| 1. |
- Zhou, Yueting, et al.
(författare)
-
Self-calibrated NICE-OHMS based on an asymmetric signal : theoretical analysis and experimental validation
- 2023
-
Ingår i: Optics Express. - : Optical Society of America. - 1094-4087. ; 31:17, s. 27830-27842
-
Tidskriftsartikel (refereegranskat)abstract
- As an ultra-sensitive detection technique, the noise-immune cavity enhanced optical heterodyne molecular spectroscopy (NICE-OHMS) technique has great potential for assessment of the concentration of trace gases. To determine gas concentrations at the ppt or lower level with high accuracy, it is desirable that the technique exhibits self-calibration (or calibration-free) capabilities. Although being sensitive, NICE-OHMS has so far not demonstrated any such ability. To remedy this, this paper provides a self-calibrated realization of NICE-OHMS that is based on a switching of the feedback target of the DeVoe-Brewer (DVB) locking procedure from the modulation frequency of the frequency modulation spectroscopy (FMS) to the cavity length, which creates an asymmetrical signal whose form and size can be used to unambiguously assess the gas concentration. A comprehensive theoretical model for self-calibrated NICE-OHMS is established by analyzing the shift of cavity modes caused by intracavity absorption, demonstrating that gas absorption information can be encoded in both the laser frequency and the NICE-OHMS signal. To experimentally verify the methodology, we measure a series of dispersion signals under different levels of absorbance using a built experimental setup. An instrument factor and the partial pressure are obtained by fitting the measured signal through theoretical expressions. Our results demonstrate that fitted values are more accurate for higher partial pressures than for lower. To improve on the accuracy at low partial pressures, it is shown that the instrument factor obtained by fitting the signal at large partial pressures (in this case, above 7.8 µTorr) can be set to a fixed value for all fits. By this, the partial pressures can be assessed with a relative error below 0.65%. This technique has the potential to enable calibration-free ultra-sensitive gas detection.
|
|
| 2. |
- Shi, Hang, et al.
(författare)
-
A Free Stale Synchronous Parallel Strategy for Distributed Machine Learning
- 2019
-
Ingår i: BDE 2019. - New York, NY, USA : Association for Computing Machinery (ACM). ; , s. 23-29
-
Konferensbidrag (refereegranskat)abstract
- With the machine learning applications processing larger and more complex data, people tend to use multiple computing nodes to execute the machine learning tasks in distributed way. However, in real world, people always encounter a problem that a few nodes in system exhibit poor performance and drag down the efficiency of the whole system. In existing parallel strategies such as bulk synchronous parallel and stale synchronous parallel, these nodes with poor performance may not be monitored and found out in time. To address this problem, we proposed a free stale synchronous parallel (FSSP) strategy to free the system from the negative impact of those nodes. Our experimental results on some classical machine leaning algorithms and datasets demonstrated that FSSP strategy outperformed other existing parallel computing strategy.
|
|
