►
From YouTube: SimPEG meeting Sept 2
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
So
we're
now
live
yeah
who
wants
to
start
us
off.
C
B
B
So
what
you
do
in
the
consortium
they
do
this
this
joint
inversion,
and
so
they
gonna
it's
not
that
I
run
inversions,
but
I
just
need
the
gradient,
so
this
can
now
be
done
via
command
line
with
one
input,
terminal
input,
arguments
and
the
other
thing
is
also
with
a
config
file
and
config
file
parsing.
So
that's
now
in
and
works
or
is
currently
being
tested
by
the
other
parties.
So
yeah,
that's
the
main
thing
I
would
say
other
than
that
pushing
me
to
finally
do
that
pay-per-view
promise.
A
Enough,
well,
it
would
be
cool
actually,
maybe
at
some
point
when
you're
happy
with
it
to
actually
walk
us
through
the
config
file,
because
it
it
would
be
nice
if
we
actually
sort
of
thinking
about
some
of
the
simpag
potentially
like
a
command
line
interface
down
the
line.
If
we
can
align
align
these
so
that
you
could
actually
send,
it
would
be
great
to
be
able
to
send
the
same
config
file
to
both
simpag.
B
A
B
A
A
That'd
be
great,
and
actually
on
that
note,
would
you
mind
just
announcing
that
subsurface
meeting
so
that
everyone
here
is
aware
of
that,
because
I
think
a
few
of
us
are
looped
in
on
slack,
but
not
necessarily
everyone.
B
Yeah,
so
this
subsurface
was
kind
of
what
came
out
of
a
hackathon
at
the
transform
where
we
want
to
have
this
package
that
can
talk
between
a
lot
of
different
packages.
At
the
moment,
mainlanders
pie,
gimli
fatiando,.
B
Myself
and
probably
some
others
and
dom
with
qh
h5.
What
is
it
yeah
and
pi
vista,
of
course,
as
well?
That's
an
important
thing
and
we're
gonna
have
a
meeting.
When
is
it
on
the
we
shifted
it
monday
28th?
B
D
B
No,
you
didn't
receive
it
in
the
sl.
I
thought
I
pinged
you
yeah,
because
florian
wagner
from
pie
gimli,
he
actually
asked
what
is
the
status.
So
that
was
the
reason
I
thought
we
could
make
a
catch-up
meeting,
but
then
he
is
on
holiday
on
the
14th.
So
it's
two
weeks
later
at
the
time.
I
don't
know,
what's
your
time,
it's
here
at
six,
so
it's
like
nine
o'clock,
yeah.
B
Yeah,
so
anyone
is
is
more
than
welcome
to
join
in
and
silently
or
very
verbose
opinions,
it's
really
at
still
at
a
very
early
stage.
B
The
interesting
thing
is
that
miguel
from
chempai-
I
don't
know
who
knows
that
they
have,
he
is
about
to
finish
his
phd
and
they
built
a
start
up
for
senpai.
So
that's
a
company
now
where
chempai
is
going
to
remain
open
source,
it's
mainly
they
do
consulting
or
implement
feature.
That's
the
and
some
research
money.
So
one
part
of
their
proposal
is
actually
to
build
something
like
this
for
easier
interaction
with
genpai
models
to
other
tools.
So
he
has
in
that
way
a
serious
interest
that
he
has
to
do
it.
B
Baine
is
also
very
interested
from
the
vtk
side
and
I
think
dom
also
has
sort
of
the
green
light
from
his
boss.
So
we
hope
it's
actually
going
to
happen
somehow
yeah
and
it
would
be
cool
if
you
then
can
go
with
some
magnetics
from
fatiango
to
simpeg
or
between
your
code
and
my
code
or
by
pie.
Gimli
at
some
point,
even
with
finite
element,
meshes
and
then
go
to
sim
page
and
back
and
and
all
that
stuff.
D
B
A
Yeah
thanks,
you
did
that'll,
be
that'll,
be
good,
and
if
you
don't
mind,
could
you
when
the
meeting
gets
closer,
maybe
the
day
or
so
before?
If
you
could
just
drop
a
reminder
for
us?
Yes,.
B
A
Smells
good
and
if
you
want
to
post
it
on
the
like
simple
youtube
or
the
gsi
youtube
you're
welcome
to
or
if
you
got
somewhere
else
in
mind.
That's
that's
fine,
but
yeah.
B
F
I
got
things
to
distract
me
and
like
with
the
merging
of
the
pgi
branch,
as
I
say,
was
gonna
be
complicated
and,
like
I
realized
that
like,
if
I
do
it
for
like
a
bit
and
then
I
leave
it
for
a
few
days,
I
don't
remember
what
I
tried
it's
just
like
starting
over,
but
then
yeah
I've
had
things
keeping
me
out
of
that,
but
mostly
reading
papers
doing
we're
working
a
bit
of
the
present
the
presentation,
based
on
your
comments
last
last
week,
thanks
again
everyone
for
that.
F
Oh
hi,
daniel
in
seoul
in
soggy's,
picture,
yeah
and
and
working
on,
like
proof
for
like
the
different
journal
like
yeah.
Finally
done.
We
finally
settled
down
on
on
the
on
the
proof
with
the
interpretation
like
we
got
like
four
round
to
make
it
right,
but
it's
done.
F
E
A
Excellent
dom
do
you
have
any
updates.
D
No
new
coding-
I
saw
joe
your
your
request
say,
but
I
just
saw
you
you
merged
it
in
the
app.
So
I'm
sorry
about
that.
D
D
G
D
If
it
starts
breaking
sounds
good,
otherwise
my
plan
and
in
the
short
term
will
be
to.
D
Will
be
to
finally,
my
just
because
people
like
I've
been
asked
to
nick
basically
once
wants
me
to
move
on
so
but
what's
preventing
me
to
do
it,
and
I
will
need
to
do
a
full
request,
for
it
is
the
is
basically
the
the
paralyzation
over
the
tiling
right,
which
currently
we
don't
have
on
fourteen,
and
then
we
need
the
m1d
to
be
fully
merged
in.
D
So
I'm
not
sure
what
the
status
is.
How
far
we
are
from
from
this
to
happen,
I
guess
devin's
not
here,
but.
H
D
Good
man
yeah-
I
can
definitely
pitch
in
two
months
once
I'm
at
that
point
and
yeah
joe,
so
for
the
I'm
guessing
for
the
parallelization
on
the
tiles.
We're
just
gonna
have
an
extra
an
extra
replacement
in
the
yeah.
E
What's
the
what's
the
parallelization
flag.
D
G
D
For
now,
I'm
probably
gonna
call
you
when
I'm
when
I'm
at
at
that
point
yeah
and
otherwise
the
the
carbon,
the
big
carbon
project,
the
simplex
going
to
be
shining.
We
finished
you
know
over,
like
60
inversions,
over
a
bc,
so
there
will
be
more
promo
for
for
snpeg.
That's
what
is.
H
That
carbon
project.
D
Yeah,
it's
the
they
want
to
estimate
how
much
carbon
sequestration
potential
there
is
in
bc,
so
they
basically
inverted.
They
selected
100
sites
that
they
knew
that
they
were
some
sort
of
units
and
then
they
wanted
to
have
a
volume
estimate
of
all
those
all
those
units
in
bc.
So
that's
a
nd
ru
in
the
ru
project.
D
Yeah
it's
going
to
come.
It's
probably
going
to
come
up
like
before
the
end
of
the
year
for
sure.
H
Yeah
and
you
use
the
simplex
dc
code.
A
That's
also
a
good
motivation.
It
would
be
nice
in
the
docs.
If
we
get
like
a
publications
page
or
like
projects
using
simpeg,
I
don't
actually
think
it
strictly
needs
to
be
publications,
but
it
would
be
great
if
we
actually
have
pointers
out
to
the
world
of
like,
where
simpeg's
being
used
for
various
things,
yeah
yeah.
D
Once
I
have
a
hyperlink
I
will,
I
can
add
it
for
sure.
C
I
We
also
just
tested
out
the
1d
with
just
a
project
we
just
had
going
on
here
a
couple
weeks
ago
in
northern
saskatchewan
and
yeah,
the
1d
code
seems
to
be
doing
really
well,
where
it
matches
with
some
of
the
some
of
the
gradient
maps
that
we've
had
previously
there,
and
we
did
a
dipole
dipole
through
there
and
yeah.
It
all
seems
to
be
everything
seems
to
be
driving
all
done
with
synthetics
but
yeah.
They
they
agree
the
dc
and
the
mt.
I
Well,
we're
we've
only
got
like
five
stations
right
now.
We
had
to
stop
because
of
some
weather,
but
when
we
get
more
stations,
I'll
be
putting
it
into
the
3d
code
and
then
have
a
yeah
test
that
one,
I
guess
to
see
how
well
it
combines
or
how
well
it
compares.
I
The
I
think
we
only
went
up
to
about
130
hertz
right
now,
yeah,
so
the
the
inversion
that
we
did
was
yeah.
It
wasn't,
including
any
high
frequency,
yet
we're
going
to
do
that.
Next.
C
I
Me
that
that's
true,
I
didn't
look
at
that
very
clearly,
but
that
would
make
sense
it's
going
to
be
the
high
frequency.
I
Oh
okay,
yeah
sorry
we're
up
at
about
a
thousand
hertz
up
there,
so
yeah
not
super
high
yeah
other
than
that.
Yeah
I've
been
messing
with
those
solvers
quite
a
bit.
This
has
been
yeah.
It's
been
quite
difficult.
I
I
pretty
much
found
that
any
of
the
solvers
like
even
toll
mod,
mumps
and
cardiso.
If
I
run
it
on
my
you
know,
30
core
machine
as
opposed
to
my
128
core
machine.
There
is
no
gain.
It
says
it's
using
the
cores
and
it's
using
everything,
but
you
get
no
gain
and
I'm
maybe
thinking
that's,
maybe
just
our
problem.
I
Just
yeah
the
dc
pro
or
I
guess
all
of
them.
I've
been
testing
it
with
mt
as
well.
Another
couple
things
is
that
monks
is
pretty
fast.
I
It
does
the
ldl
there,
joe,
that
you
were
saying
that
troll
mod
does
it
does
it
for
the
positive,
definite
matrices,
but
we
don't
have
that
every
time
so
in
general,
pardeso's
still
the
better
solver,
because
just
we
can
throw
any
anything
at
it
and
it'll
just
chug
through
it,
whereas
mumps
and
well
yeah
mumps
just
doesn't
handle
anything
other
than
positive,
different,
definite
or
just
generally
general
symmetrics
really
well
anything
else.
It's
pretty.
It
slows
down
a
lot,
so
that's
kind
of
unfortunate.
I
Then
I've
tried
parallelizing
over
the
solvers
kind
of
like
what
we
did
back
like
year
or
so
ago.
There
dom
just
trying
it
with
mumps
and
just
setting
it
to
like
one
thread.
That's
all
it
needs.
I
can
have
full
control
over
moms,
but
still
there's
something
that
they're
they're
battling
over
in
the
background,
because
if
I
gask
it
or
if
I
multi-processing
whatever
module,
I
use
they
both
well
one.
Is
they
can't
pickle
it
and
then
desk?
It
just
dumps
the
cores.
It's
just.
A
John,
if
you
have
a
minimal
example
of
that,
there's
an
issue
on
desk
that
they're
actually
starting
to.
I
think,
look
into
this.
I
think
I
posted
it
I'll,
dig
it
up
again
because
that
might
be
actually
insightful
for
for
them
to
be
able
to
have
a
look
at
because
what
they
like.
A
As
far
as
I
understand
sort
of
the
idea
that
they
were
basically
looking
at,
is
to
actually
just
try
and
monitor
the
progress
of
a
single
computation
and
then
see
if
threads
are
over
subscribed
or
if
cpu
is
oversubscribed
and
then
use
that
to
basically
say
like
oh
yeah,
there's
some
sort
of
other
parallelization
going
under
the
hood.
We
can't
also
oversubscribe
on
top
of
that.
So
if
you
have
a
minimal
example,
it
might
be.
A
E
I
That
much
gains,
but
it
was
something
chol
mod,
is
a
pretty
good
solver
except
I
have.
If
I
have
the
problem
too
big
it.
It
just
says
the
problem's
too
big
and
it
can't
handle
it.
So
that's
that
might
be,
though,
because
I
think
I
can
actually
pass
just
like
the
upper
triangle
and
that
might
give
us
a
little
bit
more
room,
but
then
again
it.
You
know
there
is
a
chance
that
it
might
just
get
too
big
and
that's
only
for
positive,
definite
matrices
as
well,
which
we
don't
have
in
every
case.
I
Yeah
and
again
they're
all
like
all
the
all
of
them.
I
think
actually
told
mod,
though,
is
parallelized
in
that
sense,
but
moms
and
parties
so
they're,
all
just
yeah,
sequential
through
the
to
the
factorization.
G
G
At
some
point,
the
factorization
just
kind
of
you
can't
parallelize
it
too
much
more.
It's
a
very
it's
an
interesting
process,
the
the
way
that
it
gets
solved
and
parallel
they
have
to
it's
a
graph
theory.
It's
really
interesting.
I
Yeah,
I
think
that's
exactly
what
those
starch
and
matey's
packages
are.
They
they're
just
like.
I
G
I
They
need
mpi
to
do
that,
though,
when
you
compile
the
sequential
forms
that
they
just
can't
utilize
that
for
some
reason,
but
maybe
but
the
troll
mod
though
it
looks
like
it's
actually
doing
a
lot
in
parallel
during
that
phase.
So
yeah.
I
Yeah
mom's
mumps
was
good
for
that
and
troll
mod
up
to
a
certain
size
as
well
pardeso,
actually
yeah
it
didn't.
When
I
looked
at
the
code,
it
almost
looks
like
he's,
he's
doing
them
individually,
like
he's
actually
just
calling
to
solve,
or
was
it
dave
who
wrote
that
yeah
yeah,
it
almost
seems
like
he's,
not
actually
solving
all
the
right-hand
sides
that
was
like
sending
them
all
to
partiso
he's
actually
just
going
through
and
chopping
through.
I
I
I
Started
thinking
about
how
I
I
think
you
guys
talked
about
this
before,
but
you
know
having
like
parallelizing,
say
over
taking
one
simulation
and
then
just
having
like
a
bunch
of
sub
simulations
that
are
just
like
one
source,
its
own
mesh
and
things
like
that.
I
was
playing
around
with
it
a
little
bit
and
it
looks
like
if
I
can.
I
I
can
launch
two
simulations
and
pardiso
doesn't
complain,
but
I
haven't
looked
at
any
quite
the
results
yet
to
see
if
or
run
it
enough
to
see
if
they
are
talking
or
switching
up
like
we
saw
before
where
we,
the
values,
just
weren't
coming
back
or
not
all
the
values.
I
H
G
Of
each
other,
but
if
you're
trying
to
solve
call
the
same
solver
instance
because
it
holds
this,
it
holds
a
bunch
of
like
work,
arrays
and
memory
that
are
unique
to
each
instance
of
the
solver.
So
if
you're
trying
to
call
it
multiple
times
since
those
work
arrays
that
get
overwritten
with
each
other,
it
gets
all
messed
up.
That's.
I
Got
to
be
it
yeah
yeah,
so
I'm
just
going
to
try
and
avoid
that
if
we
can
set
that
up
and
then
that
might
set
open
to,
we
could
even
do
it
by
frequencies
or
sources
whatever
you
want
and
then
send
its
own
mesh
or
whatever
you
could
build
from
there.
But
if
it
works,
I
still
gotta
test
that.
D
That's
the
way
to
do
it,
john
for
sure,
and
then
after
this,
you
just
need
to
make
sure
that
you're
not
sending
too
many
of
those
at
the
same
time,
so
that
you
don't
run
out
of
memory
right
because
now
you're
creating
a
bunch
of.
I
E
I
E
I
B
Sorry,
just
one
quick
question
with
a
very
very
little
chance:
do
you
run
that
in
a
notebook.
I
A
Thanks,
john
joe,
would
you
like
to
go
next.
G
Sure,
like
we
were
talking
about
with
dom
earlier,
I
got
a
few
of
the
volume
averaging
things
a
little
bit
more
ironed
out.
As
far
as
you
can.
We
can
make
a
few
assumptions
when
they
share
the
same
base
mesh.
G
So
I
have
that
working
for
arbitrary
tree
meshes
as
long
as
they
have
the
same
base.
Mesh
doesn't
there's
no
requirements
on
like
respective
levels
of
each
cell
in
it
so
dom.
The
original
thing
you
had
to
put
together
required
that
every
cell
in
the
local
mesh
was
the
same
or
lower
likes,
but
that
is
not
a
requirement
anymore.
So
it's
a
little
bit
more
general
as
well
as
far
as
that
is
concerned.
G
A
few
is
still
issues
as
far
as
testing
and
it
seems
to
be
similar
to
some
of
the
tests
that
have
been
failing
on
travis
recently,
and
I
still
cannot
figure
out
why
they're
failing
I
have
my
next
step
is
to
get
a
ubuntu
virtual
image,
booted
up
and
start
testing
inside
of
that
to
see
if
I
can
start
reproducing
it,
but
I
haven't
had
any
luck
reproducing
the
seg
faults
on
any
of
the
machines
I've
tried.
G
Last
week
I
brought
up
a
little
bit
about
the
new
inverse
testing.
Has
anyone
got
a
chance
to
look
through
that.
G
E
G
E
G
A
Yeah
three
three
o'clock
would
be
great
for
me.
G
I'm
working
on
the
release,
notes
right
now.
I'm
thinking
it's
a
little
bit
more
than
a
patch,
so
I
think
it's
zero
five
zero.
A
E
B
G
There's
a
few
small
polar
clips
in
there
that
we
should
at
least
check,
I
think,
they're.
We
can
look
at
some
of
those
on
friday
too,
and
see
what
see
the
state
of
them
and
close
them
up
if
they're,
outdated
or
double
check
on
a
few
things
so
we'll
deal
with
that
on
friday
then
mean
time,
I'm
going
to
get
that
release,
notes
out
and
push
and
tag
it,
and
it
should
all
automatically
go
to
everywhere.
G
G
Submitting
one
of
the
papers
for
my
thesis
on
the
second
work
that
was
outside
of
the
thesis
I
should
hopefully
have
both
those
in
this
week.
G
Combining
and
denoising
gravity
and
gravity
gradient
data
together
using
so
it's
like
optimally
weighting
them
in
an
inversion
is
kind
of
the
thing.
The
idea
so
kind
of
combining
you
know,
surface
gravity
with
airborne
gravity,
gradient
data.
At
the
same
time,
an
optic
figuring
out
optimal
weights
between
the
two
data
sets
and
also
regularizing
the
model,
at
the
same
time,
to
get
a
better
overall
picture,
supplement
the
long
wave,
the
missing
long
wavelength,
information
and
gravity
gradient
data.
D
E
D
No,
I
was
saying
very
useful.
We
should
expand,
extend
this
to
mag
right,
pretty
quick,
because,
after
the
talk
of
your
boss,
john
people
are.
D
To
do
more,
more
mag,
radiometry.
D
Yeah,
I
put
it
up
on
youtube
I'll.
I
can
post
them
that'd
be
great.
E
G
A
Excellent
well,
look
forward
to
seeing
it
you'll
have
to
send
us,
send
the
links
around
when
you
get
it
out
sure
cool,
well,
soggy
and
daniel.
H
Nothing
in
right
on
my
end,
it's
like
that
kind
of
rebounding
that
1d
code
and
3d
code.
H
H
Yeah,
that's
that's!
Let's
pretty
good
at
that
with
lindsay's
example.
It
was
interesting
that
peter
was
using
the
log
resistivity,
so
yeah,
it's
a
so
you
need
to
be
careful
what
you're
going
to
average,
because
you
can
average
resistivity,
conductivity
or
log
over
this
tv
depending
upon
the
problem.
Your
choice
of
property
is
important,
so
yeah
like
it's
great,
to
have
a
voluntary,
proper
choice
of
what
are
you
going
to
interpolate,
it's
quite
important
depending
upon
the
problem
yeah,
I
don't
have
any
updates.
H
B
D
H
H
A
H
Food
he's
having
solids
but
absolutely.
C
I
can
walk
that's
good,
actually,
I'm
standing
now
in
front
of
my
screen.
I
can
actually
stand
on
one
foot.
I
can
get
lots
of
exercise
excellent.
Now.
I
think
we
were
we've
been
consumed
over
the
last
week
with
all
of
the
geophysics
stuff,
so
the
stuff
for
the
scg.
So
really
congratulations
to
the
whole
team
here
for
getting
all
of
those
in
that's
yeah
four
talks
all
pre-recorded,
so
I
think
we're
pretty
much
good
to
go.
So
we
can
catch
our
breath
on
that
one
and
look
at
other
things.
At
this
point.
A
Any
updates-
I
I
don't
know
what
I
do.
No
I've
been
working
on
the
well
started
working
on
the
marla
model
and
I'm
excited
joe
that
the
discretize
updates
will
be
coming
out
because
I
was
running
into
issues
compiling
compiling
the
new
code
when
I
think
you
were
still
in
progress
so
give
that
a
go
and
hopefully
be
able
to
show
some
of
this
week.
I
also
was
the
recording
of
the
scg
stuff,
took
a
really
long
time,
pre-recording
a
talk
when
there's
no
one
watching.
A
You
is
a
really
weird
thing
to
do
so.
It
took
me
the
better
part
of
an
afternoon
to
just
like
finally
stop
thinking
about
it
and
get
through
the
entire
talk,
but
anyways
it's
done,
and
I
think
it
was
okay.
Doug
said
it
was
okay,
I
couldn't
watch
it,
so
it
was
great.
A
So
I
could
send
that
around
if
folks
are
interested,
because
I
sort
of
been
actually
trying
to
work,
make
some
more
progress
on
the
uxo
problem
recently
with
with
that
and
kind
of
thinking,
through
some
new
examples
that
black
tusk
sent
along
so
nothing
too
much
to
report
there.
But
it's
it's
kind
of
in
progress.
C
C
Send
the
links
to
all
four
talks
around
because
they're
simple
is
involved
in
all
of
them.
A
Yeah
so
I
think
they're
on
dropbox.
I
don't
think
we
can
post
them
on
youtube.
I
don't
actually
know
because
it's
scg,
but
they'll
probably
be
upset
about
that.
If
we
make
all
of
our
talks
for
you
and
they
want
you
to
pay
for
the
conference
so,
but
we
can
send
him
around
internally
on
on
slack
for
sure
yeah.
I
think
that's
about
it.
B
B
B
Yeah
he
did
the
masterpiece
with
us,
which
was
a
couple
of
months
with
fugro
for
detecting
bombs,
mainly
offshore,
like
that.
The
target
would
be
here
offshore,
the
netherlands,
on
the
coast,
where
there
are
a
lot
of
bombs,
usually
from
fighters
that
left
and
dropped
before
going
home
or
after
the
war
that
were
dumped
into
the
water
and
now
with
all
the
wind
middle
offshore
windmill
activity.
B
B
E
A
Excellent,
I
guess
actually
one
other
update
that
I
would
be
curious
to
hear
folks,
thoughts
on
so
through
swung
or
thinking
about
doing
an
event
similar
to
transform
but
more
focused
on
research,
probably
sort
of
end
november
middle
end
of
november
timeline
and
I'm
meeting
with
a
few
folks
tomorrow,
steve
and
and
a
couple
others
who
are
sort
of
championing
the
organization
of
this,
but
just
trying
to
think
through,
like
what
would
be
a
an
interesting,
useful
event
that
you'd
be
excited
to
participate
in
with
this
community
sort
of
thinking.
A
So
the
transform
really
focused
on
tools
and
what
sorts
of
tools
either
training
like
you
know,
providing
tutorials
and
things
like
that,
or
also
the
hackathon
and
building
new
new
resources
for
the
community.
A
D
D
I
don't
know
which
is
kind
of
usual,
the
usual
conference
thing,
but
if
we
could
pack
it
in
a
more
geoscience
open
source
form
yeah.
That
would
be
super
useful
to
me.
Yeah,
because
I
use
I
can't
keep
track
of
you
know.
I
can't
stay
on
song
and
read
all
the
threads
there's
a
million
a
day,
so
I
lose
track
of
the
good
stuff
that
is
being
posted.
A
Yeah
and
what
would
you
think
about
that
then
followed
by
discussion?
Do
you
think
that
would
be
like
if
we
sort
of
have
a
theme?
So
if
we
maybe
have,
I
don't
know
about
a
two
or
three
hour
event,
we
spend
45
minutes
on
lightning
talks
and
then
have
time
to
chat
afterwards.
D
I
think
that's
a
bit
tough,
because
when
you
have
a
big
group,
it
it
just
you
know,
swamps,
say
there's
only
one
or
two
that
that
speaks
tons
and
minerals.
Don't.
I
think
it
would
be
interesting
if
the
people
who
give
lightning
talks
if
they
could
be
in
their
own
respective
channels
or
like
yeah
slack,
not
stacks
like.
F
D
A
That
would
be
cool,
so
in
a
sense,
it's
like
a
lightning
talk,
plus
poster
virtual
poster
session
in
a
sense
or
virtual
coffee,
whatever
chance
to
chat
with
people
yeah
yeah,
okay,
excellent.
I
appreciate
it
I'll.
A
Perfect:
okay!
Well
thanks
I'll
I'll
bring
this
to
folks
tomorrow
and
if,
in
the
meantime,
if
you
have
any
other
ideas
or
things
that
you'd
like
to
see,
feel
free
to
just
ping
me
and
send
a
message,
and
I
can.
I
can
bring
that
up
to
folks.
B
Yeah,
I
definitely
wouldn't
think
about
it.
I
wouldn't
do
45
minutes,
lightning
talks
and
then
discussions,
because
probably
one
or
two
topic
would
take
overhand
and
the
others
will
get
forgotten.
It
would
have
to
be
lightning
talk
and
short
discussion,
like
nick
talk,
short
discussion
or,
like
dom
said
it
with
breakout
rooms
or
stuff
like
that.
E
I
I
For
dc,
I'm
working
off
of
that,
but
mt
I'm
still
on
my
simulation
empty
branch.
E
A
H
I
have
a
meeting
at
11
30,
so
I
yeah.
I
only
have
a
15
minutes.