23
Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson Department of Chemical Engineering University of California, Santa Barbara Complex Fluids Design Consortium Annual Meeting January 23, 2006

Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

  • Upload
    others

  • View
    5

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Beyond the Mean Field:Efficiently Implementing theComplex Langevin Method

Erin M. LennonGlenn H. Fredrickson

Department of Chemical EngineeringUniversity of California, Santa Barbara

Complex Fluids Design Consortium Annual MeetingJanuary 23, 2006

Page 2: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Disorder in Self-Assembly

Dean et al; Macromolecules 2003

Goal: Increaseepoxy toughness

Amphiphilic blockcopolymers

5% copolymerloading yields 40x to70x increase infracture resistance

Page 3: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Mean Field Results

Model as a blend ofcopolymer andhomopolymer

Hypothetical micellarregion

Matsen; Phys. Rev. Letters 1995

Homopolymer Volume Fraction

Frac

tion

A in

A-B

Dib

lock

?

Page 4: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Mean Field Assumption

Assume a single configuration dominates thepartition function

Fluctuations are suppressed

!

Z = D[W ]e"H [W ]# $ e"H [W*]

Page 5: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Complex Langevin Solution Stochastic sampling of fields

Simultaneous excitation and relaxation

!

"

"tW (r,t) = #$

%H[W ]

%W (r,t)

&

' (

)

* + +,(r,t)

!

"(r,t) = 0

"(r,t)"(r',t') = 2#$(r % r')$(t % t')

Page 6: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Numerical Limitations

System stability and accuracy require restrictively smallstep sizes

Page 7: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Novel Method: SDE Splitting

2nd order Semi-Implicit method

!

dx = (A(x) + B(x))dt +"

xn +1 = x

n +#t

2A

xn+1 + B

˜ x + A

xn + B

xn[ ] +"

˜ x = xn + #t A

xn + B

xn[ ] +"

Page 8: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Novel Method: SDE Splitting

Example:

!

W˜ n = W

n + "t#H

#W

n

+$

ˆ W n +1 =

1

2 + "t ˆ g T

ˆ W n + ˆ $ + "t ˆ g T +1( ) ˆ W

˜ n + "t# ˆ H

#W

˜ n %

& ' '

(

) * *

Page 9: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Method Comparison

Page 10: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Diblock Copolymer Model

!

H[W±] = C dr[(2 f "1)W"# " iW+[+W"

2($N)"1]"V lnQ]

Pressure and Difference Fields

01f

Page 11: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Appearance of Micelles

Page 12: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Shift in the Order Parameter

C = 240

Page 13: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Shift in the Order Parameter

C = 60

Page 14: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Shift in the Order Parameter

C = 30

Page 15: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Future Work

Continue work on efficient numerical methods

Extend simulations into larger systems

Address physical phenomena intractable bycurrent methodology

Page 16: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Diblock/Homopolymer Blends

Matsen; Phys. Rev. Letters 1995

Homopolymer Volume Fraction

Frac

tion

A in

A-B

Dib

lock

?A-B Diblock

A Homopolymer

Page 17: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Micelles in Disorder

New ODT

Unbinding transition

Critical MicelleConcentration

Fluctuations

SCFT Experiments

Page 18: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Diblock as Surfactant

Ternary Blend

Diblock/Homopolymer Blend

Fluctuations

SCFT ExperimentsA-B Diblock

A HomopolymerB Homopolymer

Page 19: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Microemulsions

Bates et al; PRL 1997

Fluctuation Destroy Unbinding Transition

Narrow Region!

Page 20: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Stabilizing Microemulsions

PolydispersityA-B asymmetriesChain Length Variations

Hillmyer et al; J Phys Chem B 1999

Page 21: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Stabilizing MicroemulsionsDiblocks

Symmetric

Varied Length

Varied Composition

+ =

Homopolymers

?

Hillmyer et al; J Phys Chem B 1999

Page 22: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Summary

Improved Numerical MethodsFull Complex Langevin Dynamics tractable

Application to Neat Diblock System

Future Work with Diblock Homopolymer Blends

Page 23: Beyond the Mean Field - UCSB MRSECghf/cfdc_2006/lennon_cfdc_2006.pdf · Beyond the Mean Field: Efficiently Implementing the Complex Langevin Method Erin M. Lennon Glenn H. Fredrickson

Acknowledgements

Hector Ceniceros Carlos Garcia-Cervera Kirill Katsov Eric Cochran George Mohler

Happy Birthday Paul Langevin!Born today, 134 years ago