Defining precision of floating point number in Python

Question

I have some python code here

Code:

 t = 1
 xt = array([ -1.,-1. ])

 xt[0] = theta[0][0]* P_all_States[t-1][0] + theta[1][0]*P_all_States[t-1][1]

 xt[1] = theta[0][1]* P_all_States[t-1][0] + theta[1][1]*P_all_States[t-1][1]

 P_all_States[t][0] = xt[0]
 P_all_States[t][1] = xt[1]
 print ("Added probability distribution for time t = " + str(t) + " to P_all_States")
 print P_all_States

Output:

Added probability distribution for time t = 1 to P_all_States [[0.6, 0.4], [0.62000000000000011, 0.38000000000000006], [-1.0, -1.0], [-1.0, -1.0], [-1.0, -1.0], [-1.0, -1.0]]

How can I get the floating point numbers to round to 2 sigficant digits?

These are all floats, so Im not looking to convert these to strings, I want them to remain as floats

Thanks

related: [Pretty-printing of numpy.array](http://stackoverflow.com/q/2891790/4279), namely `np.set_printoptions(precision=2)` — jfs, Oct 13 '12 at 20:06

score 3 · Answer 1 · answered Oct 13 '12 at 20:09

I am not sure what you mean, but if you mean how you can modify their value to exactly 0.62, etc. : this is not possible with floats, because of their nature. You could use decimals instead of floats to get around this issue. If you mean how to print the floats with only 2 decimals, use:

print '{:3.2f}'.format(somefloat)

score 2 · Accepted Answer · answered Oct 13 '12 at 20:05

2

>>> round(0.62000000000000011,2)
0.62

answered Oct 13 '12 at 20:05

Zashas

736
4
11

1

You should note that this doesn't actually result in the value `0.62`. `print decimal.Decimal(round(0.62000000000000011,2))` -> `0.61999999999999999555910790149937383830547332763671875`. – senderle Oct 13 '12 at 20:29

Rudolf Mühlbauer · Answer 3 · 2012-10-13T20:21:45.480

0

To convert the whole list, combine @Zashas answer with:

rounded = [ [round(a[0],2), round(a[1],2)] for a in P_all_States ]

edited Oct 13 '12 at 20:21

answered Oct 13 '12 at 20:10

Rudolf Mühlbauer

2,511
16
18

it is a list of list, so the above won' t do the job, need something like: rounded = [ [round(a[0],2), round(a[1],2)] for a in P_all_States ] – John Peters Oct 13 '12 at 20:12
sorry - did not realize that. well, you corrected it. thanks! – Rudolf Mühlbauer Oct 13 '12 at 20:20

Defining precision of floating point number in Python

3 Answers3