•Gated Recurrent Unit is proposed to extract informative features from raw financial data.•Reward function is designed with risk-adjusted ratio for trading strategies for stable returns in the ...volatile condition.•Two adaptive stock trading strategies are proposed for quantitative stock trading.•The system outperforms the Turtle trading strategy and achieve more stable returns.
The increasing complexity and dynamical property in stock markets are key challenges of the financial industry, in which inflexible trading strategies designed by experienced financial practitioners fail to achieve satisfactory performance in all market conditions. To meet this challenge, adaptive stock trading strategies with deep reinforcement learning methods are proposed. For the time-series nature of stock market data, the Gated Recurrent Unit (GRU) is applied to extract informative financial features, which can represent the intrinsic characteristics of the stock market for adaptive trading decisions. Furthermore, with the tailored design of state and action spaces, two trading strategies with reinforcement learning methods are proposed as GDQN (Gated Deep Q-learning trading strategy) and GDPG (Gated Deterministic Policy Gradient trading strategy). To verify the robustness and effectiveness of GDQN and GDPG, they are tested both in the trending and in the volatile stock market from different countries. Experimental results show that the proposed GDQN and GDPG not only outperform the Turtle trading strategy but also achieve more stable returns than a state-of-the-art direct reinforcement learning method, DRL trading strategy, in the volatile stock market. As far as the GDQN and the GDPG are compared, experimental results demonstrate that the GDPG with an actor-critic framework is more stable than the GDQN with a critic-only framework in the ever-evolving stock market.
The unmanned surface vehicle (USV) has been widely used to accomplish missions in the sea or dangerous marine areas for ships with sailors, which greatly expands protective capability and detection ...range. When USVs perform various missions in sophisticated marine environment, autonomous navigation and obstacle avoidance will be necessary and essential. However, there are few effective navigation methods with real-time path planning and obstacle avoidance in dynamic environment. With tailored design of state and action spaces and a dueling deep Q-network, a deep reinforcement learning method ANOA (Autonomous Navigation and Obstacle Avoidance) is proposed for the autonomous navigation and obstacle avoidance of USVs. Experimental results demonstrate that ANOA outperforms deep Q-network (DQN) and Deep Sarsa in the efficiency of exploration and the speed of convergence not only in static environment but also in dynamic environment. Furthermore, the ANOA is integrated with the real control model of a USV moving in surge, sway and yaw and it achieves a higher success rate than Recast navigation method in dynamic environment.
In the past two decades cooperating with Frank Laboratory of Neutron Physics (FLNP), Joint Institute for Nuclear Research (JINR) measurements of (n,
α
) reaction cross sections for
6
Li,
10
B,
25
Mg,
...39
K,
40
Ca,
54,56,57
Fe,
58
Ni,
63
Cu,
64,67
Zn,
95
Mo,
143
Nd and
147,149
Sm nuclei were performed in the MeV neutron energy region based on the 4.5 MV Van de Graaff accelerator at Peking University. In recent years, our measurements were extended in three aspects. Firstly, measurements were expanded from two-body reactions to three-body reactions such as
10
B (n, t2
α
). Secondly, the neutron energy region was extended from below 8 MeV to 8 - 11 MeV by using the HI-13 tandem accelerator of China Institute of Atomic Energy (CIAE), with which cross sections of
54,56
Fe(n,
α
)
53,51
Cr reactions were measured. Thirdly, based on the newly-built China Spallation Neutron Source (CSNS) Back-n WNS (White Neutron Source), differential and angle-integrated cross sections for
6
Li(n, t) and
10
B(n,
α
) reactions were measured in the neutron energy region from 1 eV to 3 MeV.
The particle and astrophysical xenon experiment III (PandaX-III) is aimed to search for the Neutrinoless Double Beta Decay (NLDBD) using 200-kg radio-pure high-pressure gaseous xenon time projection ...chamber (TPC) with Micromegas detectors at both ends. A small-scale prototype TPC equipped with seven Microbulk Micromegas modules has been developed. Each Micromegas module has 128 anode strip signals to be processed. Highly integrated front-end electronics composed of four front-end cards (FECs) with 1024 channels are designed to read out the charge of Micromegas anode signals digitize the waveform after shaping and send compressed data to the data collection module (DCM). The cornerstone of the front-end electronics is a 64-channel application-specific integrated circuit (ASIC) named AGET, which is based on switched capacitor arrays (SCAs). According to the test results, the integral nonlinearity (INL) of the front-end electronics is less than 1%, and the noise of each readout channel with the input floating is less than 0.9 fC on the condition of 1-<inline-formula> <tex-math notation="LaTeX">\mu \text{s} </tex-math></inline-formula> peaking time and 1-pC dynamic range. Joint tests of front-end electronics with the prototype TPC were carried out using the radioactive sources 137 Cs and 241 Am. The hit map of the Micromegas modules and the energy spectrum have been reconstructed successfully, and the results are satisfying.
The back-streaming white neutron source (Back-n) at China Spallation Neutron Source (CSNS), which has excellent energy spectrum and good time resolution, provides superior conditions for neutron ...induced light charged particle tracking experiment. To track light charged particles, CSNS has proposed a project named Multi-purpose Time Projection Chambers (MTPC) and completed a 1519-channel prototype. A multi-channel readout system based on Switched Capacitor Arrays (SCA) waveform sampling ASIC was designed to meet the requirements of the prototype MTPC. This paper presents the structure of this readout system and the results of the beam test with MTPC at CSNS. Time of Flight spectrum of neutron beam and 3D-track of neutron induced light charged particles were successfully reconstructed. These test results prove the feasibility of large-scale MTPC.
A 256-channel time-of-flight electronics system has been developed for a facility called "Back-n white neutron source (WNS)" in China Spallation Neutron Source (CSNS). This paper presents the ...structure and performance of the electronics system and test results at CSNS-WNS. A piece of lithium glass scintillator, coupled with a 256-channel multi-anode photomultiplier tube, is placed at the center of the neutron beam line as the detector. A 256-channel readout electronics system, including one front-end board (FEB), four time-to-digital converter (TDC) cards and one clock distribution module (CDM), measures the arrival time of each signal from the detector. Signals are sent to FEB from the detector through 15-cm cables and discriminated on FEB. FEB drives differential signals over 2-m cables to the TDC cards. Each TDC card has 64 channels, using a field-programmable gate array (FPGA) to measure the arrival time of each signal. The CDM supplies clock signals to the TDC cards to make them work synchronously. The TDC cards and CDM are housed in one peripheral component interconnect (PCI) eXtensions for Instrumentation chassis. The TDC cards achieve a timing precision of 3.5 ns. The electronics system meets the requirement of the experiment and is currently being used in CSNS.
A Multi-purpose Time Projection Chamber prototype based on a resistive Micromegas technique has been proposed, designed and fabricated at China Spallation Neutron Source to measure the cross sections ...of neutron-induced light-charged particle emission reactions and fission reactions at the Back-n white neutron source. For the requirements of working stability with high voltage and high granularity of readout, the prototype is fabricated by applying a resistive Micromegas with high-density pixelated anode plane. In order to fully understand the process of charge spreading and signal induction on the anode plane covered by a resistive layer, a simulation method is developed and the energy correction function is then obtained, which can be used in the further energy correction in the experimental data analysis. Measured with a 55Fe X-ray source, the uniformity of gain and avalanche gap are specially analyzed by combining the simulated gas avalanche parameters with fitting of experimental data in which the parallel plate avalanche model is adopted.