Transcript Document
Nuclear Multifragmentation
and Zipf’s Law
Wolfgang Bauer
Michigan State University
Work in collaboration with:
Scott Pratt (MSU), Marko Kleine Berkenbusch (Chicago)
Brandon Alleman (Hope College)
Nuclear Matter Phase Diagram
Two (at least) thermodynamic phase
transitions in nuclear matter:
– “Liquid Gas”
– Hadron gasQGP / chiral restoration
Problems / Opportunities:
– Finite size effects
– Is there equilibrium?
– Measurement of state
variables (r, T, S, p, …)
– Migration of nuclear system through phase
diagram (expansion, collective flow)
Structural Phase Transitions (deformation,
spin, pairing, …)
Source: NUCLEAR SCIENCE, A
Teacher’s Guide to the Nuclear
Science Wall Chart,
Figure 9-2
– have similar problems & questions
– lack macroscopic equivalent
2
History
3
Influence of Sequential Decays
Critical fluctuations
Blurring due to
sequential decays
4
Width of Isotope Distribution,
Sequential Decays
Predictions for width of
isotope distribution are
sensitive to isospin term in
nuclear EoS
Complication:
Sequential decay almost
totally dominates
experimentally observable
fragment yields
Pratt, WB, Morling, Underhill,
PRC 63, 034608 (2001).
5
Isospin: RIA Reaction Physics
Exploration of the drip lines below
charge Z~40 via projectile
fragmentation reactions
Determination of the isospin
degree of freedom in the
nuclear equation of state
Astrophysical relevance
Review:
r-process
rp-process
B.A. Li, C.M. Ko, WB,
Int. J. Mod. Phys. E 7(2),
147 (1998)
6
Cross-Disciplinary Comparison
Left: Nuclear
Fragmentation
Right: Buckyball
Fragmentation
Histograms:
Percolation Models
Similarities:
– U - shape
(b-integration)
– Power-law for
imf’s
(1.3 vs. 2.6)
– Binding energy
effects provide
fine structure
Data: Bujak et al., PRC 32, 620 (1985)
LeBrun et al., PRL 72, 3965 (1994)
Calc.: W.B., PRC 38, 1297 (1988)
Cheng et al., PRA 54, 3182 (1996)
7
Buckyball
Fragmentation
Cheng et al., PRA 54, 3182 (1996)
Binding energy
of C60:
420 eV
625 MeV
Xe35+
8
ISiS BNL Experiment
10.8 GeV p or p + Au
Indiana Silicon Strip Array
Experiment performed
at AGS accelerator of
Brookhaven
National
Laboratory
Vic Viola
et al.
9
ISIS Data Analysis
•Marko Kleine Berkenbusch
•Collaboration w. Viola group
Reaction:
p, p+Au @AGS
Very good statistics
(~106 complete events)
Philosophy: Don’t deal with energy
deposition models, but take this
information from experiment!
Detector acceptance effects
crucial
Residue
Sizes
Residue
Excitation
Energies
– filtered calculations, instead of
corrected data
Parameter-free calculations
10
Comparison:
Data & Theory
2nd Moments
Charge Yield
Spectrum
Very good agreement
between theory and data
– Filter very important
– Sequential decay corrections
huge
11
Scaling Analysis
Idea (Elliott et al.): If data
follow scaling function
T Tc
N(Z,T ) Z f Z
Tc
with f(0) = 1 (think
“exponential”), then we can use
scaling plot to see if data cross
the point [0,1] -> critical events
Idea works for theory
Note:
– Critical events present, p>pc
– Critical value of pc was corrected for
finite size of system
M. Kleine Berkenbusch et al.,
PRL 88, 0022701 (2002)
12
Effects of Detector Acceptance Filter
Unfiltered
Filtered
13
Scaling of ISIS Data
Most important: critical region and
explosive events probed in experiment
Possibility to narrow window of critical
parameters
: vertical dispersion
: horizontal dispersion
– Tc: horizontal shift
c2 Analysis to find
critical exponents
and temperature
Result:
0.5 0.1
2.35 0.05
Tc (8.3 0.2) MeV
14
Essential for Scaling of Data:
Correction for Sequential Decays
15
The Competition …
Work based on Fisher liquid
drop model
nA q0 A e
1
(A c0 A )
T
Same conclusion:
Critical point is reached
Result:
0.54 0.01
2.18 0.14
Tc (6.7 0.2) MeV
J.B. Elliott et al., PRL 88, 042701 (2002)
16
IMF Probability Distributions
Moby Dick:
IMF: word with ≥ 10 characters
Nuclear Physics:
IMF: fragment with 20 ≥ Z ≥ 3
System Size is the
determining factor
in the P(n) distributions
17
Zipf’s Law
Back to Linguistics
Count number of words in a book (in English) and
order the words by their frequency of appearance
Find that the most frequent word appears twice as
often as next most popular word, three times as
often as 3rd most popular, and so on.
Astonishing observation!
G. K. Zipf, Human Behavior and the Principle of Least Effort
(Addisson-Wesley, Cambridge, MA, 1949)
18
English Word Frequency
201
181
161
1
fn
n
141
f1
1.4
fn
121
101
81
61
41
21
1
1
21
41
61
81
101 121 141 161 181 201
n
W ord
the
of
and
a
in
to
it
is
was
to
i
for
you
he
be
with
on
that
by
at
Rank
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
British language compound, >4000 texts
2
1
1
1
1
19
1
DJIA-1st Digit
1st digit of DJIA
is not uniformly
distributed from 1
through 9!
Consequence of
exponential rise
(~6.9% annual
average
Also psychological
effects visible
20
Zipf’s Law in Percolation
Sort clusters according
to size at critical point
Largest cluster is n
times bigger than nth
largest cluster
M. Watanabe, PRE 53 (‘96)
21
Zipf’s Law in Fragmentation
Calculation with Lattice
Gas Model
Fit largest fragments
to
An = c n-
At critical T:
crosses 1
New way to detect
criticality (?)
Y.G. Ma, PRL 83 (‘99)
22
Zipf’s Law: First Attempt
N (A,T ) aA f [A (T Tc )]
<A1>/<Ar>
at Tc : f (0) 1
N (A,Tc ) aA
rank, r
23
Zipf’s Law: Probabilities (1)
Probability that cluster of size A is the largest one =
probability that at least one cluster of size A is present
times probability that there are 0 clusters of size >A
P1st (A) p1 (A) p0 ( A)
[1 p0 (A)] p0 ( A)
N(A) = average yield of size A: N(A) = aA-
N(>A) = average yield of size A: (V = event size)
N( A)
V
V
i A1
i A1
N(i) aA a ( ,1 A) a ( ,1 V )
Normalization constant a from condition:
V
A N(A) V
A1
V
a V / A1 V / HV(1 )
A1
24
Zipf’s Law: Probabilities (2)
Use Poisson statistics for individual probabilities:
N(i) e N (i )
pn (i)
n!
p0 (i) e N (i ) ; p1 (i) N(i) p0 (i); p2 (i)
n
1
2
N(i) p1 (i)...
Put it all together:
P1st (A) [1 p0 (A)] p0 ( A)
[1 e N (A) ] e[a ( ,1 A)a ( ,1V )]
Average size of biggest cluster
V
A1st A P1st (A)
A1
(Exact expression!)
25
Zipf’s Law: Probabilities (3)
Probability for given A to be 2nd biggest cluster:
P2nd (A) p2 (A) p0 ( A) p1 (A) p1 ( A)
[1 p0 (A) p1 (A)] p0 ( A) [1 p0 (A)] p1 ( A)
Average size of 2nd
biggest cluster:
V
A2nd A P2nd (A)
A1
And so on … (recursion
relations!)
26
Zipf’s Law: -dependence
20
A1 / An
Verdict: Zipf’s Law does not work
for multifragmentation, even at the
critical point! (but it’s close)
Series1
2.00
Series2
2.18
Series3
2.33
Series4
2.50
Series5
2.70
Series6
3.00
Series7
5.00
18
16
14
12
Expectation
if Zipf’s Law
was exact
10
8
6
4
2
0
1
2
3
4
5
6
7
n
Resulting distributions: Zipf Mandelbrot
8
9
10
W.B., Pratt (2005)
27
Human Genome
1-d partitioning problem of gene length distribution
on DNA
Human DNA consist of 3G base pairs on 46
chromosomes, grouped into codons of length 3 base
pairs
– Introns form genes
– Interspersed by exons; “junk DNA”
QuickT i me™ and a
T IFF (Uncompressed) decom pressor
are needed to see this picture.
28
Computer Hard Drive
Genome like a computer hard
drive.
Memory is like chromosomes.
A files analogous to genes.
To delete a file, or gene,
delete beginning.
29
Recursive Method
Number of ways a length A string can split into m pieces
with no piece larger than i.
i
mj
N A, m, i N A j, m 1, i
A
j
Probability the lth longest piece has length i
A l 1 A
m!
N A l k i, m l , i 1N A Asmall ik , k , imax
Asmall k 0 l 1 k!l k !m l !
Pl , i
N A, m, imax
30
Simulation
Random numbers are
generated to determine
where cuts are made.
Here length is 300 and
number of pieces is 30.
31
Assumption: Relaxed Total Size
The number of pieces falls exponentially.
i
ni Ce
From this assumption the average piece size is obtained.
1
i
Also, the average size of the longest piece.
2A
P i ln
i
1
32
Power Law – Percolation Theory
Assumes pieces fall according to a power law.
na Ca
Average length of piece N is:
N 1
1
1 C 1
N
P
N 1
33
Gene Data
Alleman, Pratt, WB 2005
Data from
Chromosomes 1, 2, 7,
10, 17, and Y.
Plotted against
Exponential and
Power Law models in
Green.
34
Summary
Scaling analysis (properly corrected for decays and
feeding) is useful to extract critical point
parameters.
“Zipf’s Law” does not work as advertised, but
analysis along these lines can dig up useful
information on critical exponent , finite size
scaling, self-organized criticality
Gene length distribution as a 1d partitioning
problem is interesting and not solved
Research funded by US National Science Foundation
Grant PHY-0245009
35