►
From YouTube: SimPEG meeting May 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).
A
Perfect
well,
can
we
like
to
start
slow
this
week,
I.
B
Can't
I
don't
have
anything
to
do
it
for
this
week.
Okay,.
C
C
E
C
F
F
So
we
have
good
examples
and
tutorials
organized
on
the
on
the
website
now
and
then
I
started
to
put
in
a
little
bit
of
work
towards
that
eeehm
Tim
Peggy
in
1d
and
I'm
kind
of
at
a
point
now,
where
I
have
to
make
some
I
think
important
decisions
and
I
thought.
Maybe
I
could
use
a
bit
of
input
from
the
group,
so
I'm
really
just
thinking
about
the
the
sources,
the
receivers
and
and
the
survey
and
I've
posted
a
link
to
a
basically
a
github
issue.
F
F
So
because
of
this,
the
it
makes
the
simulation
really
fast,
but
it
also
means
that
the
way
it
was
it
was
set
up
right
now
is
all
of
these
properties
that
we
would,
you
know,
define
sources
and
receivers
individually,
like
the
the
dipole
moment
or
the
the
you
know,
radius
of
the
loop.
That
kind
of
stuff
is
all
just
being
defined
in
a
survey
class,
so
there's
no
sources
or
receiver
class.
F
It's
just
all
in
the
survey
and
for
for
this
it
makes
sense,
but
it's
not
really
how
we
organize
things
in
simple,
and
we
also
have
waveforms
that
are
kind
of
in
their
own
dot
pie,
and
so
just
I
guess
we.
The
challenge
is
that
we
don't
want
to
kind
of
have
this
be
defining
each
source
and
receiver
pair
individually,
like
we
might
do
in
in
simple,
because
then
we
would
just
be
looping
over
everything
and
the
code
isn't
as
fast,
but
at
the
same
time
it's
not
really
following
the
class
structure
of
simp
X.
F
So
before
I
went
in
and
started
changing
too
much
stuff.
I
wanted
to
get
some
input
and
I
think
it
is
possible
to
look
through
the
list
of
sources
and
receivers
and
say:
oh
okay,
each
source
receivers
if
they're
using
the
same
source.
The
same
receiver
like
let's,
let's
kind
of
do
all
the
computations
at
at
once,
but
yeah
I
could
use
some
input
on
that.
B
F
B
B
Would
say:
let's
generalize
it,
so
we
can
do
any
possible
cases
and
the
airborne
it's
just
one
specific
case
of
that
I
would
I
would
think
that
would
be
the
next
way
forward.
There
was
no
point
that
at
bringing
it
in,
as
is,
if
it's
not
you
know,
they
can't
do
all
the
things
that
the
other
surveys
are
doing.
Yeah.
F
Do
you
most
people
feel
we
really
want
this
to
follow
the
exact
same
class
structure
that
we
have
with
all
of
the
rest
of
our
parts
of
simpang
and
to
be
able
to
do
all
of
those
things
like
completely
generalizable?
Be
able
to
have
any
combination
of
sources
and
receivers
and
input
and
create
a
survey
and
be
able
to
to
simulate
data?
Is
that
what
we
all
want.
A
I'm
not
actually
entirely
sure
and
I
think
that
this
is
also
something
where
we
should
get
Saudis
input
I
mean
because
I
think
it
is
fine
to
have
things
that
are
fit
for
purpose
and
like
this
is
solving
a
real,
tangible
problem
right
now,
and
so
we
haven't
yet
encountered
the
case
where
somebody
wants
something
different,
I
sort
of
hesitate
to
generalize
before
we
actually
have
a
new
problem
that
we
need
to
solve.
I
mean
I.
Think
it's
great.
A
If
things
are
consistent,
where
they
can
be
simple,
but
I
I
don't
know
if
you
need
to
introduce
extra
complexity
in
terms
of
sources
and
receivers.
If
this
really
is
just
focused
on
airborne
e/m,
and
so
maybe
this
is
something
to
talk
about
with
with
soggy
too,
because
what
we
could
think
about
doing
is
maybe
actually
even
renaming.
This
is.
A
F
Able
to
put
in
any
combination
of
sources
and
receivers
as
an
academic
exercise,
you
know
we're
really
trying
to
reach
schools.
Universities
get
them
using
this
to
simulate
data
and
understand
geophysics.
It's
really
nice.
If
you
can
just
define
a
source
and
say:
oh
I,
want
to
measure
the
x
y&z
component
and
simulate
it
up
and
make
a
basic
plot,
but
just
like
we
did
with
the
the
the
DC
resistivity
1d
right.
F
What
joke
did
too
to
look
and
say:
oh
I'm
gonna
make
this
much
faster
because
we're
reusing
a
bunch
of
the
the
fields
for
these
dipoles
and
it's
just
a
superposition
we're
just
adding
the
effects.
There
could
be
something
like
that
where
it
recognizes
that
oh
I'm
doing
a
standard,
airborne,
e/m
survey
and
so
I'm
not
going
to
loop
over
a
bunch
of
stuff
I'm
going
to
just
do
this
all
simultaneously.
I
think
that
there's
a
way
that
we
could
serve
both
purposes
if
we
we
put
in
some
work.
F
F
H
Audience
out
there,
which
is
not
a
perhaps
as
large
as
the
as
the
airborne
community,
but
certainly
a
number
of
ground-based
surveys
that
are
that
are
practical
and
these
are
being
carried
out
and
they'll
be
carried
out
with
a
location
of
a
transmitter
and
then
multiple
receivers
that
are
offset
from
that,
and
sometimes
both
the
transmitter
orientation
and
the
receiver
orientations
are
altered
to
try
to
get
different
sensitivities.
So
I
think
there's
an
intermediate
ground
where
there's
still
the
potential
for
the
use
of
this
kind
of
code.
F
Yeah
I've
always
had
trouble
trying
to
use
3d
codes
for
some
of
these
ground-based
large
loop
surveys,
just
because
the
the
sensitivity
you're
putting
that
source
right
on
the
ground
and
those
cells
are
really
likely
to
change.
I
wonder
if
this
be
something
that
might
give
you
a
result,
that's
easier
to
to
get.
H
And
always
mean
one
of
the
you
know
really
good
uses
for
the
1d
algorithms
is
just
for
that
for
that
forward.
Modelling
at
first
you
know
a
first
pass
inversion,
just
to
help
calibrate
what
the
3d
results
might
might
be,
and
just
to
look
at
normalization.
The
data
I
mean
there's
just
quite
a
few
there's
quite
a
few
reasons
for
working
with
a
1d
algorithm
before
you
go
to
3d,
even
though
maybe
your
final
inversion
is
gonna,
be
on
3d
yeah.
D
I
would
be
very
interested
to
play
a
stage
version
as
well,
if
possible,
but
that
would
be
then
for
dipoles
and
for
loops.
It's.
F
We
actually
have
the
kernel
for
loops
as
well,
so
you
would
be
able
to
put
in
the
actual
yeah
you'd
only
be
able
to
go
and
predict
the
response.
At
a
point,
though,
so
there's
no
I
don't
think,
there's
a
receiver
class
that
would
integrate
the
electric
fields
over
the
pathway
of
a
receiver
loop,
but
we
do
have
loop
sources
and
we
do
have
wire
sources.
I
believe
I
think
you
could
make
a
loop
with
an
arbitrary
shape.
F
G
F
F
A
F
D
F
G
G
The
class
that
we've
that
we
kind
of
used
to
everything
with
you
were
in
touch
with
Franklin
about
the
future
of
that
plan
about
that
package
and,
what's
kind
of
what
they're
going
to
be
using
it
for
and
it
doesn't
sound
like
it's
gonna,
be
really
maintained.
Much
anymore,
updated,
it'll,
just
kind
of
like.
If
there's
a
few
pull
requests,
they
might
pull
him
in
and
things
like
that,
but
I,
don't
I'm,
not
sure
if
it'll
be
well
supported
moving
forward.
G
My
dad
take
yeah,
basically
I
know.
We
were
talking
about
last
week
that
we
about
3.5,
3.6
and
higher.
Had
this
thing
you
can
annotate
things
with
types
three.
So
to
give
users
hints
at
thing
expected
inputs
to
functions,
and
we
return
to
our
user
function,
and
it's
part
of
this
part
well
hide
antics,
takes
that
another
step
and
actually
will
enforce
and
coerce
variables
into
those
expected
types.
G
So
I
and
it's
it's
something
that
is.
It
looks
like
it's
very
wide
like
there's
a
lot
of
use.
Other
people
using
as
well
outside
of
just
this
and
I
can't
force
a
lot
more
things.
That
properties
cannot
tell
you
can't
force
ality
calls
to
functions
of
things
about
this
class
value.
So
if
we're
expecting
a
function
to
have
a
certain
say
it
on
the
function
like
we
can
define
it
as
part
of
a
function
instead
of
like
validating
what
is
inside
of
it.
G
That's
something
if
you
guys
want
to
play
around
with
pedantic
I
would
encourage
you
to
do
that.
The
only
thing
is
that
numpy
arrays
are
not
explicitly
supported,
so
I
had
to
come
up
with
a
simple
class
that
will
expect
a
numpy
array
which
is
pretty
important
to
us
right.
We
need
things
to
be
able
work
with
nope
I
said
not.
B
G
G
It
does
a
little
bit
more
than
getters
and
setters,
and
it's
like
one
of
those
things.
That's
it's
essentially
just
automating
a
bunch
of
the
getters
and
setters,
which
we
don't
necessarily
need
to
rewrite.
Every
single
time.
I
mean
those
are
nice,
but
it
also
provides
automated
serialization
to
dictionaries
to
Jason's,
which
is
something
that
we
won't.
C
G
G
Saw
SOG
you
just
posted
another
pull
request
on
that
or
two
simulation
to
fix
a
bug
in
the
IPCA,
with
IP
for
a
parent
toward
ability
on
the
3-d
case,
I'm,
not
sure
if
it
needs
to
come
in
just
yet,
but
we
can
get
that
in
later
as
well.
Let's
I'd
like
to
focus
on
getting
a
simulation
thing
in
then
we
can
accept
you
more
bugs
if
they
pop
up
do.
A
D
A
A
E
A
A
G
A
E
F
F
Know
if
anyone
knows
knows
that
right
now,
we've
got
the
the
3d
MT
branch
and
then
there's
an
empty
kind
of
like
a
finished
MT
1d
branch.
That
is
that's
kind
of
off
of
that
one.
And
so
if
I
was
hoping
to
kind
of
bring
those
things
together
and
bring
those
in,
but
not
really
sure
what
the
status
of
Mt
3d
is.
C
C
F
Didn't
I
guess:
yeah
I'll,
just
I'll
have
to
go
and
look
back
through
the
commit
history,
because
last
I
remembered
we
were
trying
to
figure
out
how
to
do
basically
implement
the
the
boundary
conditions
and
use
the
same
approach
that
the
UBC
Fortran
codes
use.
But
I
don't
know
if
that
ever
got
validated.
G
I'm
not
sure
I
would
have
to
expect
that,
if
anything
might
be
worth
it
just
to
switch,
the
branch
I
could
try
to
merge
it
in
the
master.
Now,
instead,
if
you're
comfortable
with
the
em1
deep
or
the
1d
side
of
it,
although
I
know
at
one
point,
we
did
talk
about
getting
it
as
part
of
geo.
Anna
like
at
least
the
machinery
is
part
of
geo
Anna
instead
of
him
and
then
just
using
that
in
yeah
in
the
1d
code,
I.
F
C
C
A
So
I've
got
a
couple
updates.
First
I
want
to
thank
folks
for
giving
some
feedback
and
input
on
the
on
the
blog.
If
you
have
any
last
words,
you
can
get
them
in
today
and
then
we'll
get
it
folks.
They
probably
by
the
end
of
the
day
or
so
I
know,
started
discourse
post
and
just
kind
of
send
that
around
and
I
guess.
A
So
a
big
thanks
to
soggy
who's
already
come
up
with
and
run
sort
of
the
2d
DC
and
IP
inversions,
but
I
started
sketching
together
in
notebook
and
I
would
actually
appreciate
just
thoughts
on
sort
of
my
general
flow
I'll
show
you
so
I
dropped
the
the
github
link
just
into
into
the
notes.
I'll
just
show
you
that
right
now,
I
don't
have
my
environment
set
up
to
actually
run
this
okay.
C
A
So
my
basic
thinking
is
I
really
want
to
go
from
sort
of
square.
One
with
this
and
I
know:
Devon
Sun
Watts
in
the
tutorial,
so
I'm
gonna
be
pulling
lots
of
your
text
in
here.
But
basically,
I
was
thinking
is
let's
just
start
directly
from
the
data
file.
So
here's
the
data
file,
we're
gonna
work
with.
A
Let's
actually
just
look
at
what
that
data
file
is,
and
we
actually
run
the
notebook
and,
of
course
like
shortens
this,
but
it's
still
hopeful
I
think
to
actually
get
a
sense
of
like
this
is
what
we're
starting
with
and
then
go
in
and
actually
I
know.
We've
got
these
utilities,
but
I
actually
thought
for
a
tutorial.
A
It
is
kind
of
a
good
framing
for
this,
so
what
this
code
sort
of
or
what
this
readin
does
is
very
similar
actually
to
what
the
utility
does
I
removed
a
few.
If
statements
and
things
like
that,
that
worked
for
more
general
functions-
and
this
is
this-
will
read
in
data
from
century-
and
that's
probably
it
I'm
and
that's
fine
I,
think.
A
Do
that
I
need
to
actually
check
that
they
would
actually
read
in
this
file
these
this
data
file
is,
if
it
it's
the
older
UVC
version
and
I,
don't
actually
know
if
we
have
encountered
that
before
in
any
of
the
other
examples
that
we've
done
yeah
so
I
don't
know.
Do
you
folks
think
that
that
is
like
a
reasonable
thing
to
do,
or
would
you
rather
just
see
like
we've
got
a
utility
to
do
this?
Let's
focus
on
simple
I.
F
A
I,
also
build
good.
Oh
sorry,.
E
I
was
gonna,
I
was
gonna,
say
like
when
Sarah
was
at
exhibiting.
Events
were
from
Belgium
when
she
was
at
CBC
like,
even
though
we
had
new
teams
to
input
data.
It
took
her
a
long
time
to
get
the
data
in
and
being
evil
twin.
So
I
think
this
is
great
with
what
you
didn't
and
that
we
didn't
leave
I
agree
with
Devon
like
we
need.
We
need
that
Instituto.
You
and
I
would
even
suggest
a
contact,
server
and
poker
like
each.
She
can
give
you
some
input
on
that.
She.
F
C
A
Just
the
geometry
is
a
bit
unusual,
so
I
actually
went
in
and
wrote
up
a
little
function
to
plot
a
widget
that
actually
shows
us
how
this
one
for
this
specific
survey
was
books,
which
was
constructive
because
at
first
I
actually
wondered
if
there
was
a
bug
in
the
way
that
we
were
calculating
all
the
mid
points
and
stuff
there's
no
bugs
it
was
just
a
different
geometry,
but
then
go
through
now.
A
The
four
simulations
and
inversions
so
I'll
take
a
look
at
what
saw
he
did
there
and
bring
in
some
of
his
work
in
that
sense.
So
that'll
take
us
through
the
2d
inversion.
Stop
sharing
my
screen.
We
have
three
hours
for
this,
so
it's
a
fairly
good
time,
but
we
do
need
to
have
a
break
because
folks,
well
I
certainly
wouldn't
want
to
sit
in
front
of
my
computer
screen.
First
listening
to
somebody
drone
on
so
we'd
have
a
bit
of
a
break.
So
we
have
the
century
data.
A
G
A
A
That
would
be
great
and
I'll
just
sort
of
plug
away
on
some
of
those
things,
and
hopefully
we
can
actually
ideally
get
this
merged
like
before
before
the
tutorial,
so
that
we
can
actually
get
it
deployed
on
kondeh,
forge
so
that
if
folks
want
to
be
downloading
this
and
running
it
locally,
any
issues
so
I'll
try
and
keep
that
pull
request.
Pretty
tightly
scoped.
A
A
So
we'll
get
that
in
there,
but
right
now,
the
way
that
it's
set
up
is
I've
actually
for
developers
you
can
actually
install
what's
called
a
pre-commit
hook,
so
what
we
can
do
is
actually
those
wimpy
walks
through
this
I
was
thinking
of
having
a
demo
but
yeah.
C
C
A
So
what
we,
what
you
can
do
is
we've
installed
a
pre-commit
hook.
So
actually,
if
you
see,
let
me
see
if
files
at
all
yeah
so
there's
this
file
here
called
pre-commit,
config
camel
and
all
its
gonna
do.
Is
it's
going
to
run
black
using
python
3.7?
A
So
if
you
have
python
3.7
installed
what
you
can
do
inside
of
the
simpang
inside
of
the
base
like
inside
of
the
root
directory
for
simple,
if
you
do
kondeh
install
pre-commit
and
black,
so
those
are
the
two
dependencies
that
you
need
and
then,
if
you
do
pre-commit
install
what
it'll
do
is
before
you
actually
run
a
before.
You
actually
commit
code
it'll
check.
If
black
his
run,
it's
not
it'll
run
black
and
until
you
commit
again
so
actually
let
me
see
if
we
can
do.
G
C
A
Okay,
so
let's
just
go:
do
something
like
this
and
I
got
a
long
line
and
I've
saved
that,
and
so
what
I'm
gonna
do
is
if
I
do
sorry,
that's
a
bit
bigger
now,
can
you
guys
see
that
okay,
good.
B
C
A
G
A
A
A
So
that's
something
to
look
into,
but
what's
nice
is
is
then
you
don't
have
to
manually
run
black
and
you
don't
have
to
actually
remember
to
sort
of
do
the
formatting
is
that
this
will
just
take
care
of
it,
for
you
so
give
it
a
try
and
if
you
run
into
any
issues,
feel
free
to
ping
Joe
and
I,
but
hopefully
this
will
actually
just
kind
of
streamline
the
the
formatting
side
of
things
so
that
we
don't
have
to
aim
folks
to
say:
can
you
please
run
black
on
your
PR
before
we.
A
So
what
did
you
do
and
they
so?
It
should
have
actually
split
up
this
string,
but
it
does
things
like
using
double
quotes.
Instead
of
single
quotes,
it's
opinionated
on
where
brackets
are
so
here.
The
open
and
closed
brackets
are
on
the
same,
like
the
bracket
closes
on
the
same
the
line
started,
it
should
make
sure
that
line
like
comments
or
code
doesn't
go
too
far
off
to
the
right.
So
there's
a
line
length
of
88
is
what
black
uses
by
default.
So
it's
basically
just
like,
and.
H
C
A
G
F
B
B
G
G
F
G
F
G
F
Yeah
I've
been
using
sublime
text,
but
I
tried
to
get
it
set
up
and
it
really
wasn't
as
easy
as
they've
done
some
stuff
in
sublime
on
how
you
would
set
up
a
linter.
That's
that's
completely
different
from
how
he
figured
it
out
before.
I
was
just
wondering
if
somebody
played
around
with
it,
but
this
this
seems
fine.