aboutsummaryrefslogtreecommitdiffstats
path: root/fg21sim/extragalactic/clusters/halo.py
blob: 072cdffe8912d6558be98c82647287aa77c80240 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
# Copyright (c) 2017 Weitian LI <liweitianux@live.com>
# MIT license

"""
Simulate (giant) radio halos following the "statistical
magneto-turbulent model" proposed by Cassano & Brunetti (2005).

References
----------
[1] Cassano & Brunetti 2005, MNRAS, 357, 1313
    http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
"""

import logging

import numpy as np
import astropy.units as au
import astropy.constants as ac
import scipy.interpolate
import scipy.integrate
import scipy.optimize

from .cosmology import Cosmology
from .solver import FokkerPlanckSolver


logger = logging.getLogger(__name__)


class HaloSingle:
    """
    Simulate a single (giant) radio halos following the "statistical
    magneto-turbulent model" proposed by Cassano & Brunetti (2005).

    First, simulate the cluster merging history from the extended
    Press-Schecter formalism using the Monte Carlo method; then derive
    the merger energy and turbulence energy as well as its spectrum;
    after that, calculate the electron acceleration and time  evolution
    by solving the Fokker-Planck equation; and finally derive the radio
    emission from the electron spectra.

    References
    ----------
    [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
        http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C

    Parameters
    ----------
    M0 : float
        Present-day (z=0) mass (unit: Msun) of the cluster.
    configs : `ConfigManager`
        A `ConfigManager` instance containing default and user configurations.
        For more details, see the example configuration specifications.

    Attributes
    ----------
    mec : float
        Unit for electron momentum (p): mec = m_e * c, p = gamma * mec,
        therefore value of p is the Lorentz factor.
    cosmo : `~Cosmology`
        Adopted cosmological model with custom utility functions.
    mtree : `~MergerTree`
        Merging history of this cluster.
    """
    # Merger tree (i.e., merging history) of this cluster
    mtree = None
    # Unit for electron momentum (p), thus its value is the Lorentz factor
    mec = ac.m_e.cgs.value*ac.c.cgs.value  # [g cm / s]
    # Mean molecular weight
    # Ref.: Ettori et al, 2013, Space Science Review, 177, 119-154, Eq.(6)
    mu = 0.6
    # Atomic mass unit (i.e., a.m.u.)
    m_atom = ac.u.cgs.value  # [g]
    # Common units conversion
    # TODO: move these to a separate module/class
    Msun2g = au.solMass.to(au.g)
    kpc2cm = au.kpc.to(au.cm)
    keV2erg = au.keV.to(au.erg)
    Gyr2s = au.Gyr.to(au.s)

    def __init__(self, M0, configs):
        self.M0 = M0  # [Msun]
        self.configs = configs
        self._set_configs()

    def _set_configs(self):
        """
        Set up the necessary class attributes according to the configs.
        """
        comp = "extragalactic/halos"
        self.zmax = self.configs.getn(comp+"/zmax")
        # Mass threshold of the sub-cluster for a significant merger
        self.merger_mass_th = self.configs.getn(comp+"/merger_mass_th")
        self.radius_halo = self.configs.getn(comp+"/radius_halo")
        self.magnetic_field = self.configs.getn(comp+"/magnetic_field")
        self.eta_t = self.configs.getn(comp+"/eta_t")
        self.eta_e = self.configs.getn(comp+"/eta_e")
        self.pmin = self.configs.getn(comp+"/pmin")
        self.pmax = self.configs.getn(comp+"/pmax")
        self.pgrid_num = self.configs.getn(comp+"/pgrid_num")
        self.buffer_np = self.configs.getn(comp+"/buffer_np")
        self.time_step = self.configs.getn(comp+"/time_step")
        self.injection_index = self.configs.getn(comp+"/injection_index")
        # Cosmology model
        self.H0 = self.configs.getn("cosmology/H0")
        self.OmegaM0 = self.configs.getn("cosmology/OmegaM0")
        self.cosmo = Cosmology(H0=self.H0, Om0=self.OmegaM0)
        logger.info("Loaded and set up configurations")

    def simulate_mergertree(self):
        """
        Simulate the merging history of the cluster using the extended
        Press-Schechter formalism.
        """
        raise NotImplementedError

    def calc_electron_spectrum(self):
        """
        Calculate the relativistic electron spectrum by solving the
        Fokker-Planck equation.
        """
        fpsolver = FokkerPlanckSolver(
            xmin=self.pmin, xmax=self.pmax,
            grid_num=self.pgrid_num,
            buffer_np=self.buffer_np,
            tstep=self.time_step,
            f_advection=self.fp_advection,
            f_diffusion=self.fp_diffusion,
            f_injection=self.fp_injection,
        )
        p = fpsolver.x
        # Assume NO initial electron distribution
        n0_e = np.zeros(p.shape)
        tstart = self.cosmo.age(self.zmax)
        tstop = self.cosmo.age0
        n_e = fpsolver.solve(u0=n0_e, tstart=tstart, tstop=tstop)
        return (p, n_e)

    def kT_mass(self, mass):
        """
        Estimate the cluster ICM temperature from its mass by assuming
        an (observed) temperature-mass relation.

        TODO: upgrade this M-T relation.

        Parameters
        ----------
        mass : float
            Mass (unit: Msun) of the cluster

        Returns
        -------
        kT : float
            Temperature of the ICM (unit: keV)

        References
        ----------
        [1] Nevalainen et al. 2000, ApJ, 532, 694
            Ettori et al, 2013, Space Science Review, 177, 119-154
            NOTE: H0 = 50 * h50 [km/s/Mpc]
        """
        kT = 10 * (mass/1.23e15) ** (1/1.79)  # [keV]
        return kT

    def _radius_virial(self, mass, z=0.0):
        """
        Calculate the virial radius of a cluster.

        Parameters
        ----------
        mass : float
            Mass (unit: Msun) of the cluster
        z : float
            Redshift

        Returns
        -------
        Rvir : float
            Virial radius (unit: kpc) of the cluster at given redshift
        """
        Dc = self.cosmo.overdensity_virial(z)
        rho = self.cosmo.rho_crit(z)  # [g/cm^3]
        R_vir = (3*mass*self.Msun2g / (4*np.pi * Dc * rho))**(1/3)  # [cm]
        R_vir /= self.kpc2cm  # [kpc]
        return R_vir

    def _radius_stripping(self, mass, M_main, z):
        """
        Calculate the stripping radius of the sub-cluster at which
        equipartition between static and ram pressure is established,
        and the stripping is efficient outside this stripping radius.

        Note that the value of the stripping radius obtained would
        give the *mean value* of the actual stripping radius during
        a merger.

        Parameters
        ----------
        mass : float
            The mass (unit: Msun) of the sub-cluster.
        M_main : float
            The mass (unit: Msun) of the main cluster.
        z : float
            Redshift

        Returns
        -------
        rs : float
            The stripping radius of the sub-cluster.
            Unit: kpc

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(11)
        """
        vi = self._velocity_impact(M_main, mass, z) * 1e5  # [cm/s]
        kT = self.kT_mass(mass) * self.keV2erg  # [erg]
        coef = kT / (self.mu * self.m_atom * vi**2)  # dimensionless
        rho_avg = self._density_average(M_main, z)  # [g/cm^3]

        def equation(r):
            return coef * self.density_profile(r, mass, z) / rho_avg - 1

        r_vir = self._radius_virial(mass, z)  # [kpc]
        rs = scipy.optimize.brentq(equation, a=0, b=r_vir)  # [kpc]
        return rs

    def _density_average(self, mass, z=0.0):
        """
        Average density of the cluster ICM.

        Returns
        -------
        rho : float
            Average ICM density (unit: g/cm^3)

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(12)
        """
        f_baryon = self.cosmo.Ob0 / self.cosmo.Om0
        Rv = self._radius_virial(mass, z) * self.kpc2cm  # [cm]
        V = (4*np.pi / 3) * Rv**3  # [cm^3]
        rho = f_baryon * mass*self.Msun2g / V  # [g/cm^3]
        return rho

    def density_profile(self, r, mass, z):
        """
        ICM (baryon) density profile, assuming the beta model.

        Parameters
        ----------
        r : float
            Radius (unit: kpc) where to calculate the density
        mass : float
            Cluster mass (unit: Msun)
        z : float
            Redshift

        Returns
        -------
        rho_r : float
            Density at the specified radius (unit: g/cm^3)

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(13)
        """
        f_baryon = self.cosmo.Ob0 / self.cosmo.Om0
        M_ICM = mass * f_baryon * self.Msun2g  # [g]
        r *= self.kpc2cm  # [cm]
        Rv = self._radius_virial(mass, z) * self.kpc2cm  # [cm]
        rc = self._beta_rc(Rv)
        beta = self._beta_beta()
        norm = self._beta_norm(M_ICM, beta, rc, Rv)  # [g/cm^3]
        rho_r = norm * (1 + (r/rc)**2) ** (-3*beta/2)  # [g/cm^3]
        return rho_r

    @staticmethod
    def _beta_rc(r_vir):
        """
        Core radius of the beta model for the ICM density profile.

        TODO: upgrade this!
        """
        return 0.1*r_vir

    @staticmethod
    def _beta_beta():
        """
        Beta value of the beta model for the ICM density profile.

        TODO: upgrade this!
        """
        return 0.8

    @staticmethod
    def _beta_norm(mass, beta, rc, r_vir):
        """
        Calculate the normalization of the beta model for the ICM
        density profile.

        Parameters
        ----------
        mass : float
            The mass (unit: g) of ICM
        beta : float
            Beta value of the assumed beta profile
        rc : float
            Core radius (unit: cm) of the assumed beta profile
        r_vir : float
            The virial radius (unit: cm) of the cluster

        Returns
        -------
        norm : float
            Normalization of the beta model (unit: g/cm^3)

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(14)
        """
        integration = scipy.integrate.quad(
            lambda r: r*r * (1+(r/rc)**2) ** (-3*beta/2),
            0, r_vir)[0]
        norm = mass / (4*np.pi * integration)  # [g/cm^3]
        return norm

    def _velocity_impact(self, M_main, M_sub, z=0.0):
        """
        Calculate the relative impact velocity between the two merging
        clusters when they are at a distance of virial radius.

        Parameters
        ----------
        M_main : float
            Mass of the main cluster (unit: Msun)
        M_sub : float
            Mass of the sub cluster (unit: Msun)
        z : float
            Redshift

        Returns
        -------
        vi : float
            Relative impact velocity (unit: km/s)

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(9)
        """
        eta_v = 4 * (1 + M_main/M_sub) ** (1/3)
        R_vir = self._radius_virial(M_main, z) * self.kpc2cm  # [cm]
        G = ac.G.cgs.value
        vi = np.sqrt(2*G * (1-1/eta_v) *
                     (M_main+M_sub)*self.Msun2g / R_vir)  # [cm/s]
        vi /= 1e5  # [km/s]
        return vi

    def _time_crossing(self, M_main, M_sub, z):
        """
        Calculate the crossing time of the sub-cluster during a merger.

        Parameters
        ----------
        M_main : float
            Mass of the main cluster (unit: Msun)
        M_sub : float
            Mass of the sub cluster (unit: Msun)
        z : float
            Redshift where the merger occurs.

        Returns
        -------
        time : float
            Crossing time (unit: Gyr)
        """
        R_vir = self._radius_virial(M_main, z)  # [kpc]
        vi = self._velocity_impact(M_main, M_sub, z)  # [km/s]
        # Unit conversion coefficient: [s kpc/km] => [Gyr]
        # uconv = au.kpc.to(au.km) * au.s.to(au.Gyr)
        uconv = 0.9777922216731284
        time = uconv * R_vir / vi  # [Gyr]
        return time

    def _z_end(self, z_begin, time):
        """
        Calculate the ending redshift from ``z_begin`` after elapsing
        ``time``.

        Parameters
        ----------
        z_begin : float
            Beginning redshift
        time : float
            Elapsing time (unit: Gyr)
        """
        t_begin = self.cosmo.age(z_begin)  # [Gyr]
        t_end = t_begin + time
        if t_end >= self.cosmo.age(0):
            z_end = 0.0
        else:
            z_end = self.cosmo.redshift(t_end)
        return z_end

    @property
    def merger_events(self):
        """
        Trace only the main cluster, and filter out the significant
        merger events.

        Returns
        -------
        mevents : list[dict]
            List of dictionaries that records all the merger events
            of the main cluster.
            NOTE:
            The merger events are ordered by increasing redshifts.
        """
        events = []
        tree = self.mtree
        while tree:
            if (tree.major and tree.minor and
                    tree.minor.node.mass >= self.merger_mass_th and
                    tree.major.node.z <= self.zmax):
                events.append({
                    "M_main": tree.major.node.mass,
                    "M_sub": tree.minor.node.mass,
                    "z": tree.major.node.z,
                    "age": tree.major.node.age
                })
            tree = tree.major
        return events

    def _coef_acceleration(self, z):
        """
        Calculate the electron-acceleration coefficient at arbitrary
        redshift, by interpolating the coefficients calculated at every
        merger redshifts.
        """
        if not hasattr(self, "_coef_acceleration_interp"):
            # Order the merger events by decreasing redshifts
            mevents = list(reversed(self.merger_events))
            redshifts = np.array([ev["z"] for ev in mevents])
            chis = np.array([self._chi_at_zidx(zidx, mevents)
                             for zidx in range(len(redshifts))])
            self._coef_acceleration_interp = scipy.interpolate.interp1d(
                    redshifts, chis, kind="linear",
                    bounds_error=False, fill_value=0.0)
            logger.info("Interpolated acceleration coefficients w.r.t. z")
        return self._coef_acceleration_interp(z)

    def _chi_at_zidx(self, zidx, mevents):
        """
        Calculate electron-acceleration coefficient at the specified
        merger event which is specified with a redshift index.

        Parameters
        ----------
        zidx : int
            Index of the redshift where to calculate the coefficient.
        mevents : list[dict]
            List of dictionaries that records all the merger events
            of the main cluster.
            NOTE:
            The merger events should be ordered by increasing time
            (or decreasing redshifts).

        Returns
        -------
        chi : float
            The calculated electron-acceleration coefficient.
            (unit: Gyr^-1)

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(40)
        """
        redshifts = np.array([ev["z"] for ev in mevents])
        zbegin = mevents[zidx]["z"]
        M_main = mevents[zidx]["M_main"]
        M_sub = mevents[zidx]["M_sub"]
        t_crossing = self._time_crossing(M_main, M_sub, zbegin)
        zend = self._z_end(zbegin, t_crossing)
        try:
            zend_idx = np.where(redshifts < zend)[0][0]
        except IndexError:
            # Specified redshift already the last/smallest one
            zend_idx = zidx + 1
        #
        coef = 2.23e-16 * self.eta_t / (self.radius_halo/500)**3  # [s^-1]
        coef *= self.Gyr2s  # [Gyr^-1]
        chi = 0.0
        for ev in mevents[zidx:zend_idx]:
            M_main = ev["M_main"]
            M_sub = ev["M_sub"]
            z = ev["z"]
            R_vir = self._radius_virial(M_main, z)
            rs = self._radius_stripping(M_sub, M_main, z)
            kT = self.kT_mass(M_main)
            term1 = ((M_main+M_sub)/2e15 * (2.6e3/R_vir)) ** (3/2)
            term2 = (rs/500)**2 / np.sqrt(kT/7)
            if rs <= self.radius_halo:
                term3 = 1.0
            else:
                term3 = (self.radius_halo/rs) ** 2
            chi += coef * term1 * term2 * term3
        return chi

    def fp_injection(self, p, t=None):
        """
        Electron injection term for the Fokker-Planck equation.

        The injected electrons are assumed to have a power-law spectrum
        and a constant injection rate.

        Qe(p) = Ke * (p/pmin)**(-s)
        Ke = ((s-2)*eta_e) * (e_th/(pmin*c)) / (t0*pmin)

        Parameters
        ----------
        p : float
            Electron momentum (unit: mec), i.e., Lorentz factor
        t : None
            Currently a constant injection rate is assumed, therefore
            this parameter is not used.  Keep it for the consistency
            with other functions.

        Returns
        -------
        Qe : float
            Current electron injection rate at specified energy (p).
            Unit: [cm^-3 Gyr^-1 mec^-1]

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eqs.(31-33)
        """
        if not hasattr(self, "_electron_injection_rate"):
            e_th = self.e_thermal  # [erg/cm^3]
            term1 = (self.injection_index-2) * self.eta_e
            term2 = e_th / (self.pmin * self.mec * ac.c.cgs.value)  # [cm^-3]
            term3 = 1.0 / (self.cosmo.age0 * self.pmin)  # [Gyr^-1 mec^-1]
            Ke = term1 * term2 * term3
            self._electron_injection_rate = Ke
        else:
            Ke = self._electron_injection_rate
        Qe = Ke * (p/self.pmin) ** (-self.injection_index)
        return Qe

    def fp_diffusion(self, p, t):
        """
        Diffusion term/coefficient for the Fokker-Planck equation.

        Parameters
        ----------
        p : float
            Electron momentum (unit: mec), i.e., Lorentz factor
        t : float
            Current time when solving the equation (unit: Gyr)

        Returns
        -------
        Dpp : float
            Diffusion coefficient
            Unit: [mec^2/Gyr]

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(36)
        [2] Donnert 2013, AN, 334, 615
            http://adsabs.harvard.edu/abs/2013AN....334..515D
            Eq.(15)
        """
        z = self.cosmo.redshift(t)
        chi = self._coef_acceleration(z)  # [Gyr^-1]
        # NOTE: Cassano & Brunetti's formula misses a factor of 2.
        Dpp = chi * p**2 / 4  # [mec^2/Gyr]
        return Dpp

    def fp_advection(self, p, t):
        """
        Advection term/coefficient for the Fokker-Planck equation,
        which describes a systematic tendency for upward or downard
        drift of particles.

        This term is also called the "generalized cooling function" by
        Donnert & Brunetti (2014), which includes all relevant energy
        loss functions and the energy gain function due to turbulence.

        Returns
        -------
        Hp : float
            Advection coefficient
            Unit: [mec/Gyr]

        References
        ----------
        [1] Donnert & Brunetti 2014, MNRAS, 443, 3564
            http://adsabs.harvard.edu/abs/2014MNRAS.443.3564D
            Eq.(15)
        [2] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eqs.(30,36,38,39)
        """
        Hp = (abs(self._dpdt_ion(p, t)) +
              abs(self._dpdt_rad(p, t)) -
              (self.fp_diffusion(p, t) * 2 / p))
        return Hp

    def _dpdt_ion(self, p, t):
        """
        Energy loss through ionization and Coulomb collisions.

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(38)
        """
        z = self.cosmo.redshift(t)
        n_th = self._n_thermal(self.M0, z)
        coef = -3.3e-29 * self.Gyr2s / self.mec  # [mec/Gyr]
        dpdt = coef * n_th * (1 + np.log(p/n_th) / 75)
        return dpdt

    def _dpdt_rad(self, p, t):
        """
        Energy loss via synchrotron emission and IC scattering off the CMB.

        References
        ----------
        [1] Cassano & Brunetti 2005, MNRAS, 357, 1313
            http://adsabs.harvard.edu/abs/2005MNRAS.357.1313C
            Eq.(39)
        """
        z = self.cosmo.redshift(t)
        coef = -4.8e-4 * self.Gyr2s / self.mec  # [mec/Gyr]
        dpdt = (coef * (p*self.mec)**2 *
                ((self.magnetic_field/3.2)**2 + (1+z)**4))
        return dpdt

    @property
    def e_thermal(self):
        """
        Calculate the present-day thermal energy density of the ICM.

        Returns
        -------
        e_th : float
            Energy density of the ICM (unit: erg/cm^3)
        """
        mass = self.M0
        f_baryon = self.cosmo.Ob0 / self.cosmo.Om0
        kT = self.kT_mass(mass)  # [keV]
        N = mass * self.Msun2g * f_baryon / (self.mu * self.m_atom)
        E_th = kT*self.keV2erg * N  # [erg]
        Rv = self._radius_virial(mass) * self.kpc2cm  # [cm]
        V = (4*np.pi / 3) * Rv**3  # [cm^3]
        e_th = E_th / V  # [erg/cm^3]
        return e_th

    def _n_thermal(self, mass, z=0.0):
        """
        Calculate the present-day number density of the ICM thermal plasma.

        Parameters
        ----------
        mass : float
            Mass (unit: Msun) of the cluster
        z : float
            Redshift

        Returns
        -------
        n_th : float
            Number density of the ICM (unit: cm^-3)
        """
        f_baryon = self.cosmo.Ob0 / self.cosmo.Om0
        N = mass * self.Msun2g * f_baryon / (self.mu * self.m_atom)
        Rv = self._radius_virial(mass, z) * self.kpc2cm  # [cm]
        V = (4*np.pi / 3) * Rv**3  # [cm^3]
        n_th = N / V  # [cm^-3]
        return n_th