freealg 0.1.4__tar.gz → 0.1.6__tar.gz

{freealg-0.1.4/freealg.egg-info → freealg-0.1.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: freealg
-Version: 0.1.4
+Version: 0.1.6
 Summary: Free probability for large matrices
 Keywords: leaderboard bot chat
 Platform: Linux

freealg-0.1.6/freealg/__version__.py

@@ -0,0 +1 @@
+__version__ = "0.1.6"

{freealg-0.1.4 → freealg-0.1.6}/freealg/_decompress.py

@@ -11,8 +11,9 @@
 # =======
 
 import numpy
+# from scipy.integrate import solve_ivp
 
-__all__ = ['decompress']
+__all__ = ['decompress', 'reverse_characteristics']
 
 
 # ==========
@@ -134,3 +135,46 @@ def decompress(matrix, size, x=None, delta=1e-4, iterations=500, step_size=0.1,
     rho = rho.reshape(*x.shape)
 
     return rho, x, (lb, ub)
+
+
+# =======================
+# reverse characteristics
+# =======================
+
+def reverse_characteristics(matrix, z_inits, T, iterations=500, step_size=0.1,
+                            tolerance=1e-8):
+    """
+    """
+
+    t_span = (0, T)
+    t_eval = numpy.linspace(t_span[0], t_span[1], 50)
+
+    m = matrix._eval_stieltjes
+
+    def _char_z(z, t):
+        return z + (1 / m(z)[1]) * (1 - numpy.exp(t))
+
+    target_z, target_t = numpy.meshgrid(z_inits, t_eval)
+
+    z = numpy.full(target_z.shape, numpy.mean(matrix.support) - .1j,
+                   dtype=numpy.complex128)
+
+    # Broken Newton steps can produce a lot of warnings. Removing them for now.
+    with numpy.errstate(all='ignore'):
+        for _ in range(iterations):
+            objective = _char_z(z, target_t) - target_z
+            mask = numpy.abs(objective) >= tolerance
+            if not numpy.any(mask):
+                break
+            z_m = z[mask]
+            t_m = target_t[mask]
+
+            # Perform finite difference approximation
+            dfdz = _char_z(z_m+tolerance, t_m) - _char_z(z_m-tolerance, t_m)
+            dfdz /= 2*tolerance
+            dfdz[dfdz == 0] = 1.0
+
+            # Perform Newton step
+            z[mask] = z_m - step_size * objective[mask] / dfdz
+
+    return z

{freealg-0.1.4 → freealg-0.1.6}/freealg/_plot_util.py

@@ -18,6 +18,7 @@ import matplotlib
 import colorsys
 import matplotlib.ticker as ticker
 import matplotlib.gridspec as gridspec
+from ._decompress import reverse_characteristics
 
 __all__ = ['plot_fit', 'plot_density', 'plot_hilbert', 'plot_stieltjes',
            'plot_stieltjes_on_disk']
@@ -243,11 +244,42 @@ def _value_formatter(v, pos):
         return f"{m_val:.1f}"
 
 
+# ================
+# plot char curves
+# ================
+def _plot_char_curves(ax, char_curves):
+    """
+    """
+
+    curves = reverse_characteristics(char_curves['matrix'],
+                                     char_curves['z'], 4)
+    lw = 2
+    for idx in range(curves.shape[1]):
+
+        creal, cimag = curves[:, idx].real, curves[:, idx].imag
+
+        ax.plot(creal, cimag, ':', color='white', linewidth=lw,
+                alpha=0.75)
+
+        ax.annotate(
+            '',                             # no text
+            xy=(creal[-1], cimag[-1]),      # arrow tip at final point
+            xytext=(creal[-2], cimag[-2]),  # tail at penultimate point
+            arrowprops=dict(
+                arrowstyle='-|>',           # simple arrow head
+                mutation_scale=5,           # size of the head
+                color='white',
+                lw=lw, alpha=0.75,          # arrow shaft line width
+            )
+        )
+
+
 # ==============
 # plot stieltjes
 # ==============
 
-def plot_stieltjes(x, y, m1, m2, support, latex=False, save=False):
+def plot_stieltjes(x, y, m1, m2, support, latex=False, char_curves=None,
+                   save=False):
     """
     """
 
@@ -275,8 +307,8 @@ def plot_stieltjes(x, y, m1, m2, support, latex=False, save=False):
         ax0.plot([lam_m, lam_p], [eps, eps], '-', linewidth=1, color='black')
         ax0.set_xlabel(r'$\mathrm{Re}(z)$')
         ax0.set_ylabel(r'$\mathrm{Im}(z)$')
-        ax0.set_title(r'(a) Principal Branch on $\mathbb{H}$ and ' +
-                      r'$\mathbb{H}^{-}$')
+        ax0.set_title(r'(a) Principal Branch on $\mathbb{C}^{+}$ and ' +
+                      r'$\mathbb{C}^{-}$')
         ax0.set_yticks(numpy.arange(y_min, y_max+0.01, 0.5))
         ax0.set_xlim([x_min, x_max])
         ax0.set_ylim([y_min, y_max])
@@ -291,12 +323,16 @@ def plot_stieltjes(x, y, m1, m2, support, latex=False, save=False):
         ax1.plot([lam_p, x_max], [eps, eps], '-', linewidth=1, color='black')
         ax1.set_xlabel(r'$\mathrm{Re}(z)$')
         ax1.set_ylabel(r'$\mathrm{Im}(z)$')
-        ax1.set_title(r'(b) Principal Branch on $\mathbb{H}$, Secondary ' +
-                      r'Branch on $\mathbb{H}^{-}$')
+        ax1.set_title(r'(b) Principal Branch on $\mathbb{C}^{+}$, Secondary ' +
+                      r'Branch on $\mathbb{C}^{-}$')
         ax1.set_yticks(numpy.arange(y_min, y_max+0.01, 0.5))
         ax1.set_xlim([x_min, x_max])
         ax1.set_ylim([y_min, y_max])
 
+        # Plot characteristic curves
+        if char_curves is not None:
+            _plot_char_curves(ax1, char_curves)
+
         pos = ax1.get_position()
         cbar_width = 0.013
         pad = 0.013
@@ -503,9 +539,9 @@ def plot_samples(x, rho, x_min, x_max, samples, latex=False, save=False):
 
         fig, ax = plt.subplots(figsize=(6, 3))
 
-        bins = numpy.linspace(x_min, x_max, samples.size // 10)
+        bins = numpy.linspace(x_min, x_max, samples.size // 15)
         _ = ax.hist(samples, bins, density=True, color='silver',
-                    label='Samples histogram')
+                    edgecolor='none', label='Samples histogram')
         ax.plot(x, rho, color='black', label='Exact density')
         ax.legend(fontsize='small')
         ax.set_ylim(bottom=0)

{freealg-0.1.4 → freealg-0.1.6}/freealg/distributions/kesten_mckay.py

@@ -21,6 +21,7 @@ try:
     from scipy.integrate import cumtrapz
 except ImportError:
     from scipy.integrate import cumulative_trapezoid as cumtrapz
+from scipy.stats import qmc
 
 __all__ = ['KestenMcKay']
 
@@ -429,8 +430,8 @@ class KestenMcKay(object):
     # sample
     # ======
 
-    def sample(self, size, x_min=None, x_max=None, plot=False, latex=False,
-               save=False):
+    def sample(self, size, x_min=None, x_max=None, method='qmc', plot=False,
+               latex=False, save=False):
         """
         Sample from distribution.
 
@@ -448,6 +449,12 @@ class KestenMcKay(object):
             Maximum of sample values. If `None`, the right edge of the support
             is used.
 
+        method : {``'mc'``, ``'qmc'``}, default= ``'qmc'``
+            Method of drawing samples from uniform distirbution:
+
+            * ``'mc'``: Monte Carlo
+            * ``'qmc'``: Quasi Monte Carlo
+
         plot : bool, default=False
             If `True`, samples histogram is plotted.
 
@@ -503,9 +510,17 @@ class KestenMcKay(object):
         inv_cdf = interp1d(cdf, xs, bounds_error=False,
                            fill_value=(x_min, x_max))
 
-        # Sample and map
-        u = numpy.random.rand(size)
-        samples = inv_cdf(u)
+        # Draw from uniform distribution
+        if method == 'mc':
+            u = numpy.random.rand(size)
+        elif method == 'qmc':
+            engine = qmc.Halton(d=1)
+            u = engine.random(size)
+        else:
+            raise ValueError('"method" is invalid.')
+
+        # Draw from distribution by mapping from inverse CDF
+        samples = inv_cdf(u).ravel()
 
         if plot:
             radius = 0.5 * (self.lam_p - self.lam_m)

{freealg-0.1.4 → freealg-0.1.6}/freealg/distributions/marchenko_pastur.py

@@ -20,6 +20,7 @@ try:
     from scipy.integrate import cumtrapz
 except ImportError:
     from scipy.integrate import cumulative_trapezoid as cumtrapz
+from scipy.stats import qmc
 
 __all__ = ['MarchenkoPastur']
 
@@ -435,8 +436,8 @@ class MarchenkoPastur(object):
     # sample
     # ======
 
-    def sample(self, size, x_min=None, x_max=None, plot=False, latex=False,
-               save=False):
+    def sample(self, size, x_min=None, x_max=None, method='qmc', plot=False,
+               latex=False, save=False):
         """
         Sample from distribution.
 
@@ -454,6 +455,12 @@ class MarchenkoPastur(object):
             Maximum of sample values. If `None`, the right edge of the support
             is used.
 
+        method : {``'mc'``, ``'qmc'``}, default= ``'qmc'``
+            Method of drawing samples from uniform distirbution:
+
+            * ``'mc'``: Monte Carlo
+            * ``'qmc'``: Quasi Monte Carlo
+
         plot : bool, default=False
             If `True`, samples histogram is plotted.
 
@@ -509,9 +516,17 @@ class MarchenkoPastur(object):
         inv_cdf = interp1d(cdf, xs, bounds_error=False,
                            fill_value=(x_min, x_max))
 
-        # Sample and map
-        u = numpy.random.rand(size)
-        samples = inv_cdf(u)
+        # Draw from uniform distribution
+        if method == 'mc':
+            u = numpy.random.rand(size)
+        elif method == 'qmc':
+            engine = qmc.Halton(d=1)
+            u = engine.random(size)
+        else:
+            raise ValueError('"method" is invalid.')
+
+        # Draw from distribution by mapping from inverse CDF
+        samples = inv_cdf(u).ravel()
 
         if plot:
             radius = 0.5 * (self.lam_p - self.lam_m)

{freealg-0.1.4 → freealg-0.1.6}/freealg/distributions/wachter.py

@@ -20,6 +20,7 @@ try:
     from scipy.integrate import cumtrapz
 except ImportError:
     from scipy.integrate import cumulative_trapezoid as cumtrapz
+from scipy.stats import qmc
 
 __all__ = ['Wachter']
 
@@ -432,8 +433,8 @@ class Wachter(object):
     # sample
     # ======
 
-    def sample(self, size, x_min=None, x_max=None, plot=False, latex=False,
-               save=False):
+    def sample(self, size, x_min=None, x_max=None, method='qmc', plot=False,
+               latex=False, save=False):
         """
         Sample from distribution.
 
@@ -451,6 +452,12 @@ class Wachter(object):
             Maximum of sample values. If `None`, the right edge of the support
             is used.
 
+        method : {``'mc'``, ``'qmc'``}, default= ``'qmc'``
+            Method of drawing samples from uniform distirbution:
+
+            * ``'mc'``: Monte Carlo
+            * ``'qmc'``: Quasi Monte Carlo
+
         plot : bool, default=False
             If `True`, samples histogram is plotted.
 
@@ -506,9 +513,17 @@ class Wachter(object):
         inv_cdf = interp1d(cdf, xs, bounds_error=False,
                            fill_value=(x_min, x_max))
 
-        # Sample and map
-        u = numpy.random.rand(size)
-        samples = inv_cdf(u)
+        # Draw from uniform distribution
+        if method == 'mc':
+            u = numpy.random.rand(size)
+        elif method == 'qmc':
+            engine = qmc.Halton(d=1)
+            u = engine.random(size)
+        else:
+            raise ValueError('"method" is invalid.')
+
+        # Draw from distribution by mapping from inverse CDF
+        samples = inv_cdf(u).ravel()
 
         if plot:
             radius = 0.5 * (self.lam_p - self.lam_m)

{freealg-0.1.4 → freealg-0.1.6}/freealg/distributions/wigner.py

@@ -21,6 +21,7 @@ try:
     from scipy.integrate import cumtrapz
 except ImportError:
     from scipy.integrate import cumulative_trapezoid as cumtrapz
+from scipy.stats import qmc
 
 __all__ = ['Wigner']
 
@@ -416,8 +417,8 @@ class Wigner(object):
     # sample
     # ======
 
-    def sample(self, size, x_min=None, x_max=None, plot=False, latex=False,
-               save=False):
+    def sample(self, size, x_min=None, x_max=None, method='qmc', plot=False,
+               latex=False, save=False):
         """
         Sample from distribution.
 
@@ -435,6 +436,12 @@ class Wigner(object):
             Maximum of sample values. If `None`, the right edge of the support
             is used.
 
+        method : {``'mc'``, ``'qmc'``}, default= ``'qmc'``
+            Method of drawing samples from uniform distirbution:
+
+            * ``'mc'``: Monte Carlo
+            * ``'qmc'``: Quasi Monte Carlo
+
         plot : bool, default=False
             If `True`, samples histogram is plotted.
 
@@ -490,9 +497,17 @@ class Wigner(object):
         inv_cdf = interp1d(cdf, xs, bounds_error=False,
                            fill_value=(x_min, x_max))
 
-        # Sample and map
-        u = numpy.random.rand(size)
-        samples = inv_cdf(u)
+        # Draw from uniform distribution
+        if method == 'mc':
+            u = numpy.random.rand(size)
+        elif method == 'qmc':
+            engine = qmc.Halton(d=1)
+            u = engine.random(size)
+        else:
+            raise ValueError('"method" is invalid.')
+
+        # Draw from distribution by mapping from inverse CDF
+        samples = inv_cdf(u).ravel()
 
         if plot:
             radius = 0.5 * (self.lam_p - self.lam_m)

{freealg-0.1.4 → freealg-0.1.6}/freealg/freeform.py

@@ -631,7 +631,8 @@ class FreeForm(object):
     # stieltjes
     # =========
 
-    def stieltjes(self, x, y, plot=False, latex=False, save=False):
+    def stieltjes(self, x, y, plot=False, latex=False, char_curves=None,
+                  save=False):
         """
         Compute Stieltjes transform of the spectral density over a 2D Cartesian
         grid on the complex plane.
@@ -655,6 +656,10 @@ class FreeForm(object):
             If `True`, the plot is rendered using LaTeX. This option is
             relevant only if ``plot=True``.
 
+        char_curves : numpy.array, default=None
+            If ``plot=True``, also plot characteristic curves starting from
+            these locations.
+
         save : bool, default=False
             If not `False`, the plot is saved. If a string is given, it is
             assumed to the save filename (with the file extension). This option
@@ -742,7 +747,9 @@ class FreeForm(object):
         m2[mask_m, :] = -m1[mask_m, :] + self._glue(z[mask_m, :])
 
         if plot:
-            plot_stieltjes(x, y, m1, m2, self.support, latex=latex, save=save)
+            plot_stieltjes(x, y, m1, m2, self.support, latex=latex,
+                           char_curves={'matrix': self, 'z': char_curves},
+                           save=save)
 
         return m1, m2
 

{freealg-0.1.4 → freealg-0.1.6/freealg.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: freealg
-Version: 0.1.4
+Version: 0.1.6
 Summary: Free probability for large matrices
 Keywords: leaderboard bot chat
 Platform: Linux

freealg-0.1.4/freealg/__version__.py

@@ -1 +0,0 @@
-__version__ = "0.1.4"