CODEWITHSUNDEEP
pythonfloating-accuracy
Max
Max

Reputation: 191

Curious Python behaviour when approaching x.0

I was playing around with a simple recursive formula and noticed that the code

p = 2.0

while p < 3.0:
    print p
    p = (6+p)**(0.5)

will print

*snip*
...
2.99999999952
2.99999999992
2.99999999999
3.0
3.0
3.0
3.0
3.0
3.0

Of course there will be some kind of approximation between 2.99999999999 and 3.0 (and before that) but what is actually happening here? For me it seems odd that the floating point 3.0 will be interpreted as something not quite 3.0 but still close enough to be called 3.0, several times in a row.

Am I doing something wrong, code-wise, here or is my interpretation correct? I if so, why is this happening?

Upvotes: 2

Views: 88

Answers (2)

NPE
NPE

Reputation: 500893

If you change the print statement like so:

print '%.20f' % p

all will become clear:

2.00000000000000000000
2.82842712474619029095
2.97126692250060076006
2.99520732546189538681
2.99920111454065274614
2.99986684946859805123
2.99997780816268644344
2.99999630135816763854
2.99999938355963147174
2.99999989725993687628
2.99999998287665592400
2.99999999714610909862
2.99999999952435159045
2.99999999992072519106
2.99999999998678745783
2.99999999999779776161
2.99999999999963273822
2.99999999999993871569
2.99999999999998978595
2.99999999999999822364
2.99999999999999955591

Upvotes: 5

Fred Foo
Fred Foo

Reputation: 363817

This is due to the way print formats your float. Try printing with

print("%.40f" % p)

Upvotes: 3

Related Questions