Probabilistic Interval-Valued Computation: Toward a Practical Surrogate for Statistics Inside CAD Tools

Interval methods offer a general fine-grain strategy for modeling correlated range uncertainties in numerical algorithms. We present a new improved interval algebra that extends the classical affine form to a more rigorous statistical foundation. Range uncertainties now take the form of confidence intervals. In place of pessimistic interval bounds, we minimize the probability of numerical "escape"; this can tighten interval bounds by an order of magnitude while yielding 10-100 times speedups over Monte Carlo. The formulation relies on the following three critical ideas: liberating the affine model from the assumption of symmetric intervals; a unifying optimization formulation; and a concrete probabilistic model. We refer to these as probabilistic intervals for brevity. Our goal is to understand where we might use these as a surrogate for expensive explicit statistical computations. Results from sparse matrices and graph delay algorithms demonstrate the utility of the approach and the remaining challenges.

[1]  Rob A. Rutenbar,et al.  Interval-valued reduced order statistical interconnect modeling , 2004, ICCAD 2004.

[2]  David Overhauser,et al.  Full-chip verification methods for DSM power distribution systems , 1998, Proceedings 1998 Design and Automation Conference. 35th DAC. (Cat. No.98CH36175).

[3]  William Feller,et al.  An Introduction to Probability Theory and Its Applications , 1951 .

[4]  Sachin S. Sapatnekar,et al.  Statistical Timing Analysis Considering Spatial Correlations using a Single Pert-Like Traversal , 2003, ICCAD 2003.

[5]  Michael Orshansky,et al.  An efficient algorithm for statistical minimization of total power under timing yield constraints , 2005, Proceedings. 42nd Design Automation Conference, 2005..

[6]  Rob A. Rutenbar,et al.  Toward efficient static analysis of finite-precision effects in DSP applications via affine arithmetic modeling , 2003, Proceedings 2003. Design Automation Conference (IEEE Cat. No.03CH37451).

[7]  Gene H. Golub,et al.  Matrix computations (3rd ed.) , 1996 .

[8]  Rob A. Rutenbar,et al.  Probabilistic interval-valued computation: representing and reasoning about uncertainty in dsp and vlsi design , 2005 .

[9]  De Figueiredo,et al.  Self-validated numerical methods and applications , 1997 .

[10]  Lars Hedrich,et al.  Analog circuit sizing based on formal methods using affine arithmetic , 2002, ICCAD 2002.

[11]  G. Marsaglia.  Ratios of Normal Variables and Ratios of Sums of Uniform Variables , 1965 .

[12]  Daniel P. Lopresti,et al.  Interval methods for modeling uncertainty in RC timing analysis , 1992, IEEE Trans. Comput. Aided Des. Integr. Circuits Syst..

[13]  Robert V. Hogg,et al.  Introduction to Mathematical Statistics. , 1966 .

[14]  Rob A. Rutenbar,et al.  Probabilistic interval-valued computation: toward a practical surrogate for statistics inside CAD tools , 2006, 2006 43rd ACM/IEEE Design Automation Conference.

[15]  L. Pileggi,et al.  Asymptotic probability extraction for non-normal distributions of circuit performance , 2004, ICCAD 2004.

[16]  Daniel Zwillinger,et al.  CRC standard mathematical tables and formulae; 30th edition , 1995 .

[17]  M. Figueiredo,et al.  A method to extract non-linear principal components of large datasets - an application in skill transfer , 2005, Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005..

[18]  K. Ravindran,et al.  First-Order Incremental Block-Based Statistical Timing Analysis , 2004, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems.

[19]  Rob A. Rutenbar,et al.  Fast interval-valued statistical interconnect modeling and reduction , 2005, ISPD '05.