plotting.py

# -*- coding: utf-8 -*-

import os
import time
import itertools

import numpy
import pandas
import matplotlib.pyplot as plt

from common import UNITS, GRID, statistics as STATISTICS
from common.utils import ring_deque, getboolenv


def monitor_datafile(datafolder, timer=1):

    datafile = os.path.join(datafolder, 'evolution.pickle')

    def plot_backlog(data, last_datalen):
        i = last_datalen + 1
        while i < len(data):
            plotting.plot(data[:i], redraw=False)
            i += 1

        last_datalen = len(data)
        plotting.redraw()
        return last_datalen

    plt.ion()
    plotting = Plotting()
    data = pandas.read_pickle(datafile)
    last_datalen = plot_backlog(data, 0)

    last_mtime = os.stat(datafile).st_mtime
    while True:
        mtime = os.stat(datafile).st_mtime
        if mtime > last_mtime:
            try:
                data = pandas.read_pickle(datafile)
                if len(data) < last_datalen:
                    print("Datafile is shorter than before, clearing the output")
                    plt.close('all')
                    plotting = Plotting()
                    last_datalen = 0

                last_datalen = plot_backlog(data, last_datalen)
                print("Plotting done at", mtime)

            except Exception as e:
                print(e)
            last_mtime = mtime
        time.sleep(timer)


class Plotting(object):
    particles = None

    def __init__(self, show=True):
        """ Initialize plots, setup basic styling """

        self.show_plots = getboolenv("SHOW_PLOTS", show)

        self.params_figure, self.plots = plt.subplots(2, 2, num=1)
        self.plots = list(itertools.chain(*self.plots))
        self.params_figure.subplots_adjust(hspace=0.5, wspace=0.5)

        self.plot_map = ['T', 'a', 'aT', 'N_eff']
        self.divider_map = [UNITS.MeV, 1, UNITS.MeV, 1]

        self.plots[0].set_title("Temperature")
        self.plots[0].set_xlabel("time, s")
        self.plots[0].set_xscale("log")
        self.plots[0].set_yscale("log")
        self.plots[0].set_ylabel("T, MeV")
        self.plots[0].set_yscale("log")

        self.plots[1].set_title("Scale factor")
        self.plots[1].set_xlabel("time, s")
        self.plots[1].set_xscale("log")
        self.plots[1].set_yscale("log")
        self.plots[1].set_ylabel("a, 1")
        self.plots[1].set_ylim(0, 1)

        self.plots[2].set_title("T * a")
        self.plots[2].set_xlabel("time, s")
        self.plots[2].set_xscale("log")
        self.plots[2].set_ylabel("T * a, MeV")
        self.plots[2].set_ylim(1, 1.1)

        self.plots[3].set_title("N_eff")
        self.plots[3].set_xlabel("time, s")
        self.plots[3].invert_xaxis()
        self.plots[3].set_xscale("log")
        self.plots[3].set_ylabel("N_eff")

        self.lines = []
        self.plots_data = []
        self.times = ring_deque([], 1e6)
        for plot in self.plots:
            self.lines.append(plot.plot([], [], 'b-')[0])
            self.plots_data.append(ring_deque([], 1e6))

        if self.show_plots:
            self.params_figure.show()

    def redraw(self):
        self.params_figure.canvas.draw()
        if self.particles:
            self.particles_figure.canvas.draw()

    def save(self, filename):
        """ Save cosmological and monitored particles plots to the file in the same folder as \
            `filename` """

        folder = os.path.split(filename)[0]
        self.params_figure.savefig(os.path.join(folder, 'plots.svg'))
        if self.particles:
            self.particles_figure.savefig(os.path.join(folder, 'particles.svg'))

    def monitor(self, map):
        """ Setup the detailed distribution function and energy density plots for specific \
            particle species """

        self.particles_figure, self.particles_plots = plt.subplots(len(map), 2, num=2)
        self.particles_figure.subplots_adjust(hspace=0.5, wspace=0.5)

        self.particles = self.particles if self.particles else []

        for i, (particle, monitor) in enumerate(map):
            self.particles.append((particle, monitor(particle, self.particles_plots[i])))

        if self.show_plots:
            self.particles_figure.show()

    def plot(self, data, redraw=True):
        """ Plot cosmological parameters and monitored particles distribution functions """

        last_t = data['t'][-1] / UNITS.s
        if last_t == 0:
            return

        self.times.append(last_t)

        for i, plot in enumerate(self.plots):
            _, xmax = plot.get_xlim()
            ymin, ymax = plot.get_ylim()

            if last_t >= xmax:
                plot.set_xlim(self.times[0], last_t * 1.1)

            last_data = data[self.plot_map[i]][-1] / self.divider_map[i]
            self.plots_data[i].append(last_data)

            if last_data >= ymax:
                plot.set_ylim(self.plots_data[i].min, 1.1 * last_data)
            if last_data <= ymin:
                plot.set_ylim(last_data / 1.1, self.plots_data[i].max)

            self.lines[i].set_data(self.times, self.plots_data[i])

        if self.particles:
            for i, (_, monitor) in enumerate(self.particles):
                monitor.plot(data)

        if redraw:
            self.redraw()


class ParticleMonitor(object):
    data = None

    def __init__(self, particle, plots):
        raise NotImplementedError()

    def plot(self, data):
        raise NotImplementedError()

    def scatter(self, plot, x, y, *args, **kwargs):
        if not self.data:
            self.data = {}
        if plot not in self.data:
            self.data[plot] = [[], []]

        self.plots[plot].scatter(x, y, *args, **kwargs)
        self.data[plot][0].append(x)
        self.data[plot][1].append(y)

    def plot_function(self, plot, t, grid, foo, *args, **kwargs):
        if not self.data:
            self.data = {}
        if plot not in self.data:
            self.data[plot] = [[], [], []]

        self.plots[plot].plot(grid, foo, *args, **kwargs)
        self.data[plot][0].append(t)
        self.data[plot][1].append(grid)
        self.data[plot][2].append(foo)


class RadiationParticleMonitor(ParticleMonitor):
    def __init__(self, particle, plots):
        self.particle, self.plots = particle, plots

        self.plots[0].set_title(particle.name)
        self.plots[0].set_xlabel("T, MeV")
        self.plots[0].invert_xaxis()
        self.plots[0].set_xscale("log")
        self.plots[0].set_ylabel("rho/rho_eq")

        self.plots[1].set_xlabel("y, MeV")
        self.plots[1].set_xlim(self.particle.grid.MIN_MOMENTUM / UNITS.MeV,
                               self.particle.grid.MAX_MOMENTUM / UNITS.MeV)
        self.plots[1].set_ylabel("f/f_eq")

    def comparison_distributions(self, data):
        T = self.particle.params.T
        aT = self.particle.aT

        rhoeq = self.particle.energy_density / (
            self.particle.dof * numpy.pi**2 / 30. * (aT / self.particle.params.a)**4
            * (7./8. if self.particle.statistics == STATISTICS.FERMION else 1.)
        )
        feq = self.particle.equilibrium_distribution(aT=aT)

        return (T, rhoeq), feq

    def plot(self, data):
        (T, rhoeq), feq = self.comparison_distributions(data)

        if not self.particle.in_equilibrium:
            ratio = numpy.vectorize(self.particle.distribution)(self.particle.grid.TEMPLATE) / feq
        else:
            ratio = numpy.ones(self.particle.grid.TEMPLATE.shape)
            rhoeq = 1.

        self.scatter(0, T / UNITS.MeV, rhoeq, s=1)

        age_lines(self.plots[1].get_axes().lines)
        self.plot_function(1, T, self.particle.grid.TEMPLATE / UNITS.MeV, ratio)


class EquilibriumRadiationParticleMonitor(RadiationParticleMonitor):
    def comparison_distributions(self, data):
        T = self.particle.params.T
        aT = self.particle.params.aT
        # T = data['T'][-1]
        # aT = data['aT'][-1]

        rhoeq = self.particle.energy_density / (
            self.particle.dof * numpy.pi**2 / 30 * T**4
            * (7./8. if self.particle.statistics == STATISTICS.FERMION else 1.)
        )
        feq = self.particle.equilibrium_distribution(aT=aT)

        return (T, rhoeq), feq


class EffectiveTemperatureRadiationPartileMonitor(RadiationParticleMonitor):
    def comparison_distributions(self, data):
        rho = self.particle.energy_density
        const = (
            self.particle.dof * numpy.pi**2 / 30.
            * (7./8. if self.particle.statistics == STATISTICS.FERMION else 1.)
        )

        T = self.particle.params.T
        T_eff = (rho / const)**0.25
        aT = T_eff * self.particle.params.a

        rhoeq = rho / const / (self.particle.aT / self.particle.params.a)**4
        feq = self.particle.equilibrium_distribution(aT=aT)

        self.plots[1].set_title("T ~ {:3e}".format(T_eff / UNITS.MeV))

        return (T, rhoeq), feq


class MassiveParticleMonitor(ParticleMonitor):
    def __init__(self, particle, plots):
        self.particle, self.plots = particle, plots

        self.plots[0].set_title(particle.name)
        self.plots[0].invert_xaxis()
        self.plots[0].set_xlabel("T, MeV")
        self.plots[0].set_xscale("log")
        self.plots[0].set_ylabel("rho/(n M)")

        self.plots[1].set_xlabel("y, MeV")
        self.plots[1].set_xlim(self.particle.grid.MIN_MOMENTUM / UNITS.MeV,
                               self.particle.grid.MAX_MOMENTUM / UNITS.MeV)
        self.plots[1].set_ylabel("(f-f_eq) y^2")

    def plot(self, data):
        T = data['T'][-1]

        from particles.NonEqParticle import energy_density, density

        self.scatter(0, T / UNITS.MeV,
                     energy_density(self.particle) / (self.particle.mass * density(self.particle)),
                     s=1)

        age_lines(self.plots[1].get_axes().lines)

        yy = self.particle.grid.TEMPLATE * self.particle.grid.TEMPLATE / UNITS.MeV**2

        f = self.particle._distribution
        feq = self.particle.equilibrium_distribution()
        self.plot_function(1, T, self.particle.grid.TEMPLATE / UNITS.MeV, yy*(f-feq))


class EquilibrationMonitor(ParticleMonitor):
    def __init__(self, particle, plots):
        self.particle, self.plots = particle, plots

        self.plots[0].set_title(particle.name)
        self.plots[0].set_xlabel("a")
        self.plots[0].set_xscale("log")
        self.plots[0].set_ylabel("max|I|")

        self.plots[1].set_xlabel("a")
        self.plots[1].set_xscale("log")
        self.plots[1].set_ylabel("numerator, MeV^-1")

    def plot(self, data):
        a = data['a'][-1]

        from particles.NonEqParticle import numerator

        self.scatter(0, a, numpy.max(numpy.fabs(self.particle.collision_integral)) * UNITS.MeV, s=1)
        self.scatter(1, a, numerator(self.particle) * UNITS.MeV, s=1)


class AbundanceMonitor(ParticleMonitor):
    def __init__(self, particle, plots):
        self.particle, self.plots = particle, plots

        self.plots[0].set_title(particle.name)
        self.plots[0].set_xlabel("T, MeV")
        self.plots[0].invert_xaxis()
        self.plots[0].set_xscale("log")
        self.plots[0].set_yscale("log")
        self.plots[0].set_ylabel("rho fraction")

        self.plots[1].set_xlabel("T, MeV")
        self.plots[1].invert_xaxis()
        self.plots[1].set_xscale("log")
        self.plots[1].set_yscale("log")
        self.plots[1].set_ylabel("n a^3")

    def plot(self, data):
        T = data['T'][-1]

        total_rho = data['rho'][-1]
        rho = self.particle.energy_density
        self.scatter(0, T / UNITS.MeV, rho / total_rho, s=1)

        density = self.particle.density * self.particle.params.a**3 / UNITS.MeV**3
        self.scatter(1, T / UNITS.MeV, density, s=1)


class DensityAndEnergyMonitor(ParticleMonitor):
    def __init__(self, particle, plots):
        self.particle, self.plots = particle, plots

        self.plots[0].set_title(particle.name)
        self.plots[0].set_xlabel("T, MeV")
        self.plots[0].invert_xaxis()
        self.plots[0].set_xscale("log")
        self.plots[0].set_ylabel("rho a^4")

        self.plots[1].set_xlabel("T, MeV")
        self.plots[1].invert_xaxis()
        self.plots[1].set_xscale("log")
        self.plots[1].set_ylabel("n a^3")

    def plot(self, data):
        T = data['T'][-1]

        rho = self.particle.energy_density * self.particle.params.a**4 / UNITS.MeV**4
        self.scatter(0, T / UNITS.MeV, rho, s=1)

        density = self.particle.density * self.particle.params.a**3 / UNITS.MeV**3
        self.scatter(1, T / UNITS.MeV, density, s=1)


def age_lines(lines):
    """ Slightly decrease the opacity of plotted lines until they are barely visible.\
        Then, remove them. Saves up on memory and clears the view of the plots. """
    for line in lines:
        alpha = line.get_alpha() or 1.
        if alpha < 0.1:
            line.remove()
        else:
            line.set_alpha((line.get_alpha() or 1.) * 0.8)


def plot_integrand(integrand, name, p0, filename=__file__):
    """ Save a 3D plot of the distribution function integrand into a file. """

    from matplotlib import cm
    from mpl_toolkits.mplot3d import Axes3D

    fig = plt.figure(3)
    ax = fig.gca(projection='3d')
    plt.cla()
    X, Y = numpy.meshgrid(GRID.TEMPLATE, GRID.TEMPLATE)
    Z = numpy.array([integrand([x, y]) for x, y in zip(numpy.ravel(X), numpy.ravel(Y))])\
        .reshape(X.shape)

    ax.plot_surface(X, Y, Z, rstride=1, cstride=1, alpha=0.1)
    ax.contourf(X, Y, Z, zdir='z', offset=numpy.amin(Z), cmap=cm.coolwarm)
    ax.contourf(X, Y, Z, zdir='x', offset=ax.get_xlim()[0], cmap=cm.coolwarm)
    ax.contourf(X, Y, Z, zdir='y', offset=ax.get_ylim()[1], cmap=cm.coolwarm)

    ax.set_xlabel('p1')
    ax.set_ylabel('p2')
    ax.set_title('{} p0 = {}'.format(name, p0 / UNITS.MeV))

    plt.savefig(os.path.join(os.path.split(filename)[0], 'logs/plt_{}.svg'.format(p0 / UNITS.MeV)))


def plot_points(points, name):
    """ Draw a scatter plot for a number of `points` tuples `(x, y)` """
    plt.figure(4)
    plt.title(name)
    plt.scatter(*zip(*points))
    plt.show()

if __name__ == '__main__':

    import argparse
    parser = argparse.ArgumentParser(description='Monitor data files and plot')
    parser.add_argument('--folder', required=True)
    args = parser.parse_args()
    monitor_datafile(args.folder)