show
you
this
like
really
nice
app
and
you
can
they
get
the
shortcode.
Then
you
can
download
the
app
you
can
run
it
and
play
with
it.
So
that
is
a
really
nice
thing.
Then
they
have
kind
of
another
level
of
their
docks
where
they
kind
of
because
it
is
so
interactive,
it's
all
embedded.
H
B
E
E
A
A
C
E
A
I
B
A
A
Permeable
material,
so
other
I,
don't
think
that
would
affect
you
guys,
it's
really
nice.
So
that's
we
fixed
this,
so
I
hadn't
have
pulled
the
in
addition
to
this
map,
I
also
identical
leaf
ones.
We
wanted
to
test
it
along
on
time
against
all
day
and
make
sure
that
those
very
working
so
now
they're
so.