An Implementation of a Hop eld Network
نویسندگان
چکیده
EARTH is a multithreaded program execution and architecture model that hides communication and synchronization latencies through ne-grain multithreading. EARTH provides a simple synchronization mechanism: a thread is spawned when a pre-speciied number of synchronization signals are received in its synchronization slot { signaling the fact that all dependences required for its execution are satissed. This simple synchronization mechanism is an essential primitive in Threaded-C { a multithreaded language designed to program applications on EARTH. The EARTH synchronization mechanism has been eeciently implemented on a number of computer platforms, and has played an essential role in the support of a large number of parallel applications on EARTH. An interesting open question has been: is such a simple mechanism suucient to satisfy the synchronization needs of the large set of applications that EARTH can implement? Or could the EARTH programming model beneet from the implementation of more elaborate synchronization mechanism? In such case, what are the beneets and tradeoos of adding this mechanisms to EARTH? This paper describes the implementation of I-structures under the EARTH execution and architecture model. An I-structure is a data structure that allows for the implementation of a lenient computation model. A read operation can be issued to an element of an I-structure before it is known that the corresponding write operation has produced the value. We also introduce a new parallel kernel based on the Hoppeld Network and demonstrate how the I-structure support on EARTH can be utilized. We nish presenting a complete Threaded-C program to solve the Hoppeld kernel is also presented.
منابع مشابه
The Hop eld-style Network as a Maximal-Clique Graph Machine
The Hop eld-style network, a variant of the popular Hop eld neural network, has earlier been shown to have xed points (stable states) that correspond 1-1 with the maximal cliques of the underlying graph. The network sequentially transforms an initial state (set of vertices) to a nal state (maximal clique) via certain greedy operations. It has also been noted that this network can be used to sto...
متن کاملExtending the power and capacity of constraint satisfaction networks
This work focuses on improving the Hop eld network for solv ing optimization problems Although much work has been done in this area the performance of the Hop eld network is still not satisfactory in terms of valid convergence and quality of solutions We address this issue in this work by combing a new activation function EBA and a new relax ation procedure CR in order to improve the performanc...
متن کاملCompetitive neural network to solve scheduling problems
Most scheduling problems have been demonstrated to be NP-complete problems. The Hop"eld neural network is commonly applied to obtain an optimal solution in various di!erent scheduling applications, such as the traveling salesman problem (TSP), a typical discrete combinatorial problem. Hop"eld neural networks, although providing rapid convergence to the solution, require extensive e!ort to deter...
متن کاملParallel Programming on Hop eld Nets
We describe a simple general purpose condition-action type parallel programming language and its implementation on Hoppeld-type neural networks. A prototype compiler performing the translation has been implemented .
متن کاملSome Theorems Concerning the Free Energy of (Un)Constrained Stochastic Hop eld Neural Networks
General stochastic binary Hop eld models are viewed from the angle of statistical mechanics. Both the general unconstrained binary stochastic Hop eld model and a certain constrained one are analyzed yielding explicit expressions of the free energy. Moreover, conditions are given for which some of these free energy expressions are Lyapunov functions of the corresponding di erential equations. In...
متن کاملSome Afterthoughts on Hopfield Networks
The present paper investigates four relatively independent issues each in one section which complete our knowledge regarding the computational aspects of popular Hop eld nets In Section the computational equivalence of convergent asymmetric and Hop eld nets is proved with respect to the network size In Section the convergence time of Hop eld nets is analyzed in terms of bit representations In S...
متن کامل