Python错误：“ numpy.float64”对象没有属性“ append”

Question

I am trying to run a simulation in which I do the following:

1. Take 2000 random samples from a uniform distribution between 0 and 1
2. Calculate the difference, du, between any sample and the one chosen before it
3. Use that difference to calculate r=EXP(-a*du)
4. Compare another random sample z to the calculated value of r
5. Create a list of the random samples for which r>z, and discard all others
6. Repeat this process until 2000 samples have been "accepted"

Here is what I have thus far. When I run this code, I receive the error message "'numpy.float64' object has no attribute 'append'". Any ideas on how to fix this issue?

import numpy as np                                                          
import matplotlib.pyplot as plt                                             
import math                                                                 

NP=np.random.uniform(0,1,size=(2000,))                                      
a=np.linspace(0.1,2,num=20)                                                 

for i in range(len(a)):                                                     
    dr = []                                                                 
    du = []                                                                 
    for j in range(1999):                                                   
        du=N[j+1]-N[j]                                                      
        r=math.exp(-a[i]*du)                                                
        z=np.random.uniform(0,1)                                            
        if r>z:                                                             
            du.append(N[j+1])                                               
            dr.append(r)

Answer 1

You are using the same variable name, du , for two distinct concepts (the container for all the valid du values, and each individual du value per iteration).

Change your code to du_values = [] and du_values.append(N[j + 1]) and it should work.

As an aside, there appears to be a typo in your code - you define the original array as NP , but later refer to it as N .

Answer 2

Note that since you did not provide any code, the code I am going to show you is not FOR you but code I have used and SOUNDS like you could easily adopt to to your needs.

 # this will simulate 1000 different combinations of my portfolio
    for x in range(1000):
        weights = np.random.random(len(tickers))
        weights /= np.sum(weights)
        portfolio_returns.append(np.sum(weights * log_returns.mean()) * 250)
        portfolio_volatilities.append(np.sqrt(np.dot(weights.T, np.dot(log_returns.cov() * 250, weights))))

What this code is doing is short is 1000x it is creating random numbers and weights for my data points, and then appending the result.The code won't be uniform as it is random, but if you really wanted uniform you could simply step through the increments. However I don't think you want uniform, but rather a large enough sample size to not have it be thrown off by outliers.

A way to compare the results would be something like this.

simple_return = (mydata / mydata.shift(1)) - 1

Comparing random samples should be pretty straightforard, so if you do need help follow up. And with numpy arrays you can also filter/remove items based on criteria.

Sorry if these do not answer exactly what you are wanting, but it should get you headed in the right direction.