Anna Bernasconi

Logic Minimization and Testability of 2-SPP Networks

The 2-SPP networks are three-level EXOR-AND-OR forms, with EXOR gates being restricted to fan-in 2. This paper presents a heuristic algorithm for the synthesis of these networks in a form that is fully testable in the stuck-at fault model (SAFM). The algorithm extends the EXPAND-IRREDUNDANT-REDUCE paradigm of ESPRESSO in heuristic mode, and it iterates local minimization and reshape of a solution until no further improvement can be achieved. This heuristic could escape from local minima using a LAST_GASP-like procedure. Moreover, the testability of 2-SPP networks under the SAFM is studied, and the notion of EXOR-irredundancy is introduced to prove that the computed 2-SPP networks are fully testable under the SAFM. Finally, this paper reports a large set of experiments showing high-quality results with affordable run times, handling also examples whose exact solutions could not be computed.

The average sensitivity of square-freeness

Abstract. We study combinatorial complexity characteristics of a Boolean function related to a natural number theoretic problem. In particular we obtain an asymptotic formula, having a linear main term, for the average sensitivity of the Boolean function deciding whether a given integer is square-free. This result allows us to derive a quadratic lower bound for the formula size complexity of testing square-free numbers and a linear lower bound on the average decision tree depth. We also obtain lower bounds on the degrees of exact and approximate polynomial representations of this function.

Sensitivity vs. block sensitivity (an average-case study)

In this note we study the relationship between the average versions of the sensitivity and of the block sensitivity of Boolean functions.

On Projecting Sums of Products

This paper introduces a new bounded multi-level algebraic form, called projected sum of products (P-SOP), based on projections of minimal SOP forms onto subsets of the Boolean space. After a standard two-level logic minimization, this technique can be used as a very fast postprocessing step for further minimizing the circuit area, increasing the depth of the network by only a constant value. The proposed synthesis algorithms have been implemented and tested with interesting results, which show how about 75% of standard Espresso benchmarks benefit from this postprocessing phase.

Room allocation: a polynomial subcase of the quadratic assignment problem

The quadratic assignment problem (QAP) is among the hardest combinatorial optimization problems. Very few instances of this problem can be solved in polynomial time. In this paper we address the problem of allocating rooms among people in a suitable shape of corridor with some constraints of undesired neighborhood. We give a linear time algorithm for this problem that we formulate as a QAP.

On the complexity of balanced Boolean functions

This paper introduces the notions of balanced and strongly balanced Boolean functions and examines the complexity of these functions using harmonic analysis on the hypercube. The results are applied to derive a lower bound related to AC0 functions.

On the average sensitivity of testing square-free numbers

We study combinatorial complexity characteristics of a Boolean function related to a natural number theoretic problem. In particular we obtain a linear lower bound on the average sensitivity of the Boolean function deciding whether a given integer is square-free. This result allows us to derive a quadratic lower bound for the formula size complexity of testing square-free numbers and a linear lower bound on the average decision tree depth. We also obtain lower bounds on the degrees of exact and approximative polynomial representations of this function.

Logic Synthesis by Signal-Driven Decomposition

This chapter investigates some restructuring techniques based on decomposition and factorization, with the objective to move critical signals toward the output while minimizing area. A specific application is synthesis for minimum switching activity (or high performance), with minimum area penalty, where decompositions with respect to specific critical variables are needed (the ones of highest switching activity, for example). In order to reduce the power consumption of the circuit, the number of gates that are affected by the switching activity of critical signals is maintained constant. This chapter describes new types of factorization that extend Shannon cofactoring and are based on projection functions that change the Hamming distance among the original minterms to favor logic minimization of the component blocks. Moreover, the proposed algorithms generate and exploit don’t care conditions in order to further minimize the final circuit. The related implementations, called P-circuits, show experimentally promising results in area with respect to classical Shannon cofactoring.

Synthesis of Autosymmetric Functions in a New Three-Level Form

Abstract Autosymmetric functions exhibit a special type of regularity that can speed-up the minimization process. Based on this autosymmetry, we propose a three level form of logic synthesis, called ORAX (EXOR-AND-OR), to be compared with the standard minimal SOP (Sum of Products) form. First we provide a fast ORAX minimization algorithm for autosymmetric functions. The ORAX network for a function f has a first level of at most 2(n−k) EXOR gates, followed by the AND-OR levels, where n is the number of input variables and k is the “autosymmetry degree” of f. In general a minimal ORAX form has smaller size than a standard minimal SOP form for the same function. We show how the gain in area of ORAX over SOP can be measured without explicitly generating the latter. If preferred, a SOP expression can be directly derived from the corresponding ORAX. A set of experimental results confirms that the ORAX form is generally more compact than the SOP form, and its synthesis is much faster than classical three-level logic minimization. Indeed ORAX and SOP minimization times are often comparable, and in some cases ORAX synthesis is even faster.

Circuit and decision tree complexity of some number theoretic problems

We extend the area of applications of the Abstract Harmonic Analysis to lower bounds on the circuit and decision tree complexity of Boolean functions related to some number theoretic problems. In particular, we prove that deciding if a given integer is square-free and testing co-primality of two integers by unbounded fan-in circuits of bounded depth requires superpolynomial size.