Post on 01-Jun-2020
The
Beg
inni
ngs
Of A
Par
ticle
F
low
Alg
orith
m
�
Qui
ck s
tatu
s up
date
on
crys
tal E
CA
L
�
Gen
eral
com
men
ts
�
Cur
rent
sta
te o
f the
cod
e
�
Som
e pl
ots
�
May
ski
p so
me
if pr
esse
d fo
r tim
e
Mat
thew
Cha
rles,
U. I
owa
2003
-09-
29
Cry
stal
EC
AL
hard
war
e st
udy
�
Bei
ng d
one
by W
illia
m C
audy
(U
iow
a)
�
Usi
ng h
igh-
yiel
d P
bWO
4 cry
stal
s, s
ourc
e
�
Bee
n ta
king
dat
a vi
a os
cillo
scop
e (s
low
)
�
Fin
ally
sol
ved
DA
Q p
robl
em F
riday
;w
ill s
tart
taki
ng d
ata
via
NIM
cra
te.
�
Man
y th
anks
to P
eter
Kim
.
Cry
stal
EC
AL
hard
war
e st
udy
Pul
se h
eigh
t dis
trib
utio
nw
ith s
ourc
e (w
hite
)w
ithou
t sou
rce
(red
)ta
ken
over
sam
e tim
e in
terv
al
Par
ticle
flow
alg
orith
m
�
Bas
ic p
roce
ss:
�
Loca
te a
clu
ster
(po
ssib
ly s
eede
d)
�
“Fit”
it w
ith a
n ap
prop
riate
sha
pe
�
Sto
re r
esul
ts
�
Ass
ign
ener
gy/h
its fr
om th
e C
AL
�
Itera
te
�
Opt
imal
use
of i
nfor
mat
ion:
�
Sta
rt w
ith m
ost k
now
n/pr
edic
tabl
e cl
uste
rs
�
Mov
e on
to th
ose
with
less
see
d in
form
atio
n
�
Mov
e on
to th
ose
with
no
seed
info
rmat
ion
Flo
w a
lgor
ithm
thou
ghts
Whe
re to
sta
rt
�
MIP
-like
trac
ks, s
eede
d w
ith p
ositi
on a
nd
dire
ctio
n in
fo fr
om tr
acke
r
�
char
ged
pion
s, m
uons
, ...
�
Sho
wer
s w
ith k
now
n po
s. a
nd d
ir.
�
elec
tron
s
�
char
ged
trac
ks th
at s
how
er in
side
the
CA
L
�
Clu
ster
s w
ith s
ome
seed
info
rmat
ion
�
scat
tere
d/fr
agm
ente
d tr
acks
�
Oth
ers
�
gam
mas
etc
.
�
Neu
tral
had
rons
MIP
s: W
hat's
impl
emen
ted
�
Che
atin
g: F
ollo
win
g on
e tr
ack
at a
tim
e
�
See
d us
ing
trut
h in
form
atio
n
�
Sim
ple
trac
k ex
trap
olat
ion
algo
rithm
�
Firs
t fe
w la
yers
(un
til 4
laye
rs a
lread
y hi
t) ju
st
exte
nd M
C s
eed
in a
str
aigh
t lin
e
�
Afte
r th
at, s
trai
ght l
ine
from
hits
in la
st fe
w
laye
rs (
up to
6),
don
e w
ith p
rinci
pal a
xis
(he
p.lc
d.re
con.
clus
ter.
util.
Abs
trac
tClu
ster
)
�
Pic
k up
nea
rby
hits
in s
ame
laye
r
�
Will
ski
p ga
ps o
f up
to n
laye
rs (
usin
g 2)
�
Som
e pl
ots
(e.g
. dE
/dx)
Wha
t's n
ot im
plem
ente
d ye
t
�
Tot
al e
nerg
y de
posi
ted
by c
lust
er
�
Ene
rgy
subt
ract
ion
from
CA
L
�
How
muc
h to
take
? S
houl
d be
rou
ghly
con
sist
ent
with
MIP
, but
allo
w fo
r flu
ctua
tions
.
�
Ref
inin
g (s
econ
d pa
ss to
impr
ove
the
fit)
�
Re-
seed
ing
for
subs
eque
nt c
lust
ers
�
Rec
o in
the
pres
ence
of n
oise
, oth
er c
lust
ers
�
Nee
d to
inco
rpor
ate
Clu
ster
Ana
lysi
s fir
st
�
Tra
cks
cros
sing
from
bar
rel t
o/fr
om e
ndca
p
�
HC
AL
Wha
t's n
ext
�
Und
erst
andi
ng w
hy w
e lo
se c
lust
ers
�
Ref
ine/
seco
nd p
ass
�
Ron
's c
lust
er a
naly
zer
�
Add
noi
se, r
est o
f eve
nt
�
How
do
thre
shol
d cu
ts a
ffect
thin
gs?
�
Will
we
need
to a
llow
for
larg
er g
aps?
�
Pic
k th
resh
olds
�
Ene
rgy-
subt
ract
ion
algo
rithm
Rec
onst
ruct
a c
lust
er w
ith >
0 hi
ts
Par
ticle
dep
osite
d no
ene
rgy
in E
CA
L
MC
Par
ticle
did
n't r
ecor
d w
here
part
icle
ent
ered
CA
L
Par
ticle
dep
osite
d en
ergy
in
EC
AL
but w
e di
dn't
find
it
Pio
ns
panp
yZH
120-
0-50
0.si
o
Unr
econ
stru
ctab
le c
lust
ers?
�
Whe
re d
o th
e cl
uste
rs w
ith n
o E
CA
L hi
ts
com
e fr
om?
�
Whe
re d
o th
e cl
uste
rs w
ith E
CA
L hi
ts b
ut
no v
alid
see
d po
int/m
omen
tum
com
e fr
om?
Pio
ns w
ith n
o E
CA
L hi
ts: d
ecay
poi
nt
End
cap
End
cap
Bar
rel
Pio
ns w
ith E
CA
L hi
ts b
ut n
o se
ed p
oint
/mom
entu
m
End
cap
End
cap
Bar
rel
Muo
nsP
artic
le d
epos
ited
no e
nerg
y in
EC
AL
Par
ticle
dep
osite
d en
ergy
in
EC
AL
but w
e di
dn't
find
it
MC
Par
ticle
did
n't r
ecor
d w
here
part
icle
ent
ered
CA
L
Why
so
man
y m
uons
with
no
hits
in th
e E
CA
L?
Onl
y pl
ottin
g re
sults
for
trac
ks th
at
depo
sit
ener
gy in
E
CA
LW
e fa
iled
to fi
nd th
e se
ed(a
ppro
x 15
%)
Eve
ry la
yer
in E
CA
L hi
t(a
ppro
x 60
%)
Muo
ns: #
laye
rs in
re
cons
truc
ted
clus
ters
Why
so
man
y m
uons
bei
ng
mis
sed/
drop
ped?
All
low
m
omen
tum
.
All
ente
r th
e en
dcap
Nea
rly a
ll sh
ower
(!)
... s
o m
aybe
be
ing
scat
tere
d or
sho
wer
ing
befo
re fi
rst
activ
e la
yer.
Not
ve
rifie
d ye
t.
Onl
y pl
ottin
g re
sults
for
trac
ks th
at d
epos
it en
ergy
in E
CA
L
smal
l pea
k (n
ot s
een
for
muo
ns)ap
prox
33%
appr
ox 5
%
Pio
ns
Why
the
peak
afte
r 4
laye
rs?
(1 /
3)
�
Whe
n th
ere
are
< n
laye
rs h
it pr
evio
usly
, we
extr
apol
ate
prev
ious
bes
t-gu
ess
dire
ctio
n. W
hen
ther
e ar
e >=
n
laye
rs h
it, w
e sw
itch
to u
sing
prin
cipa
l axi
s of
hits
for
dire
ctio
n. D
efau
lt is
n
= 4
. Wha
t if
n =
5?
peak
at 5
Why
the
peak
afte
r 4
laye
rs?
(2 /
3)
�
... s
o it'
s al
gorit
hmic
. Dire
ctio
n es
timat
e fa
ils if
the
clus
ter
is q
uite
err
atic
.
�
Why
is it
see
n in
pio
ns a
nd n
ot m
uons
? B
ecau
se p
ions
are
mor
e lik
ely
to s
how
er.
�
Ver
ify b
y pl
ottin
g do
t pro
duct
:
[initi
al m
omen
tum
].[es
timat
ed m
om]
for
part
icle
s lo
st a
fter
findi
ng h
its in
n
la
yers
, for
n
= 4
,5,6
...
Why
the
peak
afte
r 4
laye
rs?
(3 /
3)
dE/d
x
�
Goo
d cr
oss-
chec
k
�
Will
be
used
for
a cu
t on
ener
gy d
epos
ited
in c
ells
�
Cal
cula
ted
for
core
hit
in e
ach
laye
r, u
sing
be
st e
stim
ate
of tr
ajec
tory
.
MIP
pea
k
Muo
ns: d
E/d
x