1. Introduction
   1. goal
      1. achieved through a Bayesian optimization process
   2. process
      1. starts with an input photograph of a given crack pattern
      2. extract statistics of the fractures, using fea- tures identified in [SDF∗ 10]
      3. perform a perceptual study
         1. to define a metric for similarity of fracture patterns
      4. an optimization approach
      5. develop a fracture modeling interface
   3. contributions
      1. A perceptual study
      2. An optimization approach
2. Related Work
   1. Aging and Weathering
      1. physically-based and phenomenological simulations
         1. DPH96, MG08
      2. data-driven techniques
         1. WTL∗ 06,GTR∗ 06
      3. combination of two
         1. [BLR∗ 11].
   2. Fracture Simulation
      1. first attempts
         1. [TF88]
      2. O’Brien and Hodgins [OH99]
         1. rich simulations of brittle fracture from conventional Finite Element Method simulation.
      3. ductile fracture
         1. [OBH02].
      4. Quasistatic analysis impact-based fracture
         1. [MMDJ01, MG04, BHTF07, ZJ10]
      5. two steps of a fracture simulation
         1. Fracture initiation
            1. Iben and O’Brien[IO06]
         2. Fracture propagation
            1. [GMD12]
      6. crack generation
         1. [Mou05]
         2. Hsien et al. [HT06]
         3. Desbenoit et al. [DGA05]
         4. [GC01]
   3. Statistical Models for Fractures
      1. [Gri21]
      2. [CBRY09]
      3. [Clo55]
      4. Valette et al. [VPLL08]
      5. Shin et al. [SDF∗ 10]
3. Statistical Pattern Similarity (Case Study)
   1. Three types of statistics from [SDF∗ 10]
      1. fragment(S1)
      2. crack(S2)
      3. junction(S3)
   2. perfom a study
      1. the interface for the user study
   3. The conditions
      1. S1
      2. S2
      3. S3
      4. S1+S2
      5. S2+S3
      6. S1+S3
      7. S1+S2+S3
   4. Results
      1. S1 & S3 > S2
4. Fracture Simulation
   1. a physically-based fracture simulation
      1. crack initiation
         1. Finite Element Method (FEM) [ZTZ05]
         2. crack open: use a stress map that evolves over time as proposed in [IO06]
         3. algorithm
      2. crack propagation and Stress Relaxation
         1. crack propagation method of [GMD12]
         2. When a crack propagates, it alleviates the stress
            1. not modeled in [GMD12].
         3. Noise
         4. Summary of Required Input
            1. a set of material properties (E, ν, ρ, Rcm )
            2. an age parameter (i.e. total simulation time)
            3. junction density
            4. parameters that control the crack initiation and propagation (dσe , rrelax , vrelax , Rcvar , l)
            5. the noise values A and f
5. Optimization of the Simulation Parameters
   1. Preprocess
      1. Material Parameters
      2. Normalization of the Statistics
      3. Range of Values for Optimized Parameters
   2. Optimization of Remaining Parameters
      1. we follow the same approach used in [BZB02] to find our parameters
      2. overview of the optimization process
6. Interactive Fracture Modeling