Bused hypercubes and other pin-optimal networks

Pin minimization is an important issue for massively parallel architectures because the number of processing elements that can be placed on a chip, board, or chassis is often pin limited. A d-dimensional bused hypercube interconnection network is presented that allows nodes to simultaneously (in one clock tick) exchange data across any dimension using only d+1 ports per node rather than 2d. Despite this near two-to-one reduction, the network also allows nodes that are two dimensions apart to simultaneously exchange data; as a result, certain routings can be performed in nearly half the time. The network is shown to be a special case of a general construction in which any set of d permutations can be performed, in one clock tick, using only d+1 ports per node. A lower-bound technique is also presented and used to establish the optimality of the network, as well as that of several other new bused networks     

Balanced parallel sort on hypercube multiprocessors

A parallel sorting algorithm for sorting n elements evenly distributed over 2^d p nodes of a d-dimensional hypercube is presented. The average running time of the algorithm is O((n log n)/p+p log 2n). The algorithm maintains a perfect load balance in the nodes by determining the (kn/p)th elements (k1,. . ., (p-1)) of the final sorted list in advance. These p-1 keys are used to partition the sorted sublists in each node to redistribute data to the nodes to be merged in parallel. The nodes finish the sort with an equal number of elements (n/ p) regardless of the data distribution. A parallel selection algorithm for determining the balanced partition keys in O(p log2n) time is presented. The speed of the sorting algorithm is further enhanced by the distance-d communication capability of the iPSC/2 hypercube computer and a novel conflict-free routing algorithm. Experimental results on a 16-node hypercube computer show that the sorting algorithm is competitive with the previous algorithms and faster for skewed data distributions     

The minimal dimension of stable faces required to guaranteestability of a matrix polytope: D-stability

The problem of determining whether a polytope P of n ×n matrices is D-stable-i.e. whether each point in P has all its eigenvalues in a given nonempty, open, convex, conjugate-symmetric subset D of the complex plane-is discussed. An approach which checks the D-stability of certain faces of P is used. In particular, for each D and n the smallest integer m such that D-stability of every m-dimensional face guarantees D-stability of P is determined. It is shown that, without further information describing the particular structure of a polytope, either (2n-4)-dimensional or (2n-2)-dimensional faces need to be checked for D-stability, depending on the structure of D. Thus more work needs to be done before a computationally tractable algorithm for checking D-stability can be devised     

Multinode broadcast in hypercubes and rings with randomlydistributed length of packets

Multinode broadcast (MNB) in a hypercube and in a ring network of processors is considered. It is assumed that the lengths of the packets that are broadcast are not fixed, but are distributed according to some probabilistic rule, and the optimal times required to execute the MNB are compared for variable and for fixed packet lengths. For large hypercubes, it is shown, under very general probabilistic assumptions on the packet lengths, that the MNB is completed in essentially the same time as when the packet lengths are fixed. In particular, the MNB is completed by time (1+δ)T_s with probability at least 1-ϵ, for any positive ϵ and δ, where T _s is the optimal time required to execute the MNB when the packet lengths are fixed at their mean, provided that the size of the hypercube is large enough. In the case of the ring, it is proved that the average time required to execute a MNB when the packet lengths are exponentially distributed exceeds by a factor of ln n the corresponding time for the case there the packet lengths are fixed at their mean, where n is the number of nodes of the ring     

Properties and performance of folded hypercubes

A new hypercube-type structure, the folded hypercube (FHC), which is basically a standard hypercube with some extra links established between its nodes, is proposed and analyzed. The hardware overhead is almost 1/n, n being the dimensionality of the hypercube, which is negligible for large n. For this new design, optimal routing algorithms are developed and proven to be remarkably more efficient than those of the conventional n-cube. For one-to-one communication, each node can reach any other node in the network in at most [n/2] hops (each hop corresponds to the traversal of a single link), as opposed to n hops in the standard hypercube. One-to-all communication (broadcasting) can also be performed in only [n/2] steps, yielding a 50% improvement in broadcasting time over that of the standard hypercube. All routing algorithms are simple and easy to implement. Correctness proofs for the algorithms are given. For the proposed architecture, communication parameters such as average distance, message traffic density, and communication time delay are derived. In addition, some fault tolerance capabilities of this architecture are quantified and compared to those of the standard cube. It is shown that this structure offers substantial improvement over existing hypercube-type networks in terms of the above-mentioned network parameters     

Algorithms and bounds for shortest paths and diameter in faultyhypercubes

In an n-dimensional hypercube Qn, with the fault set |F|<2_n-2, assuming S and D are not isolated, it is shown that there exists a path of length equal to at most their Hamming distance plus 4. An algorithm with complexity O (|F|logn) is given to find such a path. A bound for the diameter of the faulty hypercube Qn-F, when |F|<2_n-2, as n+2 is obtained. This improves the previously known bound of n+6 obtained by A.-H. Esfahanian (1989). Worst case scenarios are constructed to show that these bounds for shortest paths and diameter are tight. It is also shown that when |F|<2n-2, the diameter bound is reduced to n+1 if every node has at least 2 nonfaulty neighbors and reduced to n if every node has at least 3 nonfaulty neighbors     

A class of weak Kharitonov regions for robust stability of linearuncertain systems

Kharitonov's theorems are generalized to the problem of so-called weak Kharitonov regions for robust stability of linear uncertain systems. Given a polytope of (characteristic) polynomials P and a stability region D in the complex plane, P is called D-stable if the zeros of every polynomial in P are contained in D. It is of interest to know whether the D-stability of the vertices of P implies the D-stability of P. A simple approach is developed which unifies and generalizes many known results on this problem     

Bicoprime factorizations of the plant and their relation to right and left-coprime factorizations

In a general algebraic framework, starting with a bicoprime factorization P=N_prD^-1 N_pl, a right-coprime factorization N_p D_p^-1, a left-coprime factorization D^-1_pN_p, and the generalized Bezout identities associated with the pairs (N_p, D_p) and (D˜ _p, N˜_p) are obtained. The set of all H-stabilizing compensators for P in the unity-feedback configuration S(P, C) are expressed in terms of (N_pr, D, N _pt) and the elements of the Bezout identity. The state-space representation P=C(sI-A)^-1B is included as an example     

Vibrational feedback control in the problem of acoustic stability

The problem of absolute stability in a vibrational feedback controller is introduced and discussed. It is shown that for any rational G(s)=n(s)/d(s ) with d(s) Hurwitz and deg d(s) -deg n(s)=1 there exists a linear dynamic periodic controller that ensures, in a certain sense, the infinite sector of absolute stability. This implies that an additional dynamical element, inserted in the feedback loop, may lead to improvements in the robustness of nonlinear systems     

Efficient algorithms for list ranking and for solving graphproblems on the hypercube

A hypercube algorithm to solve the list ranking problem is presented. Let n be the length of the list, and let p be the number of processors of the hypercube. The algorithm described runs in time O(n/p) when n=Ω(p ^1+ε) for any constant ε>0, and in time O(n log n/p+log³ p) otherwise. This clearly attains a linear speedup when n=Ω(p ^1+ε). Efficient balancing and routing schemes had to be used to achieve the linear speedup. The authors use these techniques to obtain efficient hypercube algorithms for many basic graph problems such as tree expression evaluation, connected and biconnected components, ear decomposition, and st-numbering. These problems are also addressed in the restricted model of one-port communication     

Balanced realization of stable transfer function matrices

Simple formulas are presented to compute the internally balanced minimal realization and the singular decomposition of the Hankel operator of a given continuous-time p×m stable transfer function matrix E(s)/d(s). The proposed formulas involve the Schwarz numbers of d(s) and the singular eigenvalues-eigenmatrices of a suitable finite matrix. Similar results are also obtained for a given discrete-time transfer function matrix     

Odd even shifts in SIMD hypercubes

A linear-time algorithm is developed to perform all odd (even) length circular shifts of data in an SIMD (single-instruction-stream, multiple-data-stream) hypercube. As an application, the algorithm is used to obtain an O(M²+log N) time and O(1) memory per processor algorithm to compute the two-dimensional convolution of an N×N image and an M×M template on an N² processor SIMD hypercube. This improves the previous best complexity of O(M² log M+log N)     

Reduction operations on a distributed memory machine with areconfigurable interconnection network

Performing reduction operations with distributed memory machines whose interconnection networks are reconfigurable is considered. The focus is on machines whose interconnection graph can be configured as any graph of maximum degree d. The best way of interconnecting the p processors as a function of p,d and some problem- and machine-dependent parameters that characterize the ratio communication/arithmetic for the reduction operation are discussed. Experiments on transputer-based networks are in good accordance with the theoretical results     

Decentralized stabilization and pole assignment for general propersystems

Considered is the problem of finding existence conditions and a controller synthesis procedure, using decentralized control, for assigning the poles of a linear time-invariant proper system described by a state-space model (C, A, B, D), where no assumption is made regarding the structure of D. This problem has direct application to the decentralized stabilization problem, decentralized robust servomechanism problem, etc., and is a nontrivial extension to the standard decentralized problem where it is assumed that the direct feedthrough terms either are absent or have a block-diagonal structure     

Kharitonov regions: it suffices to check a subset of vertexpolynomials

The authors deal with the D-stability property of interval polynomials. In particular, they show that certain D-domains are Kharitonov regions. That is, the D-stability of interval polynomials is implied by the D-stability of all its vertex polynomials. They then proceed to show that it suffices to check the D-stability of a subset of the vertex polynomials     

Clustering on a hypercube multicomputer

Squared error clustering algorithms for single-instruction multiple-data (SIMD) hypercubes are presented. The algorithms are shown to be asymptotically faster than previously known algorithms and require less memory per processing element (PE). For a clustering problem with N patterns, M features per pattern, and K clusters, the algorithms complete in O(k+log NM ) steps on NM processor hypercubes. This is optimal up to a constant factor. These results are extended to the case in which NMK processors are available. Experimental results from a multiple-instruction, multiple-data (MIMD) medium-grain hypercube are also presented     

Balancing transformations for unstable nonminimal linear systems

It is shown that an unstable nonminimal continuous (discrete) realization (A, B, C) can be transformed via a similarity transformation into a balanced one if and only if the product of the controllability, observability Gramians is similar to a real diagonal matrix Λ. If, in addition, the eigenvalues of A , say λ, satisfy the relation λ_i+λ _j≠0(λ_iλ_j≠1) then the matrix Λ will always be positive semidefinite, and a balanced realization with its minimal part in the internally balanced form can always be obtained     

Network communication in edge-colored graphs: gossiping

A mechanism for scheduling communications in a network in which individuals exchange information periodically according to a fixed schedule is presented. A proper k edge-coloring of the network is considered to be a schedule of allowed communications such that an edge of color i can be used only at times i modulo k. Within this communication scheduling mechanism, the information exchange problem known as gossiping is considered. It is proved that there is a proper k edge-coloring such that gossip can be completed in a path of n edges in a certain time for n⩾k⩾1. Gossip can not be completed in such a path any earlier under any proper k edge-coloring. In any tree of bounded degree Δ and diameter d, gossip can be completed under a proper Δ edge-coloring in time (Δ-1)d +1. In a k edge-colored cycle of n vertices, other time requirements of gossip are determined     

Serial and parallel algorithms for the medial axis transform

An O(n²) time serial algorithm is developed for obtaining the medial axis transform (MAT) of an n×n image. An O(log n) time CREW PRAM algorithm and an O(log² n) time SIMD hypercube parallel algorithm for the MAT are also developed. Both of these use O(n²) processors. Two problems associated with the MAT, the area and perimeter reporting problem, are studied. An O(log n) time hypercube algorithm is developed for both of them, where n is the number of squares in the MAT, and the algorithms use O(n²) processors