森谱的界

设λ_k(F)是树或者森的第k大特征值,[x]是不超过x的最大整数,q是F的边独立数.本文证明了:对于1≤k≤[(q 1)/2]有λ_k(F)≥1,并且这个下界是最好可能的;对于1≤i≤[q/2],若q为偶数,则有λ[(q 1)/2] i(F)≥2cos((2iπ)/(4i 1)),若q为奇数,则有λ_([(q 1)/2] i)(F)≥2cos(((2i 1)π)/(4i 3)),     

Covering projective planar graphs with three forests

It is known that all planar graphs and all projective planar graphs have an edge partition into three forests. Gonçalves proved that every planar graph has an edge partition into three forests, one having maximum degree at most four [5]. In this paper, we prove that every projective planar graph has an edge partition into three forests, one having maximum degree at most four.     

Predicting USCS soil classification from soil property variables using Random Forest

Soil classification systems are widely used for quickly and easily summarizing soil properties and provide a shorthand method of communication between scientists, engineers, and end-users. Two of the most widely used soil classification systems are the United States Department of Agriculture (USDA) textural soil classification system and the Unified Soil Classification System (USCS). Unfortunately, not all soil map units are classified according to the USDA or USCS systems, and previous attempts to provide a crosswalk table have been inconsistent. Random Forest machine learning model was used to create a USCS prediction model using USDA soil property variables. Important variables for predicting USCS code from available soil properties were USDA soil textures, percent organic material, and available water storage. Prediction error rates less than 2% were achieved compared to error rates of approximately 40% using crosswalk methods.     

随机森林结合CatBoost的近红外光谱药品鉴别

药品质量关乎人民健康和国家命脉,随着社会经济的飞速发展对药品质量的快速、有效鉴别具有极其重要的作用。光谱分析技术具有较高的准确性、较快的分析速度且对样品不存在污染等突出优点,广泛应用在化工、石油以及医药等重要的领域。为了解决传统药品鉴别模型存在的鉴别精度低、鉴别速度不能满足实际需求且鉴别模型稳定性差的问题,采用光谱仪采集药品的近红外光谱数据达到对药品无污染鉴别的目的。结合随机森林和CatBoost对药品进行分类鉴别,以实现快速且准确的鉴别。首先采用随机森林(RF)对光谱仪采集的光谱数据进行有效特征波长的筛选,从而将药品光谱数据中的无关波长去除、筛选出最能表征样品属性的特征波长,然后以极限学习机(ELM)作为CatBoost的弱分类器分析筛选的特征波长对药品的属性鉴别。由于ELM仅只含有一个隐含层且无需多次迭代寻优保证了鉴别模型运行速度更快,CatBoost通过集成弱分类器以改善模型鉴别准确性。为对所提出的药品鉴别模型性能进行有效评估,采用随机抽取训练集的方式构造不同规模药品光谱数据并分别上进行独立实验且以10次运行结果的均值作为其最终结果,并通过与CatBoost、持向量机(SVM)、...     

Effects of timber skidding on chemical characteristics of herbaceous cover, forest floor and topsoil on skidroad in an oak (Quercus petrea L.) forest

This paper examines the effects of timber harvesting by skidding on some soil properties (sand, silt, clay, pH, organic carbon, bulk density and compaction), herbaceous cover (unit mass) and forest floor (unit weight) properties. Also N (%), P, K, Na, Ca, Mg, Fe, Zn, Cu and Mn (ppm) were determined in all herbaceous cover, forest floor and two soil depth (0–5 cm and 5–10 cm) on skidroad of an oak (Quercus petrea L.) stand in Istanbul Belgrad Forest – Turkey. In this study, obtained results are; the forest floor and the herbaceous cover amount on the skidroad have been found considerably lower than undisturbed area. There were some crucial changes in the characteristics of the soil which has been investigated down to 10 cm depth. Soil bulk density was found quite high in the samples taken from the skidroad subject to compaction compared to the ones on the undisturbed area. Nevertheless, no important difference had been detected between the skidroad and the undisturbed area at both soil depths in terms of organic carbon contents. Moreover, the soil acidity (pH) values showed noteworthy differences in the analysis of soil samples taken from both soil depths on the skidroad and on the undisturbed area. Fe and Cu contents of herbaceous samples on skidroad were significantly higher than undisturbed area. Forest floor on skidroad had significantly higher K content, and significantly lower Zn, Mn and N content compared to undisturbed area. P, Fe, Zn and Mn contents were found significantly lower in 0–5 cm soil depth on skidroad than undisturbed area. In 5–10 cm soil depth, concentrations of N, P, Fe, Zn and Mn were significantly lower, while Mg and Cu contents were significantly higher than undisturbed area. Results indicate that long-term harvest using skidding techniques on these sites had adversely affected soil cation concentrations, physical soil conditions and mass of herbaceous cover and forest floor.     

刚竹毒蛾危害下的毛竹叶片光谱特征及虫害等级检测研究

探讨刚竹毒蛾危害下的毛竹叶片光谱特征可为建立竹林生态安全监测体系提供重要的理论指导。相比于传统的多光谱数据,高光谱遥感能够准确探测不同刚竹毒蛾危害等级间寄主光谱的细微变化。然而,当前有关此方面的研究甚少,其寄主的光谱变化机理还有待进一步总结。为此,基于实测的552条竹叶光谱,分析了健康、受刚竹毒蛾危害、小年叶片之间的光谱差异,选择可反映其健康状况的特征变量,并利用XGBoost模型建立了叶片尺度的刚竹毒蛾危害检测模型。研究结果显示：（1）随着虫害等级的上升,受害叶片在可见光范围内的反射率逐渐出现“绿低红高”的特征,其近红外波段的反射率则不断降低,而短波红外的反射率则明显高于健康叶片,尤其在两个水汽吸收波段（1 450和1 940 nm）的差异最为明显;（2）小年叶片于可见光-近红外波段的反射率显著高于健康、受害叶片;（3）根据不同受害类型叶片的光谱特征可知,较之健康叶片,缺刻型叶片的光谱并未出现太大的变化,红褐色病斑型叶片在红光波段的反射率出现了一定程度的上升,灰白色病斑型叶片则已经完全失去了植被的基本光谱特征;（4）根据XGBoost模型给出的变量重要性排序可知,各特征变量的贡献度依次为PRI（光化学反射率指数）＞FDVI_{576, 717}（植被健康程度评估指数）＞NPCI（归一化色素叶绿素指数）＞DSWI（疾病水胁迫指数）＞VOG 1（红边指数1）＞RVSI（红边植被胁迫指数）＞NDWI（归一化差值水分指数）;（5）模型对刚竹毒蛾危害识别的总平均精度为74.39%,其中健康叶片的识别精度达到了94.55%,轻度危害叶片为74.93%,重度危害为84.12%,小年叶片则为71.10%,而中度危害叶片的识别精度较差,仅为33.48%。     

热红联用技术在森林火灾烟气测试中的应用

探索了一套基于热重分析仪-红外光谱分析仪联用（热红联用）技术的森林火灾烟气测试方法.该方法在对森林可燃物燃烧的热失重过程进行热重分析的同时,同步地对热解气体成分进行傅里叶红外光谱（FTIR）分析,从而实现了对燃烧过程诸因素的跟踪分析.