►
From YouTube: SimPEG meeting May 22
Description
Initial discussion on RFG presentations
A
A
A
Giving
a
link
to
someone
like
you
have
like
your
like
it's
way
bigger
than
you
go
like
you
really
assume
that
sighs
that
could
blablacar
and
such
so
you
have
actually
people
I'm
a
driver
and
I.
Have
this
rating
and
I'm
doing
this
I'm
doing
this
things
between
between
this
and
this
and
you
can.
We
just
got
the
passenger
and
try
to
find
she
find
somebody.
D
C
A
B
E
D
Gonna
care
much
more
about
like
what
it
can
be.
Unification
I
mean
giving
a
bit
of
an
overview,
is
good,
I,
think
but
yeah
I
think
the
application
and
then
sort
of
just
like
another
proof
of
the
framework,
so
in
I
think
it
was
Saudis.
Example
in
there
there's
like
lots
of
the
currents
in
DC
so
showing
actually
how,
like
the
data,
are
built
up
and
things
like
that.
People.
D
F
F
D
D
F
Yes,
I,
like
at
the
Sochi
talk
that
I
gave
in
couple
months
ago.
There
are
some
probably
there
are
some
example
that
you
can
reuse,
I,
think
I
simply
used
a
DC
or
Malling
to
show
the
currents
and
then
a
couple
of
inversion
reserved
I
used
the
l0
norm.
She
showed
you,
like
smoothing
various
parts
and
versions
of
stuff
I.
Think
there,
probably
some
stuff
that
you
can
reuse,
would.
D
C
D
A
A
How
I
think
about
the
problem
and
see
how
we
getting
at
each
step
and
ever
when
we
were
with
you,
it's
additional
information,
yeah
excellent,
also
not
too
much
wrong
either.
The
code
as
a
map
is
really
thank.
You,
okay!
Well,
that's
something
we're
able
to
do
if
you
provide
us
with
when
we,
sir,
like
here.
A
If
you
take
you
to
people
that
how
I
just
want
me
I
just
want
to
add
on
that
already
when
the
protests,
it's
like,
having
borehole
information
to
ends
up
with
even
constrain
more
access
to
shape
order
of
the
order
of
the
very
magnetic
body
at
that.
So
that's
one
wonders
estimation
estimate
more
will
be
actually
competed
framework
like
information,
you
can
add,
also
Jim
over
to
Georgie
life
information
yeah.
A
A
D
C
D
B
G
F
D
F
F
D
E
D
It
is
gonna,
take
yeah,
but
just
like
sit
down
and
pull
it
all
together
and
then
for
our
presentations.
I
mean
mine,
I
think
will
be
pretty
straightforward.
It's
like
a
sympathy,
talk
plus
looking
in
case
Airport
I
know
Ola
Airport,
so
we
did
a
1d
inversion.
Well,
we
did
a
plate.
I'm
gonna
do
some
forward
modeling,
showing
currents
and
stuff
your
time,
I'm
hoping.
Actually
we
should
follow
up
on.
This.
D
F
D
Email
and
see
if
we
can,
because,
if
not
like,
you,
can
always
sort
of
do
gifs
and
stuff
like
that,
but
I
feel
like
it's
much
more
powerful
to
actually
use
the
real
thing:
yeah,
great
yeah,
cool
and
then
free.
Your
talk,
soggy
was
there.
Where
are
the
things
you
want
to
chat
about?
There?
Do
you
feel,
like
you've,
got
a
decent
idea
of
what
needs
to
be
done
or
we're
stopped,
because
you
have
an
IP
one
right,
no.
F
F
C
F
D
You
one
thing
we
should
think
about:
I
think,
especially
for
Doug's:
abstract
is
they're
gonna
do
like
full
conference
proceedings,
so
we
can
publish
a
full
paper
with
it
and
so,
like
you,
just
take
your
abstract
and
expand
it
a
bit
right,
I
think
for
sure
for
Doug's
one.
We
should
do
that
and
then
we
can
think
about
the
other.
Two
I
think
they
were
saying
it
would
be
due
in
September
or
something
like
that.
So
it's
not
yes,.
F
F
Helped
a
lot
so
what
happened
so
I
was
using
some
of
the.
So
the
main
thing
for
1d
code
is
a
transformation
either
from
like
a
spatial
frequency
to
space
or
frequency
to
time
so
I
think
I've
used
in
the
beginning.
I
was
using
some
of
the
scripts
from
V
and
probably
some
UBC
alumni
and
yeah.
There
wasn't
like
it
was
accurate
but
like
it
wasn't
designed
very
like
to
be
optimized
in
terms
of
so
I
started
using
Peters
code,
some
part,
but
not
like
in
a
very
good
wait.
F
So
Peter
took
a
look
and
suggested
quite
a
few
things
where,
like
we
could
improve
this
B.
So,
yes,
he
actually
changed
the
Empire
mode,
and
then
there
was
a
couple
of
iterations
between
him
and
myself
because
he
had
to
change
his
coats
like
some
of
his
code
to
handle
what,
since
impact
M
on
me,
though
yeah
we
just
record,
we
went
back
and
forth.
I've
actually
got
the
pinata,
my
scone,
so
it's
pretty
fast.
It's
faster.
F
I
really
yeah,
it's
much
faster
I
wasn't
do
that
before,
but
now
actually
yeah
I
need
to
change
the
code
quite
a
bit
where
they
think
knowledge
I,
see
more
optimized,
okay,
yeah
yeah,
pull
latest
French
and
running
right,
I'm,
not
sure
how?
How
much
is
that?
How
much
faster
on
your
end
but
I
think
at
least
for
my
testing
is
actually
the
best
and
they
I
put
the
interpolation
once
so
at
the
end
of
the
day.
B
F
Like
a
that's
it,
it's
kind
of
weird
the
interpolation,
because
I
get
you
got
a
very
slight
get.
You
got
a
very
sparse
point
in
space
if
you
think
about
like
a
number
of
sounding
location,
yeah
we're
like
airborne
yeah
and
sometimes
it's
not
aligned
with
your
mesh.
So
just
doing
an
interpolation
doesn't
do
a
good
job.
If
you
do
a
simple
linear,
interpolation,
yeah
you're,
not
gonna,
get
the
good
kind
of
not
sure.
What's
good
but
you're
not
gonna
get
the
smooth
conductivity
model.
So
what
what
I
had
to
do?
F
I
just
interpolate
in
2d,
first
using
K,
treat
here
I
could
could
use
like
inverse
weight
or
whatever
that
you
can
make
it
your
model
into
a
space
smooth
then
do
a
one
interpolation
in
3d,
so
first
through
a
2d
in
the
XY
space
and
interpolate
in
zet.
Okay,
that's
what
I'm!
That's
what
I'm
doing
yeah
so
yeah
I
get
that
right,
but
there
probably
is
a
better
way.
I,
don't
even
know
what
what's
the
better
way,
so
I!
F
F
B
B
From
online
there's,
one
one
thing
that's
still
want
to
implement
is
to
be
able
to
see
the
models
as
they
come
out
or
see.
The
result
as
they
come
out,
don't
have
to
wait
till
the
end
of
until
the
end
of
the
run
tonight.
If
we
can
like,
if
either
we
create
like
a
temporary
H
5
files
or
if
both
MATLAB
and
Python
can
can
look
at
the
same
same
file
like
I,
have
a
listener
in
that
blog
or
something
that
just
keeps
pinging
dh5.
B
B
It
looks
like
you
where
you
can
do
sort
of
like
lazy
loading
but
I'm
just
not
sure
if
one's
gonna
start
taking
like
it's
going
to
lock
the
file
while
it's
looking
at
it,
I
might
begin
just
yeah
or
send
send
a
signal
to
one
software
to
the
other
saying
like
I'm
looking
at
it
now
wait
for
it
and
then
get
our
things
that
they
can
alternate
down
death
or
something
much
cleaner
than
I
having
multiple
h5
and
it's
lashing
it
at
the
end.
That's
right,
cumbersome,
yeah,.
B
D
And
then
the
only
other
thing
that
we
were
chatting
about
before
the
we
went
online
year
was
discretized
and
pushing
some
of
the
gene
autry
coding.
So
the
last
thing
I
just
wanted
to
connect
with
Joe
on
is
I.
Think
there's
some
files
in
there.
That
are
you
compile
with
scythe
on,
and
you
should
do
that
locally.
B
F
B
That's
what
I
was
going
to
tell
you
guys
is
that
we
need
to
think
we
have
one,
but
right
now
it's
on
your.
Like
one
shot,
you
know
it's
one
level
and
we
want
to
think
like
the
same
type
of
UVC
as
or
we
can
say,
I
want
like
three
three
of
the
next
level
and
for
the
next
step,
although
I'm
not
so,
we
can,
like
you,
know,
a
little
bit
better
yeah,
but
then
we
need
to
operate
differently.
B
B
B
D
A
wrapper
that
communicates
this
set
of
points
as
a
function,
so
I
think
that's
potentially
a
way
to
go
right
is
like
given
a
set
of
points.
How
do
you
want
to
refine
around
those?
So
that
is
a
function
that
then
are
like
class
or,
however,
we
want
to
structure
that
that
then
communicates
to
the
piece
of
rolling
road
that
just
interpolate
surround
or
like
refines
around
a
function.
Okay,
because
all
you
have
to
do
like
for
that
refinement
piece
to
work
is
like
it
just
I
believe
all
it
does.
This
takes.
D
B
D
Stopping
us
from
turning
that
into
a
function,
it
doesn't
have
to
be
a
mathematical
function,
is
you
know,
I
mean
it
just
has
to
be
like.
We
just
have
to
be
able
to
say,
given
an
XY
location,
should
I
refine,
yes
or
no?
That's
all
that
that
needs
to
do
so.
It
doesn't
ever
be
mathematical.
It
just
has
to
be
like
a
computational
function.
Okay,.
B
Thing,
but
how
do
we
answer
that
this
side?
But
that's
the
issue
is
I?
Okay,
let's
it's
surface!
We
only
wanted
this
to
discretize
the
next
level.
On
there
we
go
on
a
necessarily
to
go
over
and
then
it
points
all
around.
The
line
is
just
along
the
axis
of
the
line
there
they're
all
kind
of
different
cases
right.
D
D
B
D
B
F
B
Something
that
we
should
we
should
push
look
up.
Is
that
thing
you've
infrared
a
function.
This
is
just
gonna
break
to
that
the
finest
level
on
the
function,
but
there
was
nothing
to
tell
it
to
how
fast
it's
gonna
is
gonna,
expend
out
Kennedy
to
feed
another
function,
parallel
with
a
different
Archie
level.
I
think.
D
D
B
G
A
B
B
G
D
Yes,
really
yeah
cool
I'm,
not
here
next
week,
so
we
can
figure
out
what
you
guys
want
to
do.
What
would
be
beneficial
yeah
you're,
either
control
okay,
okay!
Well
then,
maybe
we
can
postpone
it
and
let's
have
something
that
comes
up
yeah
or
like,
if
you
guys
want
to
practice
a
presentation
or
something
like
that,
then
great,
but
it's
not
there
well.