sort_file

def sort_file(save_dir, save_as):
    #the tolerance for what is considered equivalent continuous parameters
    rho_tol = 3.0; alpha_tol = 5000.0; sigma_tol = 2.0
 
    #reads in full optimization data
    f = open(save_dir + save_as)
    data = []
    for line in f:
        data.append(line.split())
    f.close()
    #sorts data by score
    ranked_data = sorted(data, key=lambda dataline: float(dataline[0]))
    sorted_data = []; sorted_scores = []; sorted_stddev = []
    while len(sorted_data) < 100:
        #create rounded version of best ranked parameter set
        candidate = []
        junk = ranked_data[-1][4:]
        holder = junk[0].split('[')
        holder = holder[1].split(',')
        candidate.append(int(holder[0]))
        holder = junk[1].split(',')
        candidate.append(int(holder[0]))
        holder = junk[2].split(',')
        candidate.append(round(float(holder[0])/rho_tol)*rho_tol)
        #floor rho at 1.0
        if candidate[-1] == 0:
            candidate[-1] = 1
        holder = junk[3].split(',')
        candidate.append(round(float(holder[0])/alpha_tol)*alpha_tol)
        #floor alpha at 1.0
        if candidate[-1] == 0:
            candidate[-1] = 1
        holder = junk[4].split(',')
        candidate.append(round(float(holder[0])/sigma_tol)*sigma_tol)
        #floor sigma at 1.0
        if candidate[-1] == 0:
            candidate[-1] = 1
        holder = junk[5].split(']')
        candidate.append(int(holder[0]))
        #add to sorted lists only if distinct from other parameters
        if candidate not in sorted_data:
            sorted_data.append(candidate)
            sorted_scores.append(ranked_data[-1][0])
            sorted_stddev.append(ranked_data[-1][2])
        #remove this to check next best parameter set
        ranked_data.pop()
    #save sorted data
    g = open(save_dir + save_as + '_sorted','w')
    for k in range(len(sorted_data)):
        g.write(sorted_scores[k] + ' score, ' + sorted_stddev[k] + \
            ' stddev, ' + repr(sorted_data[k]) + '\n')
    g.close()