For a numerical analysis class I need to make a for loop that must do a couple of things:

1. Make a certain 'Dt' (a time step that gets smaller and smaller) and add this to an array called Dt_list. This must be an array containing all the different Dt's at the end of the loop

2. Call a function called "EstimateError" with multiple parameters including Dt which changes every time the loop restarts. This function returns a one dimensional array with two components: ([est_global_error_1, est_global_error_2]) This output has to also be added to an array called: err_list being a two dimensional array.

3. inside the for loop there must be a (nested) 'if' that checks whether or not one of the values in the global_trunction_error is smaller than or equal to 0.001, if this is the case the for loop must break.

Although seemingly easy, I can not get it to work properly after trying many variations. There seems to be something going wrong when I try to append a value to either one of the lists.

After the for-loop I must attain an optimal Dt which should be the last value on the Dt_list.

err_list = np.array([],[])
Dt_list = np.array([])


for i in np.linspace(0,100,101):
    Dt = (0.25)*(0.5)**(i)
    
    global_trunction_error = EstimateError(t_0,t_end,Dt,y_0, a,c,A,C,W_0)
    np.append(err_list, global_trunction_error)
    
    Dt_list = np.append(Dt_list, Dt)
    
    if (abs(global_trunction_error[0])) <= 0.001:
        if (abs(global_trunction_error[1])) <= 0.001:
            break

            
Dt_opt = Dt_list[-1]

CodePudding user response：

Your code starts off wrong:

In [199]: err_list = np.array([],[])    
In [200]: err_list
Out[200]: array([], dtype=[])

I expected this to raise an error, since the 2nd argument to np.array is supposed to be a dtype. Initially I was puzzled by that dtype=[], but then realized is a compound dtype with 0 fields. Normally a compound dtype looks like [('f0',float),('f1',int)], so I've never seen a 0 field dtype before.

Secondly you probably got an error like this:

In [204]: np.append(err_list, (1,2))
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Input In [204], in <cell line: 1>()
----> 1 np.append(err_list, (1,2))

File <__array_function__ internals>:5, in append(*args, **kwargs)

File ~\anaconda3\lib\site-packages\numpy\lib\function_base.py:4817, in append(arr, values, axis)
   4815     values = ravel(values)
   4816     axis = arr.ndim-1
-> 4817 return concatenate((arr, values), axis=axis)

File <__array_function__ internals>:5, in concatenate(*args, **kwargs)

TypeError: The DTypes <class 'numpy.dtype[int32]'> and <class 'numpy.dtype[void]'> do not have a common DType. For example they cannot be stored in a single array unless the dtype is `object`.

I probably should have just voted to close this question since you didn't provide any errors or their tracebacks.

First, np.append is not a list append clone! It returns a value; it does not change an array in-place.

Second, concatenating with a structured array requires arguments with a matching dtype.

On second thought, I am going to vote to close. Go back to your code and rewrite using lists and list appends. Appending values to arrays is error prone and inefficient.

CodePudding user response：

Thank you everyone for responding!

I found a way around my problem, I just divided the two dimensional array into two seperate lists and then combined them later on.

err_list_1 = []
err_list_2 = []
Dt_list = []


for i in np.linspace(1,100,10):
    Dt = (0.25)*(0.5)**(i)
    
    global_trunction_error = EstimateError(t_0,t_end,Dt,y_0, a,c,A,C,W_0)
    err_list_1.append(global_trunction_error[0])
    err_list_2.append(global_trunction_error[1])
    
    Dt_list.append(Dt)
    
    if (abs(global_trunction_error[0])) <= 0.001:
        if (abs(global_trunction_error[1])) <= 0.001:
            break

            
Dt_opt = Dt_list[-1]
err_list = np.array((err_list_1, err_list_2))

Dt_list = np.array((Dt_list))