SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
The objective of this presentation for the HPC mini showcase was to find effectiveness of erasure encoding done in parallel and then discover a possible solution to more cost effective storage.
In Fine-Grained Cycle Sharing (FGCS) systems, machine owners voluntarily share their unused CPU cycles with guest jobs, as long as their performance degradation is tolerable. However, unpredictable evictions of guest jobs lead to fluctuating completion times. Checkpoint-recovery is an attractive mechanism for recovering from such “failures”. Today’s FGCS systems often use expensive, high-performance dedicated checkpoint servers. However, in geographically distributed clusters, this may incur high checkpoint transfer latencies. Here we present a distributed checkpointing system called FALCON that uses available disk resources of the FGCS machines as shared checkpoint repositories. However, an unavailable storage host may lead to loss of checkpoint data. Therefore, we model the failures of a storage host and develop a prediction algorithm for choosing reliable checkpoint repositories. We experiment with FALCON in the university-wide Condor testbed at Purdue and show improved and consistent performance for guest jobs in the presence of irregular resource availability.
This viewgraph presentation describes a prioritized Luby Transform coding scheme that seeks to decode high priority data with high reliability, when decoders fail.
In this paper, we present two encoding methods for block-circulant LDPC codes. The first is an iterative encoding method based on the erasure decoding algorithm, and the computations required are well organized due to the block-circulant structure of the parity check matrix. The second method uses block-circulant generator matrices, and the encoders are very similar to those for recursive convolutional codes. Some encoders of the second type have been implemented in a small Field Programmable Gate Array (FPGA) and operate at 100 Msymbols/second.
In this paper we design high rate protograph based LDPC codes suitable for binary erasure channels. To simplify the encoder and decoder implementation for high data rate transmission, the structure of codes are based on protographs and circulants. These LDPC codes can improve data link and network layer protocols in support of communication networks. Two classes of codes were designed. One class is designed for large block sizes with an iterative decoding threshold that approaches capacity of binary erasure channels. The other class is designed for short block sizes based on maximizing minimum stopping set size. For high code rates and short blocks the second class outperforms the first class.
We study the robustness of quantum error correction in a one-parameter ensemble of codes generated by the Brownian SYK model, where the parameter quantifies the encoding complexity. The robustness of error correction by a quantum code is upper bounded by the “mutual purity” of a certain entangled state between the code subspace and environment in the isometric extension of the error channel, where the mutual purity of a density matrix ρAB is the difference $\mathcal{F}$ p ($A : B$) ≡ $\mathrm{T}$r $p^{2}_{AB}$ - $\mathrm{T}$r $p^{2}_{A}$ $\mathrm{T}$r $p^{2}_{B}$. We show that when the encoding complexity is small, the mutual purity is O(1) for the erasure of a small number of qubits (i.e., the encoding is fragile). However, this quantity decays exponentially, becoming O(1/N) for O(log N) encoding complexity. Further, at polynomial encoding complexity, the mutual purity saturates to a plateau of O(e -N ). We also find a hierarchy of complexity scales associated to a tower of subleading contributions to the mutual purity that quantitatively, but not qualitatively, adjust our error correction bound as encoding complexity increases. In the AdS/CFT context, our results suggest that any portion of the entanglement wedge of a general boundary subregion A with sufficiently high encoding complexity is robustly protected against low-rank errors acting on A with no prior access to the encoding map. From the bulk point of view, we expect such bulk degrees of freedom to be causally inaccessible from the region A despite being encoded in it.
A method of constructing certain low-density parity-check (LDPC) codes by use of relatively simple loop-free coding modules has been developed. The subclasses of LDPC codes to which the method applies includes accumulate-repeat-accumulate (ARA) codes, accumulate-repeat-check-accumulate codes, and the codes described in Accumulate-Repeat-Accumulate-Accumulate Codes (NPO-41305), NASA Tech Briefs, Vol. 31, No. 9 (September 2007), page 90. All of the affected codes can be characterized as serial/parallel (hybrid) concatenations of such relatively simple modules as accumulators, repetition codes, differentiators, and punctured single-parity check codes. These are error-correcting codes suitable for use in a variety of wireless data-communication systems that include noisy channels. These codes can also be characterized as hybrid turbolike codes that have projected graph or protograph representations (for example see figure); these characteristics make it possible to design high-speed iterative decoders that utilize belief-propagation algorithms. The present method comprises two related submethods for constructing LDPC codes from simple loop-free modules with circulant permutations. The first submethod is an iterative encoding method based on the erasure-decoding algorithm. The computations required by this method are well organized because they involve a parity-check matrix having a block-circulant structure. The second submethod involves the use of block-circulant generator matrices. The encoders of this method are very similar to those of recursive convolutional codes. Some encoders according to this second submethod have been implemented in a small field-programmable gate array that operates at a speed of 100 megasymbols per second. By use of density evolution (a computational- simulation technique for analyzing performances of LDPC codes), it has been shown through some examples that as the block size goes to infinity, low iterative decoding thresholds close to channel capacity limits can be achieved for the codes of the type in question having low maximum variable node degrees. The decoding thresholds in these examples are lower than those of the best-known unstructured irregular LDPC codes constrained to have the same maximum node degrees. Furthermore, the present method enables the construction of codes of any desired rate with thresholds that stay uniformly close to their respective channel capacity thresholds.
The original Luby Transform (LT) coding scheme is extended to account for data transmissions where some information symbols in a message block are more important than others. Prioritized LT codes provide unequal error protection (UEP) of data on an erasure channel by modifying the original LT encoder. The prioritized algorithm improves high-priority data protection without penalizing low-priority data recovery. Moreover, low-latency decoding is also obtained for high-priority data due to fast encoding. Prioritized LT codes only require a slight change in the original encoding algorithm, and no changes at all at the decoder. Hence, with a small complexity increase in the LT encoder, an improved UEP and low-decoding latency performance for high-priority data can be achieved. LT encoding partitions a data stream into fixed-sized message blocks each with a constant number of information symbols. To generate a code symbol from the information symbols in a message, the Robust-Soliton probability distribution is first applied in order to determine the number of information symbols to be used to compute the code symbol. Then, the specific information symbols are chosen uniform randomly from the message block. Finally, the selected information symbols are XORed to form the code symbol. The Prioritized LT code construction includes an additional restriction that code symbols formed by a relatively small number of XORed information symbols select some of these information symbols from the pool of high-priority data. Once high-priority data are fully covered, encoding continues with the conventional LT approach where code symbols are generated by selecting information symbols from the entire message block including all different priorities. Therefore, if code symbols derived from high-priority data experience an unusual high number of erasures, Prioritized LT codes can still reliably recover both high- and low-priority data. This hybrid approach decides not only "how to encode" but also "what to encode" to achieve UEP. Another advantage of the priority encoding process is that the majority of high-priority data can be decoded sooner since only a small number of code symbols are required to reconstruct high-priority data. This approach increases the likelihood that high-priority data is decoded first over low-priority data. The Prioritized LT code scheme achieves an improvement in high-priority data decoding performance as well as overall information recovery without penalizing the decoding of low-priority data, assuming high-priority data is no more than half of a message block. The cost is in the additional complexity required in the encoder. If extra computation resource is available at the transmitter, image, voice, and video transmission quality in terrestrial and space communications can benefit from accurate use of redundancy in protecting data with varying priorities.
We study concept-level forgetting in pretrained vision models: removing an entire semantic category so the system no longer recognizes that object in unseen images and contexts, rather than merely forgetting specific training examples. Prior work either applies blunt global projections or fine-tunes parameters, which can introduce collateral damage to unrelated features, add compute, and become unstable as forgetting strength increases. We introduce Contrastive Subnet Erasure (CSE), a training-free, encoder-centric edit that targets a compact set of channels most responsible for the class and attenuates them in a calibrated manner. The modification is algebraically folded into the subsequent layer, yielding no inference-time overhead and leaving task heads unchanged. To evaluate whether forgetting generalizes beyond the data used to specify the class, we introduce a cross dataset protocol in which the class is defined on a source dataset and performance is measured on a disjoint target dataset drawn from a different distribution with no shared images. This setup tests whether the model still fails to recognize the object when it looks different or appears in new scenes, and it helps avoid overfitting to patterns in the source dataset. Across CIFAR 10, CIFAR 100, and ImageNet under this protocol, CSE achieves stronger forgetting of the target class while better preserving non target utility than existing baselines in both single class and multi class settings. Overall, CSE provides a simple, stable, and deployment-ready mechanism for class-level unlearning in vision.
The inherent aspects and flaws surrounding space based communication is technically described and the math surrounding encoding and decoding LT Codes is examined. Utilizing LT codes as a means of reducing packet erasures due to corrupted packets on an RF link can result in higher voice quality. PESQ-MOS measure was used to analyze voice degradation over space links tested for LT codec size and number of 10ms per packet.Extensions utilizing LT codes to improve the packet erasure performance and combining the use of ASR could provide for a solid means of identifying the benefit in terms of intelligibility of voice communications in space-based networks
Deep space communications over noisy channels lead to certain packets that are not decodable. These packets leave gaps, or bursts of erasures, in the data stream. Burst erasure correcting codes overcome this problem. These are forward erasure correcting codes that allow one to recover the missing gaps of data. Much of the recent work on this topic concentrated on Low-Density Parity-Check (LDPC) codes. These are more complicated to encode and decode than Single Parity Check (SPC) codes or Reed-Solomon (RS) codes, and so far have not been able to achieve the theoretical limit for burst erasure protection. A block interleaved maximum distance separable (MDS) code (e.g., an SPC or RS code) offers near-optimal burst erasure protection, in the sense that no other scheme of equal total transmission length and code rate could improve the guaranteed correctible burst erasure length by more than one symbol. The optimality does not depend on the length of the code, i.e., a short MDS code block interleaved to a given length would perform as well as a longer MDS code interleaved to the same overall length. As a result, this approach offers lower decoding complexity with better burst erasure protection compared to other recent designs for the burst erasure channel (e.g., LDPC codes). A limitation of the design is its lack of robustness to channels that have impairments other than burst erasures (e.g., additive white Gaussian noise), making its application best suited for correcting data erasures in layers above the physical layer. The efficiency of a burst erasure code is the length of its burst erasure correction capability divided by the theoretical upper limit on this length. The inefficiency is one minus the efficiency. The illustration compares the inefficiency of interleaved RS codes to Quasi-Cyclic (QC) LDPC codes, Euclidean Geometry (EG) LDPC codes, extended Irregular Repeat Accumulate (eIRA) codes, array codes, and random LDPC codes previously proposed for burst erasure protection. As can be seen, the simple interleaved RS codes have substantially lower inefficiency over a wide range of transmission lengths.
The dramatic increase in the scale of current and planned high-end HPC systems is leading new challenges, such as the growing costs of data movement and IO, and the reduced mean time between failures (MTBF) of system components. In-situ workflows, i.e., executing the entire application workflows on the HPC system, have emerged as an attractive approach to address data-related challenges by moving computations closer to the data, and staging-based frameworks have been effectively used to support in-situ workflows at scale. However, the resilience of these staging-based solutions has not been addressed, and they remain susceptible to expensive data failures. Furthermore, naive use of data resilience techniques such as n-way replication and erasure codes can impact latency and/or result in significant storage overheads. In this article, we present CoREC, a scalable and resilient in-memory data staging runtime for large-scale in-situ workflows. CoREC uses a novel hybrid approach that combines dynamic replication with erasure coding based on data access patterns. It also leverages multiple levels of replications and erasure coding to support diverse data resiliency requirements. Furthermore, the article presents optimizations for load balancing and conflict-avoiding encoding, and a low overhead, lazy data recovery scheme. We have implemented the CoREC runtime and have deployed with the DataSpaces staging service on leadership class computing machines and present an experimental evaluation in the article. Here, the experiments demonstrate that CoREC can tolerate in-memory data failures while maintaining low latency and sustaining high overall storage efficiency at large scales.
We study the quantum error correction properties of the black hole interior in a toy model for an evaporating black hole: Jackiw-Teitelboim gravity entangled with a non-gravitational bath. After the Page time, the black hole interior degrees of freedom in this system are encoded in the bath Hilbert space. We use the gravitational path integral to show that the interior density matrix is correctable against the action of quantum operations on the bath which (i) do not have prior access to details of the black hole microstates, and (ii) do not have a large, negative coherent information with respect to the maximally mixed state on the bath, with the lower bound controlled by the black hole entropy and code subspace dimension. Thus, the encoding of the black hole interior in the radiation is robust against generic, low-rank quantum operations. For erasure errors, gravity comes within an O (1) distance of saturating the Singleton bound on the tolerance of error correcting codes. For typical errors in the bath to corrupt the interior, they must have a rank that is a large multiple of the bath Hilbert space dimension, with the precise coefficient set by the black hole entropy and code subspace dimension.
A VLSI coder is presented that can function either as an encoder or decoder for Reed-Solomon codes. VLSI is one approach to implementing high-performance Reed-Solomon decoders. There are three VLSI technologies that could be used: gate arrays, standard cells, and full custom. The first two approaches are relatively easy to implement, but are limited in both performance and density. Full-custom VLSI is used to achieve both circuit density and speed, and allows control of the amount of interconnect. Speed, which is a function of capacitance, which is a function of interconnect, is an important parameter in high-performance VLSI. A single 8.2 mm x 8.4 mm, 200,000 transistor CMOS chip implementation of the Reed-Solomon code required by the Hubble Space Telescope is reported. The chip features a 10-MHz sustained byte rate independent of error pattern. The 1.6-micron CMOS integrated circuit has complete decoder and encoder functions and uses a single data/system clock. Block lengths up to 255 bytes and shortened codes are supported with no external buffering. Erasure corrections and random error corrections are supported with programmable correction of up to 10 symbol errors. Correction time is independent of error pattern and the number of errors in the incoming message.
The design of quantum hardware that reduces and mitigates errors is essential for practical quantum error correction (QEC) and useful quantum computation. To this end, we introduce the circuit-Quantum Electrodynamics (QED) dual-rail qubit in which our physical qubit is encoded in the single-photon subspace, { | 01 〉 , | 10 〉 } , of two superconducting microwave cavities. The dominant photon loss errors can be detected and converted into erasure errors, which are in general much easier to correct. In contrast to linear optics, a circuit-QED implementation of the dual-rail code offers unique capabilities. Using just one additional transmon ancilla per dual-rail qubit, we describe how to perform a gate-based set of universal operations that includes state preparation, logical readout, and parametrizable single and two-qubit gates. Moreover, first-order hardware errors in the cavities and the transmon can be detected and converted to erasure errors in all operations, leaving background Pauli errors that are orders of magnitude smaller. Hence, the dual-rail cavity qubit exhibits a favorable hierarchy of error rates and is expected to perform well below the relevant QEC thresholds with today’s coherence times.
This paper reports a single 8.2mm by 8.4mm, 200,000 transistor CMOS chip implementation of the Reed Solomon code required by the Space Telescope. The chip features a 10 MHz sustained byte rate independent of error pattern. The 1.6 micron CMOS integrated circuit has complete decoder and encoder functions and uses a single data/system clock. Block lengths up to 255 bytes as well as shortened codes are supported with no external buffering. Erasure corrections as well as random error corrections are supported with programmable corrections of up to 10 symbol errors. Correction time is independent of error pattern and the number of errors.
No abstract provided.