How to use unwrap in this case?

Question

I am actually trying do unwrap a signal and the numpy function don't do what I want. I already searched the web for an answer of course and nothing help me unfortunatly: here is a part of the code:

• NT can be in the range of 1 to 48.
• The size of h1 is in Gb (like 10-20Gb), and I extract like 20,000,000 rows in one NT.
• I can't do it in one big process I need to divide it with NT.

for i in range(0,NT,1):#NT
    A = 0
    temp = pd.DataFrame()
    h1 = pd.read_hdf(path,'foo',start=int(Compt1*i/NT),stop=int(Compt1*(i+1)/NT))
    print('\033[0m'+'calcule d1 et d2 : NT '+str(i+1)+' / '+str(NT))
    temp = list(h1.phi)
    
    # s=temp[0]-tp
    # if s<(-np.pi) :
    #     temp[0]+=2*np.pi
    # elif s>np.pi:
    #     temp[0]-=2*np.pi
    
    # ech=NT*record_size
    
    # for y in range(1,len(temp),1):
    #     s=temp[y]-temp[y-1]
    #     if s<(-np.pi) :
    #         temp[y]+=2*np.pi
    #     elif s>np.pi:
    #         temp[y]-=2*np.pi
    # temp0=[tempi*(londe/(4*np.pi)) for tempi in temp]
    
    temp0 = np.unwrap(temp,axis=0)
    A = pd.DataFrame({
        'd1':[temp0*(londe/(4*np.pi)) for temp0 in temp],
        'd2':Q+np.cumsum(h1.deltax)})
    h1 = 0
    tp = temp[-1]
    Q = A.iloc[-1,1]
    A = A.apply(pd.to_numeric, downcast='float')
    print('\033[0m'+'enregistrement')
    with pd.HDFStore(pathout, data_columns=True) as store:
        try:
            nrows = store.get_storer('foo').nrows
        except:
            nrows = 0
        A.index = pd.Series(A.index) + nrows
        store.append('foo', A)
    print('\033[32m'+'Ok')

Here is what I got (d1 is not good, d2 is good):

https://i.stack.imgur.com/ULYFP.jpg

Last one in the link is what I want, first one is what I get and second one is what I get with zoom.

Answer 1

Problem solved: this was this line

temp0=np.unwrap(temp,axis=0)
'd1':[temp0*(londe/(4*np.pi)) for temp0 in temp],

who become

temp0=np.unwrap(temp)*(londe/(4*np.pi))
'd1':temp0,

thanks if your going to anwser