Learning to Discretize: Solving 1D Scalar Conservation Laws via Deep Reinforcement Learning

Conservation laws are considered to be fundamental laws of nature. It has broad applications in many fields, including physics, chemistry, biology, geology, and engineering. Solving the differential equations associated with conservation laws is a major branch in computational mathematics. The recent success of machine learning, especially deep learning in areas such as computer vision and natural language processing, has attracted a lot of attention from the community of computational mathematics and inspired many intriguing works in combining machine learning with traditional methods. In this paper, we are the first to view numerical PDE solvers as an MDP and to use (deep) RL to learn new solvers. As proof of concept, we focus on 1-dimensional scalar conservation laws. We deploy the machinery of deep reinforcement learning to train a policy network that can decide on how the numerical solutions should be approximated in a sequential and spatial-temporal adaptive manner. We will show that the problem of solving conservation laws can be naturally viewed as a sequential decision-making process, and the numerical schemes learned in such a way can easily enforce long-term accuracy. Furthermore, the learned policy network is carefully designed to determine a good local discrete approximation based on the current state of the solution, which essentially makes the proposed method a meta-learning approach. In other words, the proposed method is capable of learning how to discretize for a given situation mimicking human experts. Finally, we will provide details on how the policy network is trained, how well it performs compared with some state-of-the-art numerical solvers such as WENO schemes, and supervised learning based approach L3D and PINN, and how well it generalizes.     

Cefoperazone-sulbactam and risk of coagulation disorders or bleeding: a retrospective cohort study

ABSTRACT

Objectives: Limited evidence has suggested that cefoperazone-sulbactam causes coagulation disorders and bleeding.

Methods: The authors conducted a retrospective study to compare patients receiving cefoperazone-sulbactam versus those treated with cefoperazone-tazobactam or ceftazidime. Propensity-score matching was used to explore whether treatment with cefoperazone-sulbactam increased the risk of prothrombin time (PT) prolongation, coagulation disorders, and bleeding, or decreased platelets (PLT).

Results: The cohort included 23,242 patients. Among patients receiving cefoperazone-sulbactam, the risk of PT prolongation, coagulation disorders, decreased PLT, and bleeding was 5.3%, 9.2%, 15.7%, and 4.2%, respectively. Propensity-score matching analyses suggested that cefoperazone-sulbactam increased the risk of PT prolongation (aOR 2.26, 95% CI 1.61–3.18), coagulation disorders (aOR 1.81, 95% CI 1.43–2.30), and decreased PLT (aOR 1.46, 95% CI 1.25–1.72), but not increase bleeding (aOR 1.05, 95% CI 0.79–1.40) compared with ceftazidime. Patients receiving cefoperazone-sulbactam had higher risk of PT prolongation (aOR 1.53, 95% CI 1.11–2.10), coagulation disorders (aOR 1.53, 95% CI 1.21–1.95), but not decreased PLT (aOR 0.93, 95% CI 0.81–1.07) or bleeding (aOR 1.11, 95% CI 0.87–1.42), compared with those receiving cefoperazone-tazobactam.

Conclusion: Cefoperazone-sulbactam may be associated with a higher risk of PT prolongation and coagulation disorders compared with cefoperazone-tazobactam and ceftazidime.     

铜绿假单胞菌噬菌体PaP3生物学特性的研究

目的测定铜绿假单胞菌噬菌体PaP3的最佳感染复数、一步生长曲线和吸附K值及交叉吸附K值等基本生物学特性。方法按照感染复数（MOI）分别为0．0001、0．001、0．01、0．1、1和10加入噬菌体纯培养液和宿主菌，充分裂解细菌后，测定噬菌体滴度；以MOI=10的比例加入噬菌体及宿主菌，进行一步生长实验；纯化PaP3颗粒，免疫家兔，获得抗血清，通过中和反应实验测定PaP3和其抗血清之间的吸附反应常数K值。同时，利用抗血清交叉中和试验确定本室分离的三株噬菌体之间的血清学关系。结果当MOI=0．001时PaP3感染其宿主菌产生的子代噬菌体滴度最高；根据一步生长实验结果绘制一步生长曲线；通过血清交叉中和反应得出不同的吸附常数。结论PaP3最佳感染复数为0．001，感染宿主菌的潜伏期是20min，爆发期是60min，平均爆发量约为31，其抗血清反应的吸附常数K值为262。     

血清S100ββ变化与老年患者腹部手术后认知功能障碍的关系

目的观察老年患者腹部手术后S100ββ蛋白的变化以及术后认知功能障碍(POCD)的发生情况,并探讨二者的关系。方法26例65岁以上的老年患者ASAⅡ~Ⅲ级,行腹部手术。监测术前、术毕、术后6、24、48、72h血清S100ββ的变化,并评定术前及术后1周内的认知功能。结果老年患者血清S100ββ蛋白在术毕最高(P<0·01),术后6h和24h逐渐下降,但术后48h再次上升(P<0·01),术后72h回复至术前水平。26例老年患者腹部手术后1周内有7例发生POCD。POCD组与非POCD组相比,术毕及术后6h血清S100ββ蛋白水平明显增高(P<0·05)。结论老年患者腹部手术后POCD的发生与血清S100ββ蛋白的变化有密切关系。血清S100ββ蛋白可作为评估老年患者术后发生POCD的重要指标。     

影响颅脑外伤术后颅内感染的危险因素分析

目的探讨影响颅脑外伤开颅术后颅内感染的危险因素。方法采用回顾性研究比较分析了912例颅脑损伤术后出现颅内感染与未出现颅内感染组间的差异因素。结果非感染 770例,颅内感染142例(15．6％)。感染类型有无菌性脑膜炎、细菌性脑膜炎、脑室炎及脑室积脓、脑脓肿、硬膜下腔积脓、术区皮下或(和)骨瓣下化脓或合并骨髓炎、切口感染。细菌检出率占感染的27．5％。颅脑外伤术后感染与高温季节、高龄、重度以上损伤、短期内两次以上手术、连续两侧开颅术、长时间 (>5 h)手术、显微外科手术、颅底与后颅凹手术、脑室外引流、皮下或硬膜外积液以及急诊手术等因素相关(P<0．05)。结论对具有上述危险冈素的易感患者应给予更密切的关注和预防性的处理。     

超声在狭窄颈部动脉支架植入术后随访中的应用

目的评价超声在颈部动脉(颈动脉、锁骨下动脉及椎动脉)狭窄血管支架成形术后随访中的应用价值。方法应用超声随访观察47例患者的15枚颈内动脉(ICA)支架、9枚锁骨下动脉(SCA)支架及26枚椎动脉(VA)支架术后1～12个月,观察狭窄血管处支架位置、支架内及远端的收缩期峰值流速(PSV)、支架内有无血栓及再狭窄等情况,判断其疗效。结果支架置入术后超声示狭窄ICA、VA及SCA内径较术前明显增宽,内径分别由术前(2．90±0．76)mm、(2．0±0．38)mm及(4．00±1．46)mm增宽至术后的(4．20±1．78)mm、(3．4±0．51)mm、(6．15±0．98)mm;PSV分别由术前(115±15)cm/s、(181．4±32．5)cm/s、(257．67±66．41)cm/s减低至术后的(75±24)cm/s、(47．15±12．44)cm/s及(98．87±48．63)cm/s。随访中发现2例患者颈内动脉支架狭窄,1例颈内动脉支架闭塞。结论超声是评价颈部动脉狭窄血管支架成形术后疗效的无创、安全可靠的影像学检查方法。     

小颗粒致密低密度脂蛋白氧化特性及对血管内皮细胞脂质过氧化的影响

了解小颗粒致密低密度脂蛋白中抗氧化维生素含量、氧化特性 ,探讨其对血管内皮细胞脂质过氧化的影响。采用二次密度梯度超速离心的方法分离制备小颗粒致密低密度脂蛋白 ,高压液相色谱测定其抗氧化物质维生素A、E和 β 胡萝卜素 ;连续监测小颗粒致密低密度脂蛋白在 2 34nm的吸光度以测定其氧化敏感曲线 ;在血管内皮培养液中分别加入不同浓度的小颗粒致密低密度脂蛋白 ,培养后测定培养液丙二醛的浓度。实验成功分离小颗粒致密低密度脂蛋白 ,与大而轻低密度脂蛋白相比 ,维生素A、E和 β 胡萝卜素等抗氧化物质及总抗氧化能力明显减少 ,只有大而轻低密度脂蛋白的 6 5 %、39%和 2 4 % ,故小颗粒致密低密度脂蛋白氧化的延迟时间只有大而轻低密度脂蛋白的 37.5 % ,但硫代巴比妥酸反应物质值及氧化速率分别是大而轻低密度脂蛋白的 1 .7倍和 1 .2 8倍 ;培养液中丙二醛含量随加入脂蛋白的浓度和作用时间而增高 ,2 4h时小颗粒致密低密度脂蛋白 5 0mg组、1 0 0mg组、1 5 0mg组与相同剂量大而轻低密度脂蛋白组比较丙二醛显著增高 (P <0 .0 5 )。上述结果提示 :小颗粒致密低密度脂蛋白中抗氧化物质及抗氧化能力下降 ,氧化敏感性增高 ,并促进内皮细胞过氧化损伤