numpy - efrc gives error length-1 array can be converted to python -

i'm trying calculate bit error rate in python numpy. code looks this:

ebbyno = arange(0,16,1) ebno   = 10**(ebbyno/10) ber2   = (3/8)*erfc(cmath.sqrt(ebno*(2/5)))+(1/4)*erfc(3*sqrt(2/5*ebno)) 

but it's giving me error:

ber2 = (3/8)*erfc(cmath.sqrt(ebno*(2/5)))+(1/4)*erfc(3*sqrt(2/5*ebno)) typeerror: length-1 arrays can converted python scalars 

cmath not support numpy arrays:

ber2=(3/8)*erfc(sqrt(ebno*(2/5)))+(1/4)*erfc(3*sqrt(2/5*ebno)) 

you seem importing lot of functions from foo import * can trip up. using ints (for example 2/5) instead of floats equation above returns array of zero's:

>>> 2/5 0 >>> 2./5 0.4 

i suggest:

>>> import numpy np >>> import scipy.special sp >>> ebbyno=np.arange(0.,16.,1) >>> ebno=10**(ebbyno/10) >>> ber2=(3./8)*sp.erfc(np.sqrt(ebno*(2./5)))+(1./4)*sp.erfc(3*np.sqrt(2./5*ebno)) >>> ber2 array([  1.40982603e-01,   1.18997473e-01,   9.77418560e-02,          7.74530603e-02,   5.86237373e-02,   4.18927600e-02,          2.78713278e-02,   1.69667344e-02,   9.24721374e-03,          4.39033609e-03,   1.75415062e-03,   5.64706106e-04,          1.38658689e-04,   2.42337855e-05,   2.76320800e-06,          1.84185551e-07])