shifted tests to proper test routine

zdm/energetics.py

@@ -46,200 +46,7 @@ def init_igamma_splines(gammas, reinit=False,k=3):
             else:
                 igamma_splines[gamma] = interpolate.splrep(avals, numer,k=k)
 
-
-
-def test_spline_accuracy(gamma = -1.5837, Ntest = 100):
-    """
-    Function to test the accuracy of the spline interplation
-    It explores different methods of spline interpolation
-    
-    Ntest [int]: number of random values to generate
-    gamma [float]: value of the index gamma to test at
-    
-    Output: This produces three plots, being
-        spline_test.pdf: compares splines to truth
-        spline_errors.pdf: plots absolute value of errors
-        spline_rel_errors.pdf: plots absolute value of relative errors
-    """
-    global SplineMin,SplineMax,NSpline,SplineLog
-
-    # generate random values in the range of the splines
-    lnewavals = np.random.rand(Ntest) * (SplineMax - SplineMin) + SplineMin
-    lnewavals = np.sort(lnewavals)
-    newavals = 10**lnewavals
-
-
-    # generate the true values at these avalues via direct calculation
-    truth = np.zeros(Ntest)
-    for i,av in enumerate(newavals):
-        truth[i] = mpmath.gammainc(gamma, a=av)
-
-    # sets up for plotting results
-    from matplotlib import pyplot as plt
-    plt.figure()
-    ax1=plt.gca()
-    plt.plot(newavals,truth,label='truth')
-    plt.xlabel('a')
-    plt.ylabel('cumulative gamma function')
-    plt.xscale('log')
-    plt.yscale('log')
-
-    # sets up for plotting differences between truth and interpolation
-    plt.figure()
-    ax2=plt.gca()
-    plt.xlabel('a')
-    plt.ylabel('|interpolation - truth|')
-    plt.xscale('log')
-    plt.yscale('log')
-
-    # sets up for plotting relative differences between truth and interpolation
-    plt.figure()
-    ax3=plt.gca()
-    plt.xlabel('a')
-    plt.ylabel('|interpolation - truth|/truth')
-    plt.xscale('log')
-    plt.yscale('log')
-
-    # now use different spline methods to evaluate
-    # standard method: cubic, linear
-
-    SplineLog=False
-    init_igamma_splines([gamma],reinit=True,k=3)
-    result = interpolate.splev(newavals, igamma_splines[gamma])
-    ax1.plot(newavals,result,label='cubic spline, linear')
-    diff = np.abs(result - truth)
-    ax2.plot(newavals,diff,label='cubic spline, linear')
-    rdiff = diff/truth
-    ax3.plot(newavals,rdiff,label='cubic spline, linear')
-
-    # now use different spline methods to evaluate
-    # standard method: cubic, linear
-    init_igamma_splines([gamma],reinit=True,k=1)
-    result = interpolate.splev(newavals, igamma_splines[gamma])
-    ax1.plot(newavals,result,label='linear spline, linear')
-    diff = np.abs(result - truth)
-    ax2.plot(newavals,diff,label='linear spline, linear')
-    rdiff = diff/truth
-    ax3.plot(newavals,rdiff,label='linear spline, linear')
-
-
-    SplineLog=True
-    init_igamma_splines([gamma],reinit=True,k=3)
-    result = 10**interpolate.splev(lnewavals, igamma_splines[gamma])
-    ax1.plot(newavals,result,label='cubic spline, log')
-    diff = np.abs(result - truth)
-    ax2.plot(newavals,diff,label='cubic spline, log')
-    rdiff = diff/truth
-    ax3.plot(newavals,rdiff,label='cubic spline, log')
-
-
-    # now use different spline methods to evaluate
-    # standard method: cubic, linear
-    init_igamma_splines([gamma],reinit=True,k=1)
-    result = 10**interpolate.splev(lnewavals, igamma_splines[gamma])
-    ax1.plot(newavals,result,label='linear spline, log')
-    diff = np.abs(result - truth)
-    ax2.plot(newavals,diff,label='linear spline, log')
-    rdiff = diff/truth
-    ax3.plot(newavals,rdiff,label='linear spline, log')
-
-    plt.sca(ax1)
-    plt.legend()
-    plt.tight_layout()
-    plt.savefig('spline_test.pdf')
-    plt.close()
-
-    plt.sca(ax2)
-    plt.legend()
-    plt.tight_layout()
-    plt.savefig('spline_errors.pdf')
-
-    plt.sca(ax3)
-    plt.legend()
-    plt.ylim(1e-14,1)
-    plt.tight_layout()
-    plt.savefig('spline_rel_errors.pdf')
-    plt.close()
-
-
-def time_splines(gamma = -1.5837, Ntimetest = 100000, Nreps=100):
-    """
-    Function to time different methods of spline interpolation.
-    It explores different methods of spline interpolation
-    
-    gamma [float]: value of the index gamma to test at
-    Ntimetest [int]: number of random values to generate to test on
-    Nreps [int]: number of repetitions for timing
-    """
-    global SplineMin,SplineMax,NSpline
-
-    import time
-
-    # begins with very large array of values to evaluate on
-    lnewavals = np.random.rand(Ntimetest) * (SplineMax - SplineMin) + SplineMin
-    newavals = 10**lnewavals
-
-    # now use different spline methods to evaluate
-    # standard method: cubic, linear
-
-    SplineLog=False
-    t0=time.time()
-    for i in np.arange(Nreps):
-        init_igamma_splines([gamma],reinit=True,k=3)
-    t1=time.time()
-    for i in np.arange(Nreps):
-        result = interpolate.splev(newavals, igamma_splines[gamma])
-    t2=time.time()
-    dt1_ci = t1-t0
-    dt2_ci = t2-t1
-
-    # now use different spline methods to evaluate
-    # standard method: cubic, linear
-    t0=time.time()
-    for i in np.arange(Nreps):
-        init_igamma_splines([gamma],reinit=True,k=1)
-    t1=time.time()
-    for i in np.arange(Nreps):
-        result = interpolate.splev(newavals, igamma_splines[gamma])
-    t2=time.time()
-    dt1_li = t1-t0
-    dt2_li = t2-t1
-
-    SplineLog=True
-    t0=time.time()
-    for i in np.arange(Nreps):
-        init_igamma_splines([gamma],reinit=True,k=3)
-    t1=time.time()
-    for i in np.arange(Nreps):
-        result = 10**interpolate.splev(lnewavals, igamma_splines[gamma])
-    t2=time.time()
-    dt1_co = t1-t0
-    dt2_co = t2-t1
-
-    # now use different spline methods to evaluate
-    # standard method: cubic, linear
-    t0=time.time()
-    for i in np.arange(Nreps):
-        init_igamma_splines([gamma],reinit=True,k=1)
-    t1=time.time()
-    for i in np.arange(Nreps):
-        result = 10**interpolate.splev(lnewavals, igamma_splines[gamma])
-    t2=time.time()
-    dt1_lo = t1-t0
-    dt2_lo = t2-t1
-
-    print("Performing ",Nreps," spline initialisations took...")
-    print("Cubic spline in linear space:  ",dt1_ci)
-    print("Linear spline in linear space: ",dt1_li)
-    print("Cubic spline in log space:     ",dt1_co)
-    print("Linear spline in log space:    ",dt1_lo)
-
-    print("Performing ",Nreps," x ",Ntimetest," spline evaluations took...")
-    print("Cubic spline in linear space:  ",dt2_ci)
-    print("Linear spline in linear space: ",dt2_li)
-    print("Cubic spline in log space:     ",dt2_co)
-    print("Linear spline in log space:    ",dt2_lo)
-
+
 def init_igamma_linear(gammas:list, reinit:bool=False, 
                        log:bool=False):
     """ Setup the linear interpolator for gamma