►
From YouTube: SimPEG meeting July 31
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
B
B
D
C
B
C
B
Sounds
good
sorry
I
was
trying
to
go
there,
but
now
I'm
I'm
lost
again
yeah,
so,
okay,
so
the
first
function
now
I'll
go
through
a
notebook
after
because
it
is
going
to
be
the
example
for
the
for
the
actuary
this
guy
here.
But
let's
just
look
at
the
functions
or
maybe,
let's
start
with
a
notebook,
then
yeah,
all
the
notebook
is
doing,
which
will
be
the
example
is
to
create
a
Gaussian
topography.
B
Put
some
points,
some
survey
on
top
of
it
and
then
creating
create
the
mesh,
and
those
are
the
two
functions
that
I'm
proposing
that
I
would
love
that
you
guys
are
try
them
and
I
improve
them
if
possible,
Craig
has
been
working
with
those,
but
there's
a
milky
space
for
improvement.
So
the
first
one
is
to
build
a
mesh
centered
over
cloud
of
points
right,
so
we
often
want
to
just
give
how
far
we
want
to
pad
out.
A
B
Correct
yeah,
okay,
I
see
and
then
currently
what
I've
done
is
that
you
see
this
pad
pad
this
variable.
This
just
tells
you
how
far
you
want
to
go
outside
of
the
of
the
box
of
XYZ,
so
you
send
in
the
X
Y
Z
locations.
They
will
find
you
Center
it
on
that
box
and
it
will
pad
out
in
that
in
that
I
guess
six
other
dimensions
with
this
guy
here
and
dumb.
B
E
B
Then
you
have
two
options:
you
can
Center,
you
can
Center
the
mesh
in
the
middle.
Like
let's
say
you
want
padding
vertical
up,
then
you
just
Center
in
the
middle
or
Center
its
top
or
center
I'll
show
you
the
example
with
Center.
So,
for
instance,
here
I'm
telling
it
like
Center
it
on
topography
and
put
it
in
center
of
the
mesh
and
I
gave
it
some
distances.
B
B
The
second
function
is
to
refined,
because
now
we
change
a
little
bit,
how
we
were
how
we
were
breaking
it
down
out,
Joe
implemented
it,
and
he
gave
us
the
switch
to
finish
the
mesh.
Basically
Joe.
Can
you
give
a
brief
explanation
again?
What
what
they
finalize?
Does
it
basically
constructs
it
right
so.
G
As
its
building
the
mesh
its
dividing
everything,
it
basically
does
it
and
they
show
it
as
its
building
the
mesh
incrementally
it
initializes
nodes
and
cells.
But
after
it's
done,
then
it
will
go
through
and
determine
everything
else,
so
build
all
the
faces,
edges
and
finally
hanging
things
so
you
have
to,
but
it
only
does
that
once
yeah,
you
don't
know
once.
B
B
That's
right,
so
that
means
okay.
For
instance,
here
I
I
want
to
break
down
the
mesh.
I'll,
do
I'll
go
through
the
refine
tree
function
later,
but
I
want
to
discretize
courser
on
the
topography,
because
it's
a
bar
Fu
is
pretty
large
right,
so
I'm
going
on
the
only
going
to
discretize
on
the
second
and
third
actual
level,
and
but
I
don't
want
to
constrain
it
now.
B
So
I'll
hold
down
the
construction
of
the
mash
bike
by
inserting
a
false
statement
and
then
I'm
gonna
call
that
function
again
with
my
survey,
locations
and
African
augusto
gaez
good
eyes.
Finally,
at
the
highest
level
on
this
one,
so
I'm
gonna
ask
for
two
cells
around
my
ABS
locations,
and
then
this
course,
or
course
or
for
the
biography
one.
B
G
B
B
B
Right
now,
it's
points
I'll,
give
you
an
example
later,
but
basically
points
will
create
a
cloud
of
points
around
the
points
that
you're
defining
in
three
in
two
dimensional
space
to
be
able
to
break
down
to
the
highest
level.
Second
and
third,
the
five
here
means
that
I
I'm
asking
for
five
points
at
the
higher
side
level
around
around
the
topography.
Oh
I'm,.
C
B
B
B
Okay,
so
here
the
idea
is
that
what
the
Prague
Rafi
is
just
a
Gaussian-
oh
I,
haven't
put
it
here,
but
it's
a
Gaussian
and
the
idea
is
that
the
often
topography
extends
far
out
and
we
don't
want
a
high
US
discretization
on
typography
like
we
might
just
want
to
have
something
coarse
enough,
so
I'm
giving
zero
for
the
highest
luxury
level.
So
you
see
that
once
we're
out
of
the
observation
points,
maybe
I'll
output,
the
data.
B
B
B
A
B
B
B
A
B
A
A
B
B
So
I'll
I,
just
briefly
prepared
for
health,
so
I'll
I'll
recreate
the
mass
here
and
I'm
just
gonna
insert
one
point:
okay
has
to
be
a
column
for
mouth,
so
I'm,
just
good,
pretty,
creating
a
vector
of
1
by
3
and
I
find
out
that
it's
it's
broken.
I
can
only
put
right
now
ones
of
zeros.
It
doesn't
increment,
so
I'll
need
to
fix
it
to
a
to
loop
over.
But
okay,
let's
start
with
one
point.
A
E
B
C
B
G
C
C
B
E
H
B
B
B
G
H
You
look
at
the
top
left,
one
of
the
sexual
cluster
of
points
than
if
you
go
up
one
and
to
the
side
of
one.
That's
gonna
break
that
down
yeah.
You
go
to
the
bottom
left
point
in
the
center
clustered
points.
It
breaks
the
one
down
to
the
left
and
the
one
to
the
bottom,
but
it's
not
breaking
the
diagonal.
That
makes
sense
no.
C
B
C
C
B
B
B
B
G
So
one
of
my
examples-
I
was
working
with
for
the
for
my
field.
Camp
students
I
still
need
to
get
up
there.
The
idea
is
that
you
just
create
an
interpolation
object
and
then
you
can
pass
that
to
the
mesh
right
as
it
as
a
recline
type
thing
and
then
just
measure
okay,
how
far
away
from
it
has
the
surface
and
then
refined
based
off
of
that
level,
because
you
get
it
like
I
said
you
get
the
cell
center,
so
you
can
figure
out
it's
like
the
height
of
topography
at
that
center
location.
C
The
other
thing
with
that
gone
to,
if
we
go
about
it
function
based
as
Joe,
is
saying
then
it's
easily
like
it
sits.
We
easily
get
the
case
where
you
want
to
refine
a
box
or
refine
around
some
complex
target.
So
even
if
you
had
like
the
tkc
surfaces,
for
example,
we
could
then
just
define
distance
from
like
those
surfaces
as
a
function
and
then
have
that
be
how
the
mesh
is
built.
Okay,.
C
All
he
you
toner
and
okay.
C
E
G
B
E
B
D
B
C
B
G
G
G
C
Correct
so
we
do
we
build
a
2d
mesh
based
on
whatever
slice,
you're
gonna.
Look
at
one
quick
thing
with
the
stream
plots
what
we
actually
do,
cuz
stream
plots.
You
can't
do
on
a
non
uniform
mesh
right
now,
that's
a
limitation
of
that
bullet,
and
so
there's
a
lot
of
sort
of
logic,
I
guess
in
the
stream
plotting,
so
that
the
first
thing
that
we
actually
do
is
interpolate
to
the
finest
cell
refinement
and
then
build
the
stream
plot
from
that
which
works
reasonably
well.
C
What
you
have
to
be
careful
with,
though,
is,
if
you
have
like
very
disparate
scales,
it
can
take
a
really
long
time
to
do
that.
Interpolation
and
so
there's
also
some
parameters
in
there,
so
that
you
can
basically
define
a
bounding
box
and
say
that
I
actually
only
want
streams
from
like
negative
one
hundred
to
one
hundred
extends
out
kilometers
so
that
it
only
interpolates
in
that
region.
G
Basically,
we
go
about
are
the
same
same
sort
of
manner:
I
mean
I,
don't
think
we
need
to
build
a
interpolation
matrix.
You
do
any
other
fancy
other
stuff
for
this,
because
it's
just
mostly
at
one
time
off
plotting
things,
so
you
could
probably
just
get
away
with
using
a
grid
data
function
from
Sify.
B
G
B
B
H
B
B
B
B
This
guy
yeah!
That's
it
from
you!
So
guys
put
your
comments.
I
think
the
function
is
probably
the
best
way
to
go
right
for
the
refinement
mm-hm.
If
you
give
me
an
example,
I
will
I'll
review
up
because
we
probably
wouldn't
have
to
need.
We
didn't
need
all
this
all
this
stuff
to
like
create
clouds
around
it.
If
it's
just
a
function,
I'm
guessing
yeah,
you
don't
yeah.
G
G
F
E
C
Well,
well,
he's
playing
it
up
dawn
that
looks
that
looks
good
I'll,
take
a
look
at
the
PR
and
maybe,
if
Jo,
if
you
have
some
time,
if
you
could-
and
we
can
fear
out
where
things
should
live
because
I
do
think
a
lot
of
what's
been
developed
should
actually
be
in
discretized
and
so
yeah.
We
can
find
a
sensible
place
where,
at
all,
yeah.
G
B
Well,
joy,
Scott
I
was
talking
keyboards
to
go
in
there,
yeah
yeah
right
so
we're
talking
about
because
there's
a
bunch
of
stuff
in
there.
That
is
actually
not
really
related
to
the
tree
mesh
for
potential
fields.
I
was
updating
a
few
things
on
this
side,
so
we
could
probably
just
break
that
guy
and
merge
the
changes
to
your
PF
update,
Pia,
Pia
Pia,
alright,.
I
I
I
B
C
H
H
H
D
G
Okay
kind
of
distilled
about
the
DC
inversion
example
on
their
forum
loaded
in
all
their
data.
The
idea
is
that
it
was
a
winner.
They
were
doing
a
renter
array,
so
I
had
to
do
a
few
little
changes
there
to
allow
for
like
plotting
or
the
pseudo
section,
because
it
doesn't
wasn't
quite
quite
work.
Work
correctly,
basically
just
masquerades
the
winner
array
as
a
dipole-dipole
array,
because
that's
basically
what
it
is,
except
for
how
you
plot
the
pseudo
section
anyway.
So
I've
got
the
locations
of
all
the
electrodes.
G
G
G
G
C
G
Basically,
this
just
accesses
the
see
tree
like
a
sea
of
lamentation
of
the
tree
terminal
max
possible
level.
It's
a
little
bit
better
and
more
robust
to
use
this
if
you're,
building
the
functions,
because
there's
a
lot
of
logic
that
goes
in
determining
what
the
level
is,
especially
when
there's
differing
numbers
of
cells
in
each
dimension.
G
G
Typography
and
then
just
kind
of
a
fast
exit
and
then
because
I
just
used
my
interpolation
function
on
the
X
location
and
figure
out
what
the
top
of
the,
what
the
topography
would
be
at
that
look
at
that
X
location
and
then
compare
the
value
of
the
mesh
or
the
cell
Z
location
to
discography.
And
this
this
was
like
I,
specifically
defined
function
for
this
example.
But
you
could
also
do
things
like
the
cell
widths.
If
we
can
access
that.
G
The
idea
is
that
it
creates
a
Python
like
cell
object,
it's
a
Python
cell
object
from
the
trees
C
so
that
you
can
use
to
access
a
bunch
of
the
information
on
it
as
it's
being
built.
You
get
access
to
the
center
of
the
cells
and
the
width
of
the
cells
and
I
think
you
could
get
access
to
the
nodes
of
the
cell
as
well.
G
G
G
B
B
B
G
B
G
B
G
E
G
B
C
B
Do
you
guys,
like
the
I
personally,
think
more
in
terms
of
like
how
many
octree
cells,
maybe
just
because
we're
used
to
all
UBC's
working,
but
instead
of
a
distance,
it's
more
like
I,
want
to
say
three
cells?
What
do
you
guys
prepare?
Should
we
working
distance
or
number
of
cells
I
mean
you
could
convert
between
the.
G
G
C
Yeah
I
think
started
by
default,
though
working
in
distances
is
a
bit
more.
Natural
I
mean
we're
used
to
working
with
cells,
because
that's
just
the
codes
that
we've
been
exposed
to,
but
I
think
that,
like
I,
don't
know,
I
would
I
would
suspect
that
most
users
working
with
just
distances
and
like
I,
need.
You
know
these
10
meters
to
be
reasonably
well
real
refined
might
be
a
bit
more
intuitive
I,
don't
know
if
you
guys
have
yeah.
D
C
A
G
B
G
G
C
G
Inside
cells,
all
points
which
is
good
like
if
you
just
okay
I,
want
to
make
sure
that
everything's
refined
around
my
observation
points
that
this
is
at
this
level.
On
the
word,
it's
it's.
It's
useful
like
really
good
a
lot
more
powerful,
refined,
I
think
just
designing
a
proper
function
to
pass
to
it.
Yeah.
A
A
G
G
A
A
G
B
E
B
E
E
G
B
A
B
B
G
A
A
But
if
you're
actually
want
to
run
like
a
scalable
I'll
do
with
them
in
code,
we
definitely
need
to
work
on
the
field
object
and
that
yeah
I'm,
not
a
hundred
percent
sure
at
the
moment.
What
is
the
best
way
but
I
think
even
just
using
a
simple
numpy
rate
could
improve
the
efficiency
a
long
evening,
so
that
calm
be
a
good
starting
point.
But.
A
How
much
money
do
you
still
want?
There
I
think
the
kernel
corn
field
object
has
next
amount
flexibility.
It
guess
we
probably
need
to
trade
off
a
little
bit,
yeah,
some
not
ever
say
sheriff,
but
that
we
need
some
thought
sake.
I
guess,
especially
for
a
lot
of
like
PE
solvers,
but
yeah
problems
or
PC
problems.
Yeah
together.
C
Yeah,
there's
there's
some
there's
a
lot
of
things
to
rethink
through
that,
and
I
also
think
that
when
we
go
through
there,
we
should
think
about
the
API
to
the
fields
object,
because
it
is
like
it's
an
interesting
idea,
but
it's
also
super
strange
and
it's
hard
to
figure
out
like
unless
you
know
that
it's
sort
of
this
smart
dictionary
like.
Why
are
you
indexing
it
like
an
array
but
you're,
passing
the
string
and,
like
that's
strange,
yeah,.
E
J
A
A
C
Absolutely
well
I
mean,
if
that's
something
that
you
want
to
dive
into
and
even
diagnose,
where
the
issues
are
and
capture
that
in
an
issue
on
simpang
like
so
knowing
it,
we
need
we've
isolated.
The
fact
that
the
fields
object
is
a
bottleneck,
I
suspect
that
a
lot
of
that
is
just
the
way
that
we
are
querying,
and
so,
if
you
have
some
examples
that
show
that
that
would
be
no
thanks
on
my
shoes,
okay,
okay.
A
C
C
B
A
B
B
A
A
A
A
B
B
A
C
B
C
A
A
B
B
B
A
K
E
A
B
A
Kind
of
paper,
so
it
was
working
out
of
that
kind,
repeting
waveform,
so
we
usually
just
make
a
step
off.
For
instance,
it's
just
not
quite
the
case
also
available,
so
we
have
a
lot
of
different
purses
and
we're
stacking
it
outside
I'm,
considering
that
into
a
deceiver
this
year,
especially
in
version
yeah,
so
yeah
I
think
it's
a
part
of
a
research
topic,
I
think
it
yeah.
It's
done
stuff,
I'm,
finalizing
the
paper
I
and
then
I'll
push
the
probably
like
a
merge
step
that
bring
my
branches
into
that
streets.
Thanks
yeah.