TY - GEN
T1 - A Case Study in Analog Co-Processing for Solving Stochastic Differential Equations
AU - Huang, Yipeng
AU - Guo, Ning
AU - Sethumadhavan, Simha
AU - Seok, Mingoo
AU - Tsividis, Yannis
N1 - Funding Information:
ACKNOWLEDGMENT The authors would like to thank Nicolas Clauvelin, Victor Marten, and John Milios at Sendyne Corp. for their collaboration and advice. This material is based upon work supported by the National Science Foundation under Grant No. CNS-1239134, the Defense Advanced Research Projects Agency (DARPA) under Contract No. D16PC00089, and an Alfred P. Sloan Foundation Fellowship.
PY - 2019/1/31
Y1 - 2019/1/31
N2 - Stochastic differential equations (SDEs) are an important class of mathematical models for areas such as physics and finance. Usually the model outputs are in the form of statistics of the dependent variables, generated from many solutions of the SDE using different samples of the random variables. Challenges in using existing conventional digital computer architectures for solving SDEs include: rapidly generating the random input variables for the SDE solutions, and having to use numerical integration to solve the differential equations. Recent work by our group has explored using hybrid analog-digital computing to solve differential equations. In the hybrid computing model, we solve differential equations by encoding variables as continuous values, which evolve in continuous time. In this paper we review the prior work, and study using the architecture, in conjunction with analog noise, to solve a canonical SDE, the Black-Scholes SDE.
AB - Stochastic differential equations (SDEs) are an important class of mathematical models for areas such as physics and finance. Usually the model outputs are in the form of statistics of the dependent variables, generated from many solutions of the SDE using different samples of the random variables. Challenges in using existing conventional digital computer architectures for solving SDEs include: rapidly generating the random input variables for the SDE solutions, and having to use numerical integration to solve the differential equations. Recent work by our group has explored using hybrid analog-digital computing to solve differential equations. In the hybrid computing model, we solve differential equations by encoding variables as continuous values, which evolve in continuous time. In this paper we review the prior work, and study using the architecture, in conjunction with analog noise, to solve a canonical SDE, the Black-Scholes SDE.
KW - Analog computers
KW - Differential equations
KW - Stochastic processes
UR - http://www.scopus.com/inward/record.url?scp=85062783573&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85062783573&partnerID=8YFLogxK
U2 - 10.1109/ICDSP.2018.8631831
DO - 10.1109/ICDSP.2018.8631831
M3 - Conference contribution
AN - SCOPUS:85062783573
T3 - International Conference on Digital Signal Processing, DSP
BT - 2018 IEEE 23rd International Conference on Digital Signal Processing, DSP 2018
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 23rd IEEE International Conference on Digital Signal Processing, DSP 2018
Y2 - 19 November 2018 through 21 November 2018
ER -