Streaming FFT on REDEFINE-v2: An Application-Architecture Design Space Exploration

In this paper we explore an implementation of a high-throughput, streaming application on REDEFINE-v2, which is an enhancement of REDEFINE. REDEFINE is a polymorphic ASIC combining the flexibility of a programmable solution with the execution speed of an ASIC. In REDEFINE Compute Elements are arranged in an 8x8 grid connected via a Network on Chip (NoC) called RECONNECT, to realize the various macrofunctional blocks of an equivalent ASIC. For a 1024-FFT we carry out an application-architecture design space exploration by examining the various characterizations of Compute Elements in terms of the size of the instruction store. We further study the impact by using application specific, vectorized FUs. By setting up different partitions of the FFT algorithm for persistent execution on REDEFINE-v2, we derive the benefits of setting up pipelined execution for higher performance. The impact of the REDEFINE-v2 micro-architecture for any arbitrary N-point FFT (N > 4096) FFT is also analyzed. We report the various algorithm-architecture tradeoffs in terms of area and execution speed with that of an ASIC implementation. In addition we compare the performance gain with respect to a GPP.

Two-timescale Q-learning Algorithms with an Application to Routing in Networks

We propose two variants of the Q-learning algorithm that (both) use two timescales. One of these updates Q-values of all feasible state-action pairs at each instant while the other updates Q-values of states with actions chosen according to the ‘current ’ randomized policy updates. A sketch of convergence of the algorithms is shown. Finally, numerical experiments using the proposed algorithms for routing on different network topologies are presented and performance comparisons with the regular Q-learning algorithm are shown.

Redundancy resolution using a tractrix and its application to real-time simulations of hyper-redundant manipulators, snakes and tying of knots

To realistically simulate the motion of flexible objects such as ropes, strings, snakes, or human hair,one strategy is to discretise the object into a large number of small rigid links connected by rotary or spherical joints. The discretised system is highly redundant and the rotations at the joints (or the motion of the other links) for a desired Cartesian motion of the end of a link cannot be solved uniquely. In this paper, we propose a novel strategy to resolve the redundancy in such hyper-redundant systems.We make use of the classical tractrix curve and its attractive features. For a desired Cartesian motion of the `head'of a link, the `tail' of the link is moved according to a tractrix,and recursively all links of the discretised objects are moved along different tractrix curves. We show that the use of a tractrix curve leads to a more `natural' motion of the entire object since the motion is distributed uniformly along the entire object with the displacements tending to diminish from the `head' to the `tail'. We also show that the computation of the motion of the links can be done in real time since it involves evaluation of simple algebraic, trigonometric and hyperbolic functions. The strategy is illustrated by simulations of a snake, tying of knots with a rope and a solution of the inverse kinematics of a planar hyper-redundant manipulator.

A Simultaneous Deterministic Perturbation Actor-Critic Algorithm with an Application to Optimal Mortgage Refinancing

We develop a simulation-based, two-timescale actor-critic algorithm for infinite horizon Markov decision processes with finite state and action spaces, with a discounted reward criterion. The algorithm is of the gradient ascent type and performs a search in the space of stationary randomized policies. The algorithm uses certain simultaneous deterministic perturbation stochastic approximation (SDPSA) gradient estimates for enhanced performance. We show an application of our algorithm on a problem of mortgage refinancing. Our algorithm obtains the optimal refinancing strategies in a computationally efficient manner