Making this C array processing code more python (and even numpy) -

i'm trying head around amazing list processing abilities of python (and numpy). i'm converting c code wrote python.

i have text datafile first row header, , every odd row input data , every row output data. data space separated. i'm quite chuffed managed read data lists using nested list comprehensions. amazing stuff.

with open('data.txt', 'r') f:     # lines list of strings     lines = list(f)      # convert header row list of ints , info     header = map(int, lines[0].split(' '))     num_samples = header[0]     input_dim = header[1]     output_dim = header[2]     del header          # bad ass list comprehensions      inputs = [[float(x) x in l.split()] l in lines[1::2]]     outputs = [[float(x) x in l.split()] l in lines[2::2]]     del x, l, lines 

then want produce new list each element function of corresponding input-output pair. couldn't figure out how python specific optimizations. here in c-style python:

# calculate position pos_list = []; pos_y = 0 in range(num_samples):     pantilt = outputs[i];     target = inputs[i];      if(pantilt[0] > 90):         pantilt[0] -=180         pantilt[1] *= -1     elif pantilt[0] < -90:         pantilt[0] += 180         pantilt[1] *= -1      tan_pan = math.tan(math.radians(pantilt[0]))     tan_tilt = math.tan(math.radians(pantilt[1]))      pos = [0, pos_y, 0]     pos[2] = tan_tilt * (target[1] - pos[1]) / math.sqrt(tan_pan * tan_pan + 1)     pos[0] = pos[2] * tan_pan     pos[0] += target[0]     pos[2] += target[2]     pos_list.append(pos) del pantilt, target, tan_pan, tan_tilt, pos, pos_y 

i tried comprehension, or map couldn't figure out how to:

• draw 2 different lists (both input , output) each element of pos_list array
• put body of algorithm in comprehension. have separate function or there funky way of using lambdas this?
• would possible no loops @ all, stick in numpy , vectorize whole thing?

one vectorized approach using boolean-indexing/mask -

import numpy np  def mask_vectorized(inputs,outputs,pos_y):     # create copy of outputs array editing purposes     pantilt_2d = outputs[:,:2].copy()      # mask correspindig if conditional statements in original code     mask_col0_lt = pantilt_2d[:,0]<-90     mask_col0_gt = pantilt_2d[:,0]>90      # edit first column per statements in original code     pantilt_2d[:,0][mask_col0_gt] -= 180     pantilt_2d[:,0][mask_col0_lt] += 180      # edit second column per statements in original code     pantilt_2d[ mask_col0_lt | mask_col0_gt,1] *= -1      # vectorized tan_pan , tan_tilt      tan_pan_tilt = np.tan(np.radians(pantilt_2d))      # vectorized calculation for: "tan_tilt * (target[1] .." original code      v = (tan_pan_tilt[:,1]*(inputs[:,1] - pos_y))/np.sqrt((tan_pan_tilt[:,0]**2)+1)      # setup output numpy array     pos_array_vectorized = np.empty((num_samples,3))      # put in values columns of output array     pos_array_vectorized[:,0] = inputs[:,0] + tan_pan_tilt[:,0]*v     pos_array_vectorized[:,1] = pos_y     pos_array_vectorized[:,2] = inputs[:,2] + v      # convert list, if desired final output     # (keeping numpy array boost performance further)     return pos_array_vectorized.tolist() 

runtime tests

in [415]: # parameters , setup input arrays      ...: num_samples = 1000      ...: outputs = np.random.randint(-180,180,(num_samples,5))      ...: inputs = np.random.rand(num_samples,6)      ...: pos_y = 3.4      ...:   in [416]: %timeit original(inputs,outputs,pos_y) 100 loops, best of 3: 2.44 ms per loop  in [417]: %timeit mask_vectorized(inputs,outputs,pos_y) 10000 loops, best of 3: 181 µs per loop