►
From YouTube: SimPEG Meeting July 14th
Description
Weekly SimPEG meeting from July 14th, 2021
A
A
A
B
And
feel
free
to
prod
other
people
too,
who
don't
necessarily
always
come
to
the
meetings?
It
would
be
great
to
get
other
folks.
So
if
there's
somebody
that
you
know
has
published
something
or
that
you're
curious
to
hear
what
they're
up
to
it'd
be
great
to
to
reach
out
and
use
this
to
bring
in
some
other
folks,
so
don't
hesitate
to,
especially
because
we're
just
getting
this
started
to
poke
people
and
ask
them
to
to
step
up.
C
A
Either
of
those
days
are
good
with
me,
so
let's
plan
it.
Let's
try
to
start
planning
this
thing
for
sunday,
the
25th
and
everyone's
good
for
that
day
figure
out
what
exactly
we're
doing
we're
likely
going
to
the
beach.
Maybe
we
can
grill
up
some
things
and
just
kind
of
socialize
for
a
while,
and
I
don't
bring
some
I'll
bring
some
bocce
ball.
We
can
play
that.
A
Okay,
I'm
going
to
move
on
to
some
quick
reports
here,
so
I
I've
still
been
working
through
the
cross
gradient
and
I'm
getting
through
it
just
kind
of
going
through
making
sure
that
all
the
code
is,
you
know,
just
concise,
updating
up
to
snuff
and
getting
the
just
still
trying
to
get
those
excuse
me.
Those
directives,
married
together
with
some
of
the
other
things
that
we
have.
A
A
A
Oh,
you
multiply
and
then
average.
So
it's
kind
of
goes
on
lines
of.
Like
the
same
reason,
we
do
those
operators
when
we
have
take
the
pdes
right,
we
think
about
multiplying
things
together,
first
and
then
averaging
instead
of
averaging
and
then
multiplying,
because
it
tends
to
work
better,
the
second
or
the
first
way
and
looking
back
at
his
paper,
that's
how
they
had
recommended
implementing
it
as
well-
and
I
was
just
curious
if
you
had
seen
like
if
you
had
looked
it
into
any
differences
there.
A
D
The
dr
hipper,
yes
yeah,
I
remember
in
that
paper
the
doctor
people
proposed
we
can
use,
we
can
transfer
the
cross
gradient
function
to
the
vector,
multiply
minus
a
dot
product
right
right,
yeah,
that's
a
actually!
This
is
a
both
way.
It
works
well
because
I
already
derived
the
first
order
derivative
and
the
second
order
derivative
of
cross
gradient
using
the
cross
product
and
the
dot
plus
dot
product.
A
A
Go
ahead,
it
was
it's
a
kind
of
a
way
of
having
to
deal
with
those
like
we
talked
about
before,
like
having
that
weird,
plus
one
minus
one,
plus
one
minus
one
thing
in
the
no
space
operator,
and
I
was
curious
if
this
made
it
the
way
that
they
had
suggested
made
it
a
little
bit
better.
If
I
can
show
you
yeah.
A
A
A
A
B
Yeah,
just
a
quick
update,
so
we
got
the
first
newsletter
out.
So
thanks
everyone
for
help
on
that,
and
I
just
pasted
in
a
screenshot
of
the
stats.
So
it
looks
like
255
people
opened
it
and
70
people
clicked
on
a
thing.
So
so
that's
pretty
cool.
That
actually
means
you
know
we're
we're
reaching
a
pretty
a
reasonably
large
audience,
and
so
you
know,
I
think,
that's
a
positive
thing
for
the
seminars
and
also
you
know
for
any
announcements
and
things
like
that
that
folks
have
about
papers.
B
Cool
things
you've
been
up
to
we've
got
a
bit
of
an
audience
that
we
can
tell
about
that.
So
that's
kind
of
cool,
so
yeah
and
then
just
I'll
repeat
because
donuts
doggy
joined
after
I
mentioned
this,
but
for
the
next
seminars
feel
free
to
reach
out
and
please
please
do
reach
out
to
folks
that
you'd
like
to
hear
from
and
yeah
see
when
people
might
like
to
give
a
talk.
B
E
B
C
Yeah
so
been
making
some
pretty
good
progress
on
the
doc
strings.
I
think
maybe
a
final
draft
of
of
all
the
documentation
is
about
90
complete
now,
so
I
have
been
getting
some
feedback
from
joe
along
the
way
and
it's
going
pretty
well
and
yeah,
I'm
hoping
to
kind
of
find
any
little
bits
and
pieces.
I
haven't
documented
and
kind
of
complete
that
today
so
yeah
we're,
I
think,
we're
on
track
to
having
something.
Maybe
we
can
almost
release
in
a
week
or
two
seems
yeah.
C
E
So
dear,
and
I
worked
still
working
still
hard
on
this
merger
of
eg3
and
synthetic
we're
still
having
a
hard
time
this
memory
leakage,
we
actually
tried
many
different
things
and
finding
very
odd
behavior.
So
on
my
mag,
it's
it's
working!
Well
on
my
linux
machine,
it's
getting
worse
and
there's
a
little
bit
different
behavior
on
my
cluster
and
peter's
linux
machine.
So
yeah
we're
a
bit
confused
getting
out
of
ideas,
we're
hoping
to
kind
of
tackle
that
in
the
next
week,
so
yeah.
E
We
did
we
actually
ran
many
many
many
memory
profiles
and
well
still
not
catching
up.
That's
there's
an.
I
think
it's
related
to
the
the
minimization
package,
because
I
think
this
iterative
process
of
passing
vectors
to
a
jpeg
and
jt
back
and
looping
this
over.
It
seems
like
there
is
a
point
where
the
memory
can
leak,
which
is
very
hard
to
keep
track
of.
E
F
Go
ahead
yeah,
I
just
thought
the
weird
thing
is:
if
we
try
to
separate
and
if
you
run
stuff
simpek
only
there
just
doesn't
seem
to
be
an
issue
and
if
you
do
run
stuff
emg
3d
only
then
it
doesn't
seem
to
be
an
issue,
and
if
you
connect
the
two,
there
seems
to
be
this
problem
and
we
don't
exactly
know
where,
where
it
happens
between
the
connection,
because
then
it's
extremely
hard
to
say
does
it
happen
in
emg
3d?
Does
it
happen
in
simpac
somewhere?
E
F
A
It's
definitely
getting
much
more
improved.
Now,
okay,
I
really
appreciate
the
stuff
that
kind
of
like
the
camera
install
makes
it
really
nice
for
us
to
be
able
to
distribute
that,
like
I
know,
there's
some
other,
like
package
managers
like
that
that
are
doing
similar
things
they're
a
little
bit
more
open,
but
it's
it's
just
nice
to
have
all
those
out
there.
People
realize
that
it
was
a
problem
and
then
took
steps
to
address
it.
E
F
E
B
E
E
B
A
I'll
probably
be
reaching
out
and
seeing
what
we
can
reaching
out
to
a
few
people
and
make
sure
we
get
okay
for
the
meet
up
next
week
and
to
see
to
make
sure
that
we
have
everything
planned
properly,
but
other
than
that,
I
don't
think,
there's
anything
else
that
we
should.
We
need
to
go
through
today,
yeah.