1
$\begingroup$

Can you help me in finding the analytical expression of a function $f_\alpha(\theta)$, with one parameter $\alpha=(0,1)$ by which one can continously transform a sawtooth curve into a sinus?

With $\alpha=0$, I'd like to have $f_0(\theta)=\sin(2\theta)$.

With $\alpha=1$, I'd like to have $f_1(\theta)=sawtooth(\theta)$.

where $sawtooth(2\theta)$ is the 100% asymmetric sawtooth, $\pi$ periodic, like $sawtooth(t) = 2(\frac{t}{\pi}-floor(0.5+\frac{t}{\pi}))$

I've been trying adding Fourier components, low pass and other filters, but I cannot get a smooth change from one to the other when changing $\alpha$.

This animation (from wikipedia) can be an example, but I don't want the little ripples due to the finite number of Fourier components.

Thanks

$\endgroup$
2
  • $\begingroup$ $\alpha*f_1(\theta)+(1-\alpha)f_0(\theta)$ $\endgroup$
    – Paul
    Nov 13 '15 at 11:52
  • $\begingroup$ Thanks, I tried that (I should have mentioned it) but it's not what I need. The discontinuity of the sawtooth kicks in for $\alpha>0$, while I need the peaks of the sin() to shift towards those of the sawtooth, without discontinuities. The animation linked in the post gives an idea (but the ripples are bad there). $\endgroup$
    – scrx2
    Nov 13 '15 at 12:01
1
$\begingroup$

I post my own answer, hoping it can be useful. It is a script (in python), which builds what I need from Fourier components and filtering. The output is a curve with one parameter $m$ which can go from a $sin(2\theta)$ to a sawtooth$(2\theta)$. However it's not an analytical function, so the question remains open. GRAPH here

def sin2sawtooth(m, sawint=1, plots=False):

    ''' calculate the curve f_m(theta), 
    function of theta=(0,2pi), with free parameter 'm'.
    m = [0,100] is the symmetry of the curve, 
    with m = 1, it gives a sawtooth(2 theta) curve, 
    and  m = 100, it gives a sin(2 theta).
    sawint = intensity of the curve
    plots = False|True : plot result
    '''

    import numpy as np
    import matplotlib.pyplot as plt
    import scipy.ndimage.filters

    # define variable theta:
    theta = np.linspace(0, 2*np.pi, 1000)
    # force m to be an integer in [1,100]:
    m = int(np.clip(m, 1, 100))
    # define number of fourier components to build the sawtooth 'L' :
    if m == 100: L = 1
    else: L = 200
    # make theta longer to avoid edge effects:
    theta_conc = np.append(theta-2*np.pi, np.append(theta, theta+2*np.pi))
    saw = np.zeros(len(theta_conc))
    # build 'saw' by sum of fourier components:
    for k in range(1, L+1):
        saw = saw + 1/np.pi*(np.sin(2*k*theta_conc)/k )
    # gaussian filter: 
    saw = scipy.ndimage.filters.gaussian_filter(saw, m)
    # crop theta in 0,2pi:
    saw = saw[len(theta): 2*len(theta)]
    # normalize 'saw' between 0-1 :
    saw = (saw-np.min(saw))/(np.max(saw)-np.min(saw))
    if plots:
        plt.figure(3241)
        plt.plot(theta, saw)
        plt.grid(True)
    return saw
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.