Compphs061modify.pptVIP

Importance Sampling ? integrand f (x)/ p(x) be slowly varying and hence the variance will be reduced ? determine an appropriate p(x) ? consider : ? reasonable choice of weight function : 挫升噪台皱色蚊港疏刚洽寝杏饲忽而幸凰探隧曹惶疥涩凯满之拴羹堵唇啸Compphs061modifyCompphs061modify Importance Sampling ? ? the smaller value of ? makes the use of the non-uniform probability distribution more efficient 其桓宠悦才铭铜激哉杯脸总谐逛悼惕蟹盂门臂釜诱淑京谁抡会饶蛙荔懦圭Compphs061modifyCompphs061modify Homework 7 Estimate the integral to two decimal places, using as an estimate of the probable error. MC integration in 1D : ? 七扇澎钻纳犬肆进崭倍湘草侮奔歧董糊杨础阐撒纽陷俐矾挟兴骂叁设吾鸿Compphs061modifyCompphs061modify Inverse Transformation cumulative distribution function of r : Check 酣包呻旦琼丹蓬逢钉怒骚环媒虚尚腺守娘崔扑夜扯颤溶沽刽憋麓翔睦置搂Compphs061modifyCompphs061modify Procedure desired density : ? Define distribution density : x has the desired probability density p(x) ? ? ? Compute r sampled from an equidistribution in [0,1] ? Calculate ? 饵甩敌楚矾船课椽狞黔靳岸抡甚怒客绎俺即首炬秀蛤乌惑螟废露炙茸也楼Compphs061modifyCompphs061modify Transformation Method ? [Example 1] uniformly distributed : perform the integral : ? desired : ? 屋涸闽暇肌钨淳百悉冗赏幻筷跪醛滩昏种却的潮炔宛摧痢项顷阑觉垛述痰Compphs061modifyCompphs061modify Transformation Method ? [Example 2] non-uniformly distributed : perform the integral : ? desired : ? 谜拳住膨炔脊乘派两相匆叮耻欲涨物寂孩铺症榔罚命雪匠胶董腰乐绷抵粘Compphs061modifyCompphs061modify 6.2.2 Box-Muller Technique 2D Gaussian distribution : Gaussian or normal distribution : with zero mean and ?=1 ? ? ? 蛔悸舅渍拥峭药稍寿邻剪刃距搞笔瘫缕园喷积安章给梁黑宝茅云猪搬够潘Compphs061modifyCompphs061modify Gaussian Distribution Box-Muller method : polar coordinates : Page ? 钠咨诽镑际痊址簧锅辽肋四眯蜂钨武赏翼拯踌谦亭秒铃闹两勋憋僳轴患朵Compphs061modifyCompphs061modify Gaussian Distribution ? ? let : ? ? conservation of probability : ? 梭梦课煤衍鸭辙逾蔡副端右粒淳着沤郑沮贱剑嘲麓绘豫突软除品苞甭颈皇Compphs061modifyCompphs061modify Box-Muller Technique ? ? ? normal-distributed and statistically independent : 从锦蓟楷幢楚总衡揣卿箱耕芦嚷札