Question

我想知道Apache Math Commons库中是否有一个与逐个元素相乘的特性，类似于MATLAB中使用的特性，结果矩阵#Binomial option pricing (2) #We can expand the tree to more than 3rd depth. #U = exp(volatility) #D = exp(-volatility) #p = 0.5 (We have the equal chance of making or losing money) #Risk free rate = 0.02 => exp(0.02) #For those who are not familiar with data structure, I deliberately used just array. #I'll make a new code for those who are familiar with tree data structure library(igraph) #Define the variable depth<-3 #How many steps (tree depth) do you want to make rate <-0.02 #Risk Free rate volatility <- 0.35 exercise_price <- 35 stock_price <- 47.5 upside_probability<-(exp(rate)-exp(-volatility))/(exp(volatility)-exp(-volatility)) rate <- exp(rate) total_node<-2^depth-1 #Total number of node G <- graph.tree(n=total_node, children=2) #I'll make a graph whose nodes are 7, and each node has two children stock_tree <- (1:total_node) stock_tree[1]<-stock_price tree_traverse <- 2^(depth-1) -1 for(i in 1:tree_traverse) { #We are going to use mathematical trick to represent tree. stock_tree[i*2] <- stock_tree[i] * exp(volatility) stock_tree[i*2 + 1] <- stock_tree[i] * exp(-volatility) } V(G)$name <- round(stock_tree) #Name of the tree lay <- layout.reingold.tilford(G) #It's tree. You can try other shape with other layout options plot(G, layout=lay, vertex.size=15, edge.arrow.size=0.1) #Draw the tree. #As opposed to the stock price, the option pricing starts out with end nodes (bottom nodes) #I already explained the logic. Just follow it from one by one. option_price<-(1:total_node) bottom_node<-tree_traverse + 1 #In order to value the option, we need to calculate bottom line first. for(i in bottom_node:total_node) { after_option <- stock_tree[i] - exercise_price if( after_option >0 ) { option_price[i] <- after_option } else { option_price[i] <- 0 } } #Discount it back to current time while considering the probabilty of up and down for(i in tree_traverse:1) { option_price[i]<-upside_probability*option_price[i*2] option_price[i]<-option_price[i]+(1-upside_probability)*option_price[i*2+1] option_price[i]<-option_price[i]/rate } V(G)$name <- round(option_price) plot(G, layout=lay, vertex.size=15, edge.arrow.size=0.1)等于矩阵{{1}中的每个值}乘以矩阵C中的对应值。我想避免编写自己的版本，因为我知道线性代数包是针对这些类型的操作进行了高度优化的，如果已经存在（已经优化），我不想在我自己的实现上失去性能。

在MATLAB中：A和B的维度必须等于C = A.*B。

Answer 1

我用过Apache的RealVectors数学ebeMultiplication。我不知道是否有一个等效的矩阵函数，我检查过，我没有找到。至于载体：

    RealVector output = new ArrayRealVector(o); // o and y are double[]
    RealVector expected = new ArrayRealVector(y);
    RealVector errors = expected.subtract(output);    
    RealVector delta = errors.ebeMultiply(output);

Java Apache Math Commons Element-Wise矩阵乘法

1 个答案: