►
From YouTube: SimPEG meeting Aug 15
Description
updates from SciPy 2017 and the DISC presentations in Australia
A
A
C
D
C
C
D
D
D
A
In
fact,
the
last
location
kokuboro
who
is
working
on
a
1d
MT
version-
and
she
gave
a
talk
at
this
love.
Then
there
were
a
lot
of
questions
around
how
she
was
doing
the
inversion.
So
we
need
another
model
like
where,
after
her
talk
and
I
got
that
set
up-
and
she
has
since
like
emailed
us
back
and
it's
getting
version
results
alright.
So
that's
three
and
there's
also
just
in
general,
a
lot
of
interest
in
public
there's.
A
lot
of
people.
D
A
A
F
D
C
C
I'm
sort
of
positive
stop
this
coming.
What
one
thing
is
clear:
there's
there's
a
real
opportunity:
what's
the
disks
finishes
to
continue
this
kind
of
material
I
think
extend
it
because
what
you
know
right
now,
we're
you
know
like
I'm,
the
last
two
trips
over
Saudi
and
Lindsay
have
just
just
kind
of
nicely
sort
of
God.
To
just
give
me
no
brief
overview
and
then
kind
of
which
is
good.
I
mean
it.
C
C
A
C
A
A
D
B
C
B
C
D
A
A
A
So
we
think
about
the
other
good
thing
that
I
wanted
to
bring
up
was
the
Google
Form
yeah,
so
that
I
would
really
appreciate
if
everybody
can
sort
of
direct
questions
there.
So
if
you
get
an
email
from
somebody
directly
to
you,
if
you
can
say
please,
we
post
this
on
the
Google
form
and
then
click
the
reason
for
doing
that
and
I
think
most
people
get
it.
Is
that
like
this
then
becomes
supplementary
documentation
yeah,
because
we're.
E
A
A
A
There's
no
and
we'll
create
links
from
the
website,
but
there
are,
though
there
are
questions
now
that
people
are
asking
about
senpai
and
you
can
post
it
on
this
Google
forum
and
a
conversation
about
it,
and
then
this
is
searchable.
So
if
somebody's
looking
for
some
hey
I
look
for
magnetic
stuff
it
up
it
up.
You'll
hit
this
okay.
So,
for
example,
let's
see
there's
26
views
on
how
do
you
change
the
source
type
in
the
frequency
domain?
Iam
lateral?
So
some.
D
C
A
C
A
Okay,
so-
and
this
comes
in
them
in
your
email,
just
as
like
page
and
you
can
configure
your
settings
so
that
if
you
only
want
to
see
it
once
a
week
or
something
like
that,
I
don't
really
appreciate.
If
everybody
here
would
be
willing
to
keep
it
up,
every
email
that
comes
in
you
see,
because
then
that
helps
distribute
the
workload
for
answering
these
things
today.
A
A
F
Good
okay,
yeah
yeah,
so
we
good
for
DC
yeah,
that's
police,
just
to
test
the
essence
yeah
to
get
some
decent
results
and
we're
talking
yesterday
about
the
going
towards
the
title
inversion
right
then
so
we
have
conformed.
Jay
I,
just
sent
it
by
so
much
more.
The
tablet
health
strategy,
yeah
cuz,
what's
Torah,
it's
just
an
interpolation
after
yeah,
so
we
have
it
for
our
potential
fields
and
now
VC
yeah
you're
interesting
to
see.
If
we
can
no,
it's
not
master
son
to
do
that.
Also
yeah
dismisses.
B
Data,
yes,
we
generate
the
ten
mesh
and
they're
like
yeah,
we're
good,
like
a
beastly
like
pad
pretty
fast.
Okay,
like
strep
reaction.
Ever
so
we
could
Pat
three
fast.
So
each
match
could
be
relatively
small
compared
to
fools.
Reading
that,
and
so
each
problem
we're
going
to
get
a
griever
chittering
210
problem,
10
DC
problems,
for
instance.
Then,
if
I
put
that
into
a
combo
object,
you
function
at
it
as
a
day
domestic
can
can
it
handle
like.
But
my
model
is
one
like
that
and
then
I
have
a
mapping
that
I
can't
mess.
B
B
Should
be
fine
right
because
it's
like
at
one
point:
we
were
thinking
about
the
making
a
global
problem,
but
at
this
point
on
you,
don't
really
need
that
right
so
that
the
combo
projected
function
is
enough
to
handle
that
yeah
yeah.
So
that's
actually
yeah
and
that's
the
premise
simple,
that
we've
got
all
the
pieces.
I
guess
then
yeah
combo.
A
So
what
we
did
is
previously
instantaneous
baked
in
that
you
had
one
day
to
rest
fit
in
one
regularization.
Just
with
waves
got
it,
and
so
what
the
combo
ejected
function
does
is
like
art
and
says
you
can
make
up
your
objective
function,
how
everyone?
And
so,
if
you
have
multiple
regularization
when
I'm
just
back
them
up,
you
can
basically
just
write
it
down
and
then
say:
my
objective
function
is
now
Phi
1,
plus
Phi
2,
and
you
can
speak
back
in
code
and
then
under
the
pig
nose.
A
A
A
C
D
A
G
A
C
F
A
G
B
Are
a
few
things
like?
It
should
be
a
option
where,
like
that,
for
instance,
when,
when
Jace
for
when
Jace
form,
then
you
don't
want
to
reach
it
like
you,
don't
want
to
redo
the
gene
back
in
JQ
bagging,
for
instance,
so
you
need
some
sort
of
like
a
switch
where
it
triggers
in
like
once,
when
you're
generating
it
I.
B
B
C
Me
following
at
state
lion
one
set
up
so
that
you
can
also
just
occasion
make
just
get
out.
You
know
a
full
Jade
that
could
not
necessarily
the
whole
earth,
sir
full
Jade,
but
not
necessarily
with
all
the
data,
but
just
you
know
sounds
specific
just
because
you
want
to
test
this.
You
want
to
kind
of
get
a
sense
of
what
this
is
looking
like
for
that
and.
B
A
B
A
B
C
A
A
D
A
B
A
E
E
B
B
D
A
A
A
F
D
D
So
you
just
say:
I
want
to
make
sure
I
have
ten
when
I
find
it
sells
everything
except
even
per
direction.
Okay,
so
you
just
say
like
I
want
ten
fine
cells
around
each
one
unlock
each
other's
does
that
you
can
set
different,
so
25m
cells.
Then
I
want
six
of
my
next
small
cells
and
then
you
can
kind
of
control.
How
quickly
you
of
course
too,
because
sometimes
I
could
make
a
big
difference
so.
C
What's
what
would
be
the
procedure
for
actually
taking
look
at
code
like
that
and
then
just
sort
of
okay
there's
enough
to
not
transfer
it
over
to
to
to
Python
or
somehow
integrating
it
or
I
mean
weird?
If
weird
is
one
so
you'd
like
to
have
like
have
a
good
octree
machine
right
and
so
we've
got
something
and
then
there's
about
a
dozen
other
packages
out
there,
but
one
here,
that's
actually
seems
to
be
satisfactory
for
a
lot
of
problems
yep.
C
A
I
mean
I
think
the
first
thing
is
just
finding
the
smallest
piece
of
functionality
that
would
want
to
bring
in,
like
don't
think
of
it
as
too
tired
mesh
generating
package.
Think
of
like
each
function
has
its
own
thing
yeah
and
then
go
after
like
what
are
the
ones
that
are
most
useful
or
valuable
to
people.
Listen,
I,
don't
think
like
it's,
not
a
code
translation,
because
that
we
doctors
are
thought
over
different.
B
D
D
B
D
A
B
I
think
we
probably
want
to
have
a
similar
functionality
with
we
have
four
tensor
insurance,
doing
a
very
simple
algorithm
pad
like
this.
That
won't
have
a
core
like
corn
mash,
as
it
was
just
basically
a
tester
match
up
everyone's
path.
That's
like
when
I
walk
out
really
fancy
color
to
it,
refine
things,
but
then
hungry.
C
You
know,
there's
there's
basic
system:
you've
got
typography
right,
see
another
fine,
fine,
mentioning
about
typography.
You've
got
transmitter
or
a
receiver,
so
you
need
to
be
good
there
and
then
you
may
have
some.
You
know
here:
extreme
conductivity
contrast
or
something
where
you're
gonna
have
breath.
That's
some
pollination.
So
many!
Those
are
those
are
kind
of
look
base
base
elements.
Okay,
we
have
to
have
some
flying
discretization,
yeah,
of
course,
wave
them
yeah.
C
A
A
C
F
A
C
F
A
More
so
what
you
actually
know.
F
So
long
term
and
that
Canada
I
can
I
brought
something
like,
like
all
our
I,
those
are
kind
of
dispersed
everywhere
on
the
man.
You
know
so
we're
developing
that
CDC
readin
right
out.
If
you
see
because
we
needed
at
some
point
yeah,
it
makes
sense
eventually,
if
you
just
collided
all
into
one
one
little
synthetic
type
of
class
yeah
right
because
before
I'm
just
looking
for
right
now,
okay,
neither
little
Dylan-
and
you
know.
F
A
Assistance
yeah,
so
if
we
actually
could
sort
of
take
stock
of
where
things
are
I,
don't
think
we
have
too
much
duplication
discretized.
How
trying
to
call
a
few
things.
But
there
is
this:
where
you're
going
to
kiss
and
completely
undocumented
functions,
so
people
can't
find
them
yeah.
So
even
if
going
to
you're
gonna,
just
these
first
office
together
doc
string,
so
it's
a
kids
soccer
what
we've
got
or
Tiffany
good!
A
So
let
me
give
a
piece
in
matok
and
I'm
doing
something
dis,
Britt
NPM,
like
we
want
to
try
to
push
all
that
stuff
forward.
That's
a
bit
hard
to
do
just
in
terms
of
littler
quest
reviews,
and
it
takes
a
lot
of
time
right
so
down
the
road.
We
might
want
to
think
about
what
sort
of
packaging
each
of
the
minutes
early,
but.
A
A
C
F
Entity
that
that's
all
of
these-
yes,
for
instance,
like
that
static,
has
importers
or
for
observation,
these
observations,
2d
or
3d.
So
we
are
we've
already
written
those
already
em,
but
they
were
buried
under
you
need
to
create
that
survey
and
you
certainly
get
redone
or
whatever
well.
D
E
A
Like
it's
quite
sort
of
I
think
with
your
specs,
but
is
not
necessarily
that
we
want
like
a
separate
utils
package,
but
it
is
nice
to
have
things
that
where
these,
where
belong
in
the
sense
of
like
a
period,
if
you're
bringing
in
you
CDC
buggy,
surveys
like
that
should
happen
in
tcpip,
not
in
some.
Oh
okay,.
A
Think
reinsert
me
separate
inside
outside,
but
I
think
that
the
application
needs
to
be
close
to
yeah
unless
something
that
we
can
use
like
properties,
and
it
would
be
we're
just
having
a
look
at
the
open
mining
format
stuff
that
Roman
and
Franklin
have
done,
because
that
is
more
sort,
defining
a
rigorous
structure,
around
geologic
structures,
so
lines
or
holes,
services
and
being
rigorous.
Look
how
to
work
with
those
type
of
data.
And
so
they
did
a
really
good
job
at
organizing
that
and
then
basically
beyond
the
rate
parsers
that
can
bring
in
like
cheers.
A
Oh
yeah,
that's
a
good
point.
We
need
to
get
this
which
has
not
come
for
it.
So
the
people
who
knows.