shril / cubicequationsolver Goto Github PK

comparing to "zero" is problematic

try changing the first few line to

aa=abs(a)
ab=abs(b)
ac=abs(c)
tol=max((aa,ab,ac))/1099511627776 #/2**40
if (aa<tol and ab<tol):                     # Case for handling Liner Equation
    return np.array([(-d * 1.0) / c])                 # Returning linear root as numpy array.

elif (aa < tol):    

for a=-5.32282969291e-15 b=0.43709356404 c=0.819917378017 d=-0.199301067523

goes from
complex_roots: [ 8.211676669305905e+13 +0.j
-8.867187500000000e-01+353649.2408007157j
-8.867187500000000e-01-353649.2408007157j]
to
complex_roots: [ 0.2177888401916245 -2.093628352534504 ]
when numpy.roots says
complex_roots: [ 8.2116766693059141e+13 -2.0936283525344557e+00 2.1778884019162431e-01]

this came up in the context of a highly specialized pchip spline implementation I was working on

Thanks a lot for this useful code.

Would you agree to change the licence of the code to apache 2 so that it could be used in an open-source project? If yes, I'll be happy to do the corresponding pull request.

Thank you

Exactly I ended up succesfully by changing the directory for the cubicequationsolver.py file and now it returs the values for the roots in the Spyder console.

But how to plot the roots graphically by using matplotlib ?
Which commands has to be run in the Spyder console for plotting the roots graphically ?

An idea, you can squeeze some more performance juice by replacing x**2 with x*x and x**3 with x*x*x etc.

>>> from timeit import repeat
>>> repeat('findH(1, 2)', setup='def findH(g, f):\n\treturn ((g ** 2.0) / 4.0 + (f ** 3.0) / 27.0)')
[0.7711835503748148, 0.7689730036987186, 0.7664158748830658]
>>> repeat('findH(1, 2)', setup='def findH(g, f):\n\treturn ((g*g) / 4.0 + (f*f*f) / 27.0)')
[0.5918213720389645, 0.568905897817487, 0.5710548649790184]