Accelerating For Loop

[Stefan Behnel]

Şansal Birbaş, 23.02.2011 07:34:
> I needed to find the cheapest combination among given data and I developed an algorithm for this task. It works correctly. But it takes much time (nearly 2 minutes) for second function to find the result while it is just
> one second for the first function. How can I improve the calculation speed?
>[...]
>      for i in range(1,5):
>          for j in i*range(max_expander+1):
>              for k in i*range(max_expander+1):
>                  for m in i*range(max_expander+1):
>                      for r in i*range(max_expander+1):
>
>                          if (j+k+m+r)<=(i*max_expander):
>
>                              mevcut=zeros((1,5))
>                              mevcut+=X*i
>                              mevcut+=Y*j
>                              mevcut+=Z*k
>                              mevcut+=T*m
>                              mevcut+=W*r
>                              m_UI=mevcut[0][0]
>                              m_UO=mevcut[0][1]
>                              m_AO=mevcut[0][2]
>                              m_BO=mevcut[0][3]
>                              i_AI=istenen[0]
>                              i_AO=istenen[1]
>                              i_BI=istenen[2]
>                              i_BO=istenen[3]
>
>                              if (m_UI>=(i_AI+i_BI)):
>                                  if ((m_BO>=i_BO)):
>                                      kalanUO=m_UO
>                                  elif ((m_UO+m_BO)>=i_BO):
>                                      kalanUO=(m_UO+m_BO)-i_BO
>                                      if ((kalanUO+m_AO)>=i_AO):
>                                          temp=[]
>                                          temp.append(i)
>                                          temp.append(j)
>                                          temp.append(k)
>                                          temp.append(m)
>                                          temp.append(r)
>                                          fiyat=i*X[4]+j*Y[4]+k*Z[4]+m*T[4]+r*W[4]
>                                          temp.append(fiyat)
>                                          combinations.append(temp)

For this kind of computation, Cython is usually much faster than NumPy by 
itself.

http://cython.org/

http://docs.cython.org/src/tutorial/numpy.html

http://docs.cython.org/src/userguide/numpy_tutorial.html

Stefan