The Energy Shear of Protohaloes: Marcello Musso, Giulia Despali, and Ravi K. Sheth
The Energy Shear of Protohaloes: Marcello Musso, Giulia Despali, and Ravi K. Sheth
The Energy Shear of Protohaloes: Marcello Musso, Giulia Despali, and Ravi K. Sheth
ABSTRACT
As it collapses to form a halo, the shape of a protohalo patch is deformed by the initial shear field. This deformation
is often modeled using the ‘deformation’ tensor, constructed from second derivatives of the gravitational potential,
whose trace gives the initial overdensity. However, especially for lower mass protohalos, this matrix is not always
positive definite: One of its eigenvalues has a different sign from the others. We show that the evolution of a patch is
better described by the ‘energy shear’ tensor, which is positive definite and plays a direct role in the evolution. This
positive-definiteness simplifies models of halo abundances, assembly and of the cosmic web.
Key words: large-scale structure of Universe
1
Z with ϵ ≡ tr(u), which corresponds (at first order in pertur-
Kij ≡ d3 rρ(r, t)(ṙi − ṙcm,i )(ṙj − ṙcm,j ) , (3) bations) to the result that the collapse of the inertial radius
2 V
Z RI is well described by spherical collapse with overdensity ϵ.
Uij ≡ − d3 rρ(r, t)(ri − rcm,i )(∇j Φ − [∇j Φ]cm,j ) , (4) This is the basis for identifying protohalos with local maxima
V of the energy overdensity (Musso & Sheth 2021), since these
where ∇Φ is the gravitational attraction due to matter, so are minima of the collapse time of RI .
that r̈ = −∇Φ + (Λ/3)r, and [∇Φ]cm is the one of the center Naively, for a halo to form, one might imagine that all the
of mass. three axes of the inertia tensor Iij should collapse at the same
To describe the evolution of Iij with respect to the back- time as RI . Equation (8) would then require that
ground, we split the matter gravitational potential as ∇Φ =
u + uT
4πGρ̄(r/3 + ∇ϕ), where ϕ is the potential perturbation obey- AAT ≃ tr(AAT ) , (11)
ing ∇2 ϕ = δm , and δm = (ρ/ρ̄) − 1 is the matter density 2ϵ
perturbation. We then rewrite equation (2) as in which case one would have
··
Iij
·
Iij 4πGρ̄ I M 2 u + uT ϵ u + uT
+ 2H = 4kij − (uij + uji ) 2 , (5) Iij ≃ a tr(AAT ) 1 − 2D ≃ I, (12)
a2 a2 3 a 5 2ϵ 3 2ϵ
predicting that the eigenvectors of Iij and of uij + uji are
where a is the scale factor, H = ȧ/a is the Hubble parameter,
to be aligned, and their eigenvalues proportional. This pic-
I ≡ Ikk is the trace of Iij ,
ture is not entirely correct: we know in fact that protohalo
· ·
boundaries tend to follow equipotential surfaces (Musso &
Z
1 ri − rcm,i rj − rcm,j
kij ≡ d3 rρ(r, t) (6)
2 V a a Sheth 2023), which gives an independent prescription for Iij .
Figure 1. Left panel. Eigenvalues of the potential energy overdensity tensor uij . Only 5% of protohaloes have one negative eigenvalue; this
is our main result. Right panel. Eigenvalues of the mean deformation tensor. The lowest eigenvalue is negative in about 40% of the all
protohaloes, and in more than 50% of the lower mass ones, in stark contrast to the left panel.
7 1, Flora 1, Flora
2, Flora 2, Flora
6 3, Flora 2.0 3, Flora
1, Bice
5 1, Bice
2, Bice
3, Bice 1.5 2, Bice
4
P( i/ 02)
P( i/ 02)
3, Bice
3 1.0
2
0.5
1
0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 1.0 0.5 0.0 0.5 1.0 1.5 2.0
i/ 02 i/ 02
Figure 2. Left panel. Distribution of the (normalized) eigenvalues λi of the potential energy overdensity tensor. Filled and open histograms
show results for all sampled protohaloes having masses larger than 4 × 1014 h−1 M⊙ (from Flora) and larger than 1011 h−1 M⊙ (from Bice),
respectively. Right panel. Same, but now for the eigenvalues λ̄i of the traceless part of the potential energy overdensity tensor. Filled and
open histograms are very similar, indicating that differences in the left panet are mainly driven by the mass-dependence of the trace.
defined in equations (1) and (7): initial displacement field Ψ, and estimate the mean deforma-
NH
tion tensor (13) as:
X
Iˆij ≡ (q (n) − qcm )i (q (n) − qcm )j (21) NG
1 X ∂Ψj
n=1 q̂ij ≡ , (23)
P NH NG ∂qi k
(n) (n) k=1
n=1 (q − qcm )i (v − vcm )j /f DH
ûij ≡ −3 PNH (n) , (22) where NG is the number of grid cells contained within the La-
n=1 |q − qcm |2
grangian protohalo volume. This means that we also include
where D is the ΛCDM density perturbation growth factor (at empty grid cells (i.e. those not occupied by halo particles)
the redshift z of the snapshot), f = d ln D/d ln a, and n runs that are located inside the Lagrangian volume, because they
over the NH protohalo particles. Note that Iij is not divided too contribute to the total potential field that acts on the
by NH . To obtain accelerations from velocities we used the protohalo. This slightly refines the measurement one obtains
Zel’dovich approximation, in which ∇ϕ ≃ v/Ḋ = v/f DH. if one uses the particle grid points only (see Despali et al.
For each protohalo we also measure the mean deforma- 2013, for more details).
tion tensor, defined in equation (13) as the average over the The left panel of Fig. 1 shows the eigenvalues λ1 , λ2 and λ3
protohalo particles of the second spatial gradient of the grav- of ûij + ûji as a function of halo mass. In the range of masses
itational potential at each position. Since we need to take one that we sampled, only one out of the 5378 Flora protohaloes
more derivative, instead of getting the accelerations from the had one slightly negative eigenvalue. For Bice protohaloes,
particle velocities (evaluated at irregular positions), we com- which extend to much lower masses, this fraction rises to 0.5%
pute the gradient of the initial displacements that were im- including only haloes mor massive than 1012 h−1 M⊙ , and to
posed to create the initial conditions with N-GenIC (Springel nearly 5% for all haloes in the sample. Nevertheless, this
et al. 2005), which are computed on the grid. In practice, we is substantially smaller than the fractions shown in Despali
select the grid points occupied by each protohalo, take their et al. (2013) for density-based statistics – i.e. the eigenvalues
15 15
10 10
5 5
0 0
15 Flora 15 Flora
P[cos( ij)]
P[cos( ij)]
Bice 10 Bice
10
5 5
0 0
15 15
10 10
5 5
0 0
0 1 0 1 0 1 0 1 0 1 0 1
cos( ij) cos( ij)
Figure 3. Cosine of the angle θij between the i-th eigenvector of the inertia tensor and the j-th one of the energy shear (left) or of the
deformation tensor (right). The alignment with the energy shear is stronger.
3.5 0.8
R1, Flora Flora
3.0 R2, Flora 0.7 Bice
R3, Flora 2
5
2.5 R1, Bice 0.6
R2, Bice
R3, Bice 0.5
2.0
2)
P(qu2/ 02
P(Ri)
0.4
1.5
0.3
1.0
0.2
0.5 0.1
0.0 0.0
0.5 0.0 0.5 1.0 1.5 2.0 0 2 4 6 8 10
ratio of eigenvalues Ri qu2/ 2
02
Figure 5. Ratios of the i-th eigenvalue λi of the energy shear uij 2 /σ 2 for Flora (filled) and Bice (open)
Figure 6. Distribution of qu 02
to the i-th eigenvalue a2i of the inertia tensor Iij , both normalized protohaloes; smooth curve shows the expected distribution for un-
to their traces: Ri ≡ (λi /ϵ)/(a2i /I). Equation (12) would predict constrained positions (a χ2 distribution with 5 degrees of freedom).
this ratio to be 1, whereas on average R1 > 1 and R3 < 1, mean-
ing that protohaloes are slightly less elongated than this simplest
prediction.