swMATH ID: 31779
Software Authors: Wang, Mu; Gebremedhin, Assefaw; Pothen, Alex
Description: Capitalizing on live variables: new algorithms for efficient Hessian computation via automatic differentiation. We revisit an algorithm [called Edge Pushing (EP)] for computing Hessians using Automatic Differentiation (AD) recently proposed by Gower and Mello (Optim Methods Softw 27(2): 233–249, 2012). Here we give a new, simpler derivation for the EP algorithm based on the notion of live variables from data-flow analysis in compiler theory and redesign the algorithm with close attention to general applicability and performance. We call this algorithm Livarh and develop an extension of Livarh that incorporates preaccumulation to further reduce execution time—the resulting algorithm is called Livarhacc. We engineer robust implementations for both algorithms Livarh and Livarhacc within ADOL-C, a widely-used operator overloading based AD software tool. Rigorous complexity analyses for the algorithms are provided, and the performance of the algorithms is evaluated using a mesh optimization application and several kinds of synthetic functions as testbeds. The results show that the new algorithms outperform state-of-the-art sparse methods (based on sparsity pattern detection, coloring, compressed matrix evaluation, and recovery) in some cases by orders of magnitude. We have made our implementation available online as open-source software and it will be included in a future release of ADOL-C. .. 5.3 Implementing Livarhacc
Homepage: http://mpc.zib.de/archive/2016/4/Wang2016_Article_CapitalizingOnLiveVariablesNew.pdf
Keywords: algorithmic differentiation; Hessian computation; reverse mode AD; edge pushing; ADOL-C; data-flow analysis
Related Software: Livarh; edge_push_sp; ColPack; MatrixCalculus; L-BFGS; TADIFF; FADBAD++; NUTS; torchdiffeq; JuMP; Julia; ve08; LIBSVM; AMPL; hess_pat; dcc; Valgrind; ADOL-C
Cited in: 3 Documents

Citations by Year