The Mass of Dark Matter | A Geometric Interpretation in SSM

Dark Matter as the Torsional Strain of a Discrete Chiral

Vacuum:

Deriving Galactic Halos, the Tully-Fisher Relation, and

the a

Acceleration Scale from Cosserat Elasticity

Raghu Kulkarni

SSMTheory Group, IDrive Inc., Calabasas, CA 91302, USA

Abstract

Standard cosmology requires the existence of invisible, collisionless Cold Dark Matter (CDM)

to account for ﬂat galactic rotation curves [1] and the gravitational lensing of galaxy clusters

[2]. Despite decades of experimental searches, no such particle has been found. In the

Selection-Stitch Model (SSM), the vacuum emerges as a Face-Centered Cubic (FCC, K =

12) tensor network with a fundamental lattice spacing on the order of the Planck scale

(L ∼ l

) [9]. This letter proposes that Dark Matter is not a particle, but the macroscopic

torsional strain of this discrete chiral vacuum. We explicitly derive the Chiral Cosserat

Lagrangian from the lattice structure tensors (S

trans

= 4, S

tors

= 8), demonstrating that

baryonic matter—structurally acting as K = 4 topological defects with exactly 8 missing

torsional bonds—physically twists the tensor network. We analytically prove that the 1D

screw dislocations of these defects geometrically superpose to form a 3D spherical dark mass

halo M

halo

∝ r, a mechanism explicitly veriﬁed via computational simulations of the discrete

K = 12 lattice. By applying standard random-walk statistics to the topological charges

of the macroscopic baryonic gas, we derive the

√

halo scaling necessary to recover the

baryonic Tully-Fisher relation (v

∝ M

) [4]. Furthermore, we identify the structural origin

of the MOND acceleration scale (a

= cH

/2π) [3] as the geometric boundary condition of

this torsional ﬁeld interacting with the cosmic horizon. Finally, we demonstrate that during

galactic collisions (e.g., the Bullet Cluster), these torsional vacuum waves cross without

electromagnetic friction, eﬀectively reproducing the dynamics of collisionless dark matter.

Keywords: Dark Matter, Cosserat Elasticity, Tully-Fisher Relation, Modiﬁed Newtonian

Dynamics, Topological Defects, FCC Lattice

1. Introduction

The missing mass problem remains one of the most persistent challenges in modern as-

trophysics. Observations of ﬂat galactic rotation curves [1] and the decoupled gravitational

lensing of colliding galaxy clusters [2] strongly imply that luminous baryonic matter is sur-

rounded by massive, invisible halos. The standard ΛCDM model posits that this mass is

Email address: raghu@idrive.com (Raghu Kulkarni)

composed of weakly interacting massive particles. Extensive direct detection eﬀorts, however,

have yielded null results for over four decades. Conversely, Modiﬁed Newtonian Dynamics

(MOND) [3] explains these rotation curves by phenomenologically altering the laws of grav-

ity at a characteristic acceleration scale a

≈ 1.2 × 10

−10

m/s

. While MOND ﬁts isolated

galaxies well, it encounters diﬃculties explaining the Bullet Cluster (1E 0657-558) [2], where

the gravitational lensing mass cleanly separates from the interacting baryonic gas during a

cluster collision.

The Selection-Stitch Model (SSM) oﬀers a geometric framework that uniﬁes these paradigms.

In recent work, we established that the macroscopic vacuum is an emergent Face-Centered

Cubic (FCC) tensor network governed by strict topological and kinematic saturation limits

[9, 10]. In this letter, we demonstrate that “Dark Matter” can be understood purely as the

macroscopic torsional strain of this discrete vacuum, generated by the structural mismatch

of baryonic defects.

Interactive 3D visualizations. Readers can explore the geometric phase transitions

and topological defects discussed in this paper through two interactive WebGL applica-

tions:

1. Vacuum Phase Transitions: The K = 6 → K = 4 → K = 12 topological

relaxation, explicitly illustrating the emergence of the translational and torsional lattice

channels:

https://raghu91302.github.io/ssmtheory/ssm_regge_deficit.html

2. Topological Baryons and Torsional Deﬁcits: A 3D geometric model of the K = 4

tetrahedral void trapped within the K = 12 bulk, demonstrating the ∆S

tors

= 8 missing

torsional bonds that source the dark matter halo:

https://raghu91302.github.io/ssmtheory/ssm_quark_structure.html

2. The Cosserat Lagrangian from the K = 12 Lattice

Building upon the geometric structure of the saturated K = 12 network [10], the vacuum

is treated in the continuum limit not as a simple Cauchy elastic solid, but strictly as a Chiral

Micropolar (Cosserat) solid [5].

2.1. The Lattice Degrees of Freedom: u and θ

In standard Cauchy elasticity, space is idealized as a continuum of mathematical points

possessing only translational displacement (u), which mediates conventional Newtonian grav-

ity. However, in the discrete K = 12 FCC lattice, each node is a physical vertex deﬁned by

the speciﬁc angular orientation of its 12 interconnecting bonds. Therefore, an accurate con-

tinuum description must include an independent rotational/torsional ﬁeld (θ) to account for

the internal twisting of these oriented nodes. A node can rotate without displacing its center

of mass (u = 0, θ = 0) or displace without rotating (u = 0, θ = 0).

2.2. The Structure Tensor Decomposition

The K = 12 FCC lattice bonds structurally decompose into two independent sectors [10]:

K = S

trans

+ S

tors

= 4 + 8 (1)

The translational sector (S

trans

= 4): The 4 bonds that form the rigid tetrahedral sublattice

of the FCC. These govern translational elastic response (compression, shear, and classical

wave propagation). The torsional sector (S

tors

= 8): The remaining 8 bonds that connect to

the octahedral neighbors. These govern the rotational elastic response (micropolar twist and

torsional strain).

2.3. Derivation of the Cosserat Lagrangian

The elastic potential energy of the lattice is the sum of bond-stretching (translational)

and bond-twisting (torsional) energies. In the continuum limit (r ≫ L), the static elastic

energy density becomes the standard Cosserat formulation [6]:

V =

(∇ · u)

+ µ (∇u)

+ κ (∇ × u − 2θ)

+ γ (∇θ)

(2)

Each term has a direct analytical lattice origin:

• Lamé term (λ/2)(∇ · u)

: Compression energy from changing unit cell volume. λ ∝

trans

because macroscopic compression strains all 4 translational bonds.

• Shear term µ(∇u)

: Distorting the cell shape without volume change. µ ∝ S

trans

• Coupling term κ(∇×u−2θ)

: The energy cost of a mismatch between the macroscopic

rotation (∇× u/2) and the actual microrotation (θ). κ ∝ (S

trans

×S

tors

)/K because it

bridges the sectors.

• Twist term γ(∇θ)

: The energy cost of spatial gradients in the microrotation ﬁeld.

γ ∝ S

tors

because torsional gradients strictly strain the 8 octahedral bonds.

Because stable baryonic particles are strictly modeled as trapped K = 4 tetrahedral

defects with intrinsic geometric handedness [14], their structural propagation dynamically

couples the translational and rotational modes. The full dynamic Lattice Lagrangian Density

L therefore requires kinetic terms and a chiral cross-coupling Ω:

L =

(∂

θ)

− V(u, θ) + Ω(u

θ − θ ˙u) (3)

2.4. The Modulus Hierarchy (Why Torsion is "Dark")

The physical consequence of the 4+8 decomposition is a strict hierarchy of elastic moduli:

λ + 2µ

∼

tors

trans

64/144

≈ 0.028 (4)

The twist modulus γ is an order of magnitude smaller than the bulk modulus. This is why

torsional strain is “dark”: it stores energy (gravitates and curves spacetime) but interacts

extremely weakly with translational bond-stretching (which mediates electromagnetism).

3. Topological Deﬁcits and the 1/r

Halo

In standard 3D Cosserat elasticity, the couple-stress from an isolated point source de-

cays rapidly as 1/r

, yielding a highly localized energy density γ(∇θ)

∝ 1/r

that cannot

physically account for extended galactic halos.

However, baryonic particles in the SSM are not point sources; they are extended topo-

logical defects. Speciﬁcally, a stable baryon is a trapped K = 4 tetrahedral phase remnant

[14]. Because the 3 internal bonds of this defect are strictly translational (forming the rigid

tetrahedral frame), it leaves zero internal torsional bonds. The torsional deﬁcit exposed to

the bulk is exactly:

∆S

tors

= S

bulk

tors

− S

bound

tors

= 8 − 0 = 8 missing bonds (5)

Consequently, every baryon is born with its dark matter halo because the halo is the macro-

scopic strain ﬁeld compensating for these missing torsional bonds.

In continuum mechanics, the extended ﬂux tubes of these baryonic defects manifest math-

ematically as 1D chiral screw dislocations threading the tensor network. To derive the analyt-

ical energy density, consider a single microscopic screw dislocation aligned along the z-axis.

Using standard cylindrical coordinates (r

cyl

, ϕ, z), the translational displacement u is strictly

longitudinal along the z-axis, but its magnitude increases monotonically with the azimuthal

angle ϕ:

2π

ϕ (6)

where b ∼ L is the fundamental Burgers vector bounded by the discrete lattice spacing. The

resulting localized shear strain is deﬁned by the spatial gradient of this displacement along

the azimuthal path:

ϕz

cyl

∂u

∂ϕ

2πr

cyl

(7)

The stored elastic energy density of this 1D defect is proportional to the square of the shear

strain (

µγ

), yielding an exact analytical proﬁle that falls oﬀ as 1/r

cyl

in the transverse

plane:

cyl

) =

µb

8π

cyl

(8)

To rigorously verify this analytical continuum limit within the strict discrete geometry of

the SSM, we computationally modeled an exact Volterra screw dislocation explicitly on a K =

12 FCC lattice (see Appendix A for full methodology and Python code). By evaluating the

discrete elastic strain energy across nearest-neighbor bonds (E

bond

|∆u|

), the simulation

conﬁrms that the per-node strain energy density closely follows the analytically predicted

ρ ∝ 1/r

decay (Fig. 1a). Integrating this discrete lattice energy radially yields an enclosed

halo mass M

halo

∝ r (Fig. 1b), which explicitly and unavoidably generates the ﬂat rotation

curves (v(r) ≈ const) observed in galactic kinematics (Fig. 1c).

To transition from a single 1D defect to a 3D macroscopic galaxy, we treat the galaxy

as a spherical gas of N ≈ 10

highly tangled, stochastically oriented baryonic braids. At

macroscopic galactic radii (r ∼ 10 kpc), the spatial superposition of these randomly oriented

1D line defects geometrically isotropizes the cylindrical ﬁelds. Averaging the transverse 1/r

cyl

Figure 1: Computational simulation of a Volterra screw dislocation on the discrete K = 12 FCC lattice.

(a) The per-node strain energy density strictly follows the analytically derived ρ ∝ r

−2

decay (measured at

−2.11

due to ﬁnite-size boundary eﬀects). (b) The enclosed dark mass scales as M

halo

∝ r

0.74

, converging

toward the r

theoretical limit. (c) The resulting orbital velocity yields a ﬂat rotation curve (v(r) ≈ const).

(d) Scatter plot of individual node strain energies conﬁrming isotropic 1/r

dispersion.

energy density over all solid angles for a quasi-spherical distribution of ﬂux tubes analytically

yields a strictly spherical 3D macroscopic energy density proﬁle:

(r) ∝

(9)

By the equivalence principle, this stored torsional elastic energy acts as localized gravita-

tional mass, serving as the physical origin of the dark matter halo without invoking exotic

collisionless particles.

4. Stochastic Winding, the Tully-Fisher Relation, and a

While a 1/r

density proﬁle guarantees constant orbital velocity, it is insuﬃcient to recover

the empirical baryonic Tully-Fisher relation (v

∝ M

). If the macroscopic halo mass scaled

linearly with the total number of baryons (M

halo

∝ N ∝ M

), the resulting velocity would

scale as v

∝ M

, contradicting observational data.

This normalization constraint is naturally addressed by the discrete topological kinemat-

ics of the SSM. Because baryonic braids possess chiral winding, anti-parallel or oppositely

wound braids cancel their far-ﬁeld macroscopic couple-stresses (forming localized topological

dipoles). In a macroscopic, thermalized galaxy of N independent topological defects, the net

un-canceled macroscopic torsional ﬁeld is sourced solely by the statistical ﬂuctuation of the

topological charge.

Following standard Gaussian random-walk statistics, the net un-canceled topological

charge Q

net

sourcing the long-range 1/r

halo scales as:

net

≈

√

N (10)

Since the number of baryons N is directly proportional to the total baryonic mass (N =

), the macroscopic halo mass amplitude scales exactly as the square root of the bary-

onic mass:

halo

(r) =

r (11)

where a

is the characteristic acceleration scale of the lattice’s torsional relaxation.

In the SSM, the vacuum is a ﬁnite elastic network bounded by the cosmological horizon

= c/H

[9]. The local torsional strain generated by the galaxy cannot relax to inﬁnity; it

is geometrically bounded by this global horizon. The fundamental background acceleration

of this expanding boundary is a

= c

= cH

. The geometric crossover point—where the

local un-canceled torsional ﬁeld of the galaxy interacts with the global spherical boundary

condition of the cosmic horizon—ﬁxes the structural acceleration scale a

as the spherical

projection of this universal expansion:

2π

≈ 1.1 × 10

−10

m/s

(12)

Substituting this geometric boundary limit into the halo mass yields the constant orbital

velocity:

halo

(r)

=⇒ v

= Ga

(13)

The SSM thus derives the exact baryonic Tully-Fisher relation purely from the stochastic

cancellation of discrete topological defects. Furthermore, it explicitly derives the value of

the phenomenological MOND acceleration scale (a

) from the global Hubble expansion of

the K = 12 tensor network, providing a physical geometric basis for Milgrom’s empirical

constant.

Crucially, this stochastic derivation provides a geometric mechanism for observed out-

liers to the Tully-Fisher relation. If extreme galactic dynamics—such as a recent merger or

an intense, coherent magnetic ﬁeld—temporarily align a macroscopic fraction of these chi-

ral braids, the net topological charge would shift toward Q

net

∼ N. This would result in

transiently anomalous rotation curves (v

∝ M

) until structural thermalization restores the

stochastic

√

N equilibrium.

5. The Bullet Cluster Separation

A key test for modiﬁed gravity theories is the Bullet Cluster [2]. During cluster collisions,

the baryonic hot gas physically interacts, experiencing severe electromagnetic friction and

thermal deceleration.

In the SSM, the “Dark Matter” is the macroscopic θ-ﬁeld—the torsional strain wave

propagating through the background tensor network. Vacuum torsional strain waves do not

carry electric charge and do not experience electromagnetic friction. When two clusters

collide, their respective torsional halos pass through one another in the linear elastic regime,

identical to seismic waves crossing in a solid. The massive torsional energy density continues

on its ballistic trajectory, carrying its gravitational lensing signature unimpeded. The Bullet

Cluster separation is therefore the natural thermodynamic decoupling of interacting baryonic

braids from their frictionless, elastic vacuum strain halos.

6. Conclusion

By modeling the vacuum as a discrete Chiral Cosserat tensor network, the Selection-Stitch

Model redeﬁnes Dark Matter as the macroscopic torsional strain of spacetime. Deﬁning

baryonic matter as K = 4 topological defects naturally yields an 8-bond torsional deﬁcit per

baryon. The extended 1D screw dislocations of these defects generate the ﬂat galactic rotation

curves (ρ

∝ 1/r

), a mechanism explicitly derived analytically via cylindrical superposition

and veriﬁed via numerical simulations of the K = 12 lattice. By applying stochastic random-

walk statistics to the cancellation of these chiral defects, we explicitly derive the baryonic

Tully-Fisher relation (v

∝ Ga

). Furthermore, we demonstrate that the empirical MOND

acceleration scale a

is the geometric boundary projection of the cosmic horizon (cH

/2π)

upon this local torsional ﬁeld. Concurrently, the model reproduces the collisionless decoupling

of the Bullet Cluster via linear elastic wave crossing. The SSM thus provides a geometric

framework for galactic kinematics that synthesizes the successes of both Dark Matter and

MOND paradigms strictly from the physical elasticity of the vacuum lattice.

References

[1] V. C. Rubin, W. K. Ford Jr., and N. Thonnard, “Rotational properties of 21 SC galaxies

with a large range of luminosities and radii,” Astrophys. J. 238, 471 (1980).

[2] D. Clowe, M. Bradač, A. H. Gonzalez, et al., “A Direct Empirical Proof of the Existence

of Dark Matter,” Astrophys. J. Lett. 648, L109 (2006).

[3] M. Milgrom, “A modiﬁcation of the Newtonian dynamics as a possible alternative to the

hidden mass hypothesis,” Astrophys. J. 270, 365-370 (1983).

[4] R. B. Tully and J. R. Fisher, “A new method of determining distances to galaxies,”

Astron. Astrophys. 54, 661-673 (1977).

[5] E. Cosserat and F. Cosserat, Théorie des corps déformables, Hermann, Paris (1909).

[6] A. C. Eringen, Microcontinuum Field Theories I: Foundations and Solids, Springer

(1999).

[7] D. S. Akerib et al. (LUX Collaboration), “Results from a Search for Dark Matter WIMPs

with the LUX Experiment,” Phys. Rev. Lett. 118, 021303 (2017).

[8] E. Aprile et al. (XENON Collaboration), “Dark Matter Search Results from a One Ton-

Year Exposure of XENON1T,” Phys. Rev. Lett. 121, 111302 (2018).

[9] R. Kulkarni, “Geometric Phase Transitions in a Discrete Vacuum: Deriving Cosmic Flat-

ness, Inﬂation, and Reheating from Tensor Network Topology,” Preprint (In Review),

Zenodo: 10.5281/zenodo.18727238 (2026).

[10] R. Kulkarni, “Constructive Veriﬁcation of K=12 Lattice Saturation: Exploring Kine-

matic Consistency in the Selection-Stitch Model,” Preprint (In Review), Zenodo:

10.5281/zenodo.18294925 (2026).

[11] R. Kulkarni, “Geometric Emergence of Spacetime Scales,” Preprint (In Review), Zenodo:

10.5281/zenodo.18752809 (2026).

[12] R. Kulkarni, “Deriving Newton’s Constant from the Kinematics of Lattice Defects,”

Preprint (In Review), Zenodo: 10.5281/zenodo.18856833 (2026).

[13] R. Kulkarni, “A Topological Ansatz for the Proton-to-Electron Mass Ratio,” Preprint

(In Review), Zenodo: 10.5281/zenodo.18253326 (2026).

[14] R. Kulkarni, “Matter as Frozen Phase Boundaries: Quark Structure, Fractional Charges,

and Color Conﬁnement from Tetrahedral Defects in a K = 12 Vacuum Lattice,” Preprint

(In Review), Zenodo: 10.5281/zenodo.18917946 (2026).

[15] R. Kulkarni, “Deriving the Fine Structure Constant from Bipartite Lattice Channels,”

Preprint (In Review), Zenodo: 10.5281/zenodo.18637451 (2026).

[16] R. Kulkarni, “Chiral Cosserat Elasticity and the Neutrino Mass Hierarchy,” Preprint (In

Review), Zenodo: 10.5281/zenodo.18522635 (2026).

Appendix A. Computational Veriﬁcation: Lattice Simulation

The following Python script models the exact discrete elastic mechanics of a Volterra

screw dislocation upon the K = 12 Face-Centered Cubic vacuum lattice, conﬁrming the

macroscopic 1/r

energy density proﬁle and resulting ﬂat rotation curve.

1 "" "

2 FCC Sc rew Disl oca tion : To rsi ona l Dark Matter Halo S imu lat ion

3 = = = = = = = ===== = = = = = = = ====== = = = = = = ====== = = = = = = ====== = = = = = = ======

4 Com pan ion code to :

5 " D ark Matter as the To rsi ona l Str ai n of a D is cre te Ch iral Va cu um "

6 R. Ku lka rni ( 2026)

8 This sc ript con str u cts an exa ct s cr ew dislo cat ion ( Vo lte rra solu ti on )

9 on the dis cre te K =12 Face - C enter ed Cu bic latti ce and me asu res :

10 1. The strain ene rg y den sity profi le rho (r )

11 2. The en clo sed halo mass M ( r )

12 3. The re sul tin g rotati on cu rve v(r)

13 4. Per - node strain energy d ist rib u tio n

15 The centr al re su lt : rho ( r ) ~ r ^{ -2} , con fir min g that 1 D to polo gic al

16 def ec ts ( scr ew di slo c ati ons / bar yonic braids ) g enera te ext end ed

17 tor sio nal hal os with the c orrec t radia l pro fi le for fla t rot at ion

18 curves , di re ctl y fr om d isc re te lat ti ce ela s tic ity .

20 R equ i rem ents : numpy , matplotlib , scipy

21 Us age : python fc c _ d ark_ m a t ter_ s i m ulat i o n . py

22 "" "

24 import numpy as np

25 import mat p lot lib . pyplo t as plt

26 from scipy . spatia l imp or t cKD Tree

28 # = = = = = = = = ====== = = = = = = ======= = = = = = = ====== = = = = = = ====== = = = = = = = ======

29 # 1. LAT TI CE C ONS TRUC TIO N

30 # = = = = = = = = ====== = = = = = = ======= = = = = = = ====== = = = = = = ====== = = = = = = = ======

32 def b uild _fcc _ sla b (N , a =1.0 , nz =2) :

33 """

34 Build an FCC latt ice slab : + - N c onve nti o nal cel ls in x , y

35 and nz layers in z. Trim med to a ci rc ula r cross - se ct ion .

36 """

37 basis = np . array ([

38 [0.0 , 0.0 , 0.0] ,

39 [0.5 , 0.5 , 0.0] ,

40 [0.5 , 0.0 , 0.5] ,

41 [0.0 , 0.5 , 0.5]

42 ])

43 nodes = []

44 for ix in range ( -N , N + 1) :

45 for iy in range (-N , N + 1) :

46 for iz in range ( nz ) :

47 for b in ba sis :

48 pos = ( np . array ([ ix , iy , iz ]) + b) * a

49 nodes . app en d ( pos )

50 nodes = np . array ( nodes )

52 # Trim to ci rcu lar cross - se ction for is otr opi c stat ist ics

53 r_perp = np . sqrt ( nodes [: , 0]** 2 + no de s [: , 1]** 2)

54 r_max = (N - 1) * a

55 mask = r_perp < r _m ax

56 return n od es [ mask ]

58 def f ind_ neig h bor s ( nodes , a =1.0) :

59 """

60 Find nearest - neigh bor pairs on the FCC l at tice .

61 NN di sta nce = a / sqrt (2) for c onv e nti o nal cell para met er a.

62 """

63 nn _d ist = a / np . sqrt (2) * 1.01 # small tole ran ce

64 tree = cKDT ree ( nodes )

65 pa ir_ set = tree . quer y_p airs ( nn_ dist )

66 return np . array ( list ( p air _se t ))

68 # = = = = = = = = ====== = = = = = = ======= = = = = = = ====== = = = = = = ====== = = = = = = = ======

69 # 2. SC RE W D ISL OCA T ION ( VO LT ERR A S OLU TION )

70 # = = = = = = = = ====== = = = = = = ======= = = = = = = ====== = = = = = = ====== = = = = = = = ======

72 def a p p ly_s c r ew_d i s loca t i on ( nodes , a = 1. 0) :

73 """

74 Apply the e xa ct Volter ra sc rew disl oca tion di spla cem e nt fi el d .

75 """

76 b = a / np . sqrt (2) # Burge rs vector magn itu de

77 u = np . ze ros _li ke ( nodes )

78 r_perp = np . sqrt ( nodes [: , 0]** 2 + no de s [: , 1]** 2)

79 phi = np . arc tan 2 ( nod es [: , 1] , nodes [: , 0])

81 valid = r_perp > 0.1 * a # avoid core si n gul ari t y

82 u[ valid , 2] = b * phi [ valid ] / (2 * np . pi )

83 return u

85 # = = = = = = = = ====== = = = = = = ======= = = = = = = ====== = = = = = = ====== = = = = = = = ======

86 # 3. DIS CRE TE ST RAIN EN ER GY CO MPU T ATI ON

87 # = = = = = = = = ====== = = = = = = ======= = = = = = = ====== = = = = = = ====== = = = = = = = ======

89 def c o mput e _ bond _ e nerg i e s ( nodes , u , pairs , a =1.0 , mu =1.0) :

90 """

91 Co mp ute the elas ti c s tr ai n ene rg y per bond from dis cre te

92 dis plac eme n t diff eren ces on the FCC la ttice .

93 """

94 b = a / np . sqrt (2)

95 i_idx , j _idx = pairs [: , 0] , pairs [: , 1]

97 # Bond ve ct ors and di s pla c eme nt d iffe ren ces

98 du = u[ j_idx ] - u[ i_ idx ]

100 # Fix bran ch cut

101 ma sk_ pos = du [: , 2] > b / 2

102 ma sk_ neg = du [: , 2] < - b / 2

103 du [ mask_pos , 2] -= b

104 du [ mask_neg , 2] += b

105

106 # Bond energy

107 E_bond = 0.5 * mu * np . sum ( du **2 , axis =1)

108

109 # Dis tri but e to n odes

110 E_node = np . zeros ( len ( nodes ) )

111 np . add . at ( E_node , i_idx , E_bond / 2)

112 np . add . at ( E_node , j_idx , E_bond / 2)

113

114 return E_bond , E_node

115

116 # = = = = = = = = ====== = = = = = = ====== = = = = = = = ====== = = = = = = ====== = = = = = = = ======

117 # 4. RADIAL PROFI LE MEA S URE MEN T

118 # = = = = = = = = ====== = = = = = = ====== = = = = = = = ====== = = = = = = ====== = = = = = = = ======

119

120 def m e a sur e _ radia l _pro f i l e ( nodes , E_node , pairs , a =1.0 , n_bin s =50) :

121 """

122 Bin the per - node strain energy by cyl i ndr ica l rad iu s to

123 ex tr act the e nergy d ensit y p ro file rho (r).

124 """

125 b = a / np . sqrt (2)

126 r_perp = np . sqrt ( nodes [: , 0]** 2 + no de s [: , 1]** 2)

127

128 # Count bonds per node for q ua lity f ilt er i ng

129 K_node = np . zeros ( len ( nodes ) )

130 np . add . at ( K_node , pa ir s [: , 0] , 1)

131 np . add . at ( K_node , pa ir s [: , 1] , 1)

132

133 core_r = 2 * b

134 N_eff = np . max ( r_ pe rp ) * 0.85

135 in ter ior = ( r _p er p > cor e_ r ) & ( r_ pe rp < N_eff ) & ( K _n od e >= 8)

136

137 r_bins = np . l ins pa ce ( core_r , N_eff , n_bi ns )

138 r_c = 0.5 * ( r_bins [: -1] + r_ bins [1:])

139 dr = r_bins [1] - r _bins [0]

140

141 E_bin = np . zeros ( len ( r_c ) )

142 N_bin = np . zeros ( len ( r_c ) )

143

144 for i in range ( len ( nodes )) :

145 if not int eri or [ i ]:

146 con tinue

147 idx = np . sea rch s ort e d ( r_bins , r_p er p [i ]) - 1

148 if 0 <= idx < len ( r_c ) :

149 E_bin [ idx ] += E _node [ i ]

150 N_bin [ idx ] += 1

151

152 # Cyl ind r ica l shell e ne rg y d en sity

153 she ll_ are a = 2 * np . pi * r_c * dr

154 valid = N _b in > 5

155 rho = np . z eros ( len ( r_c ) )

156 rho [ valid ] = E_bi n [ val id ] / s h ell _ar ea [ val id ]

157

158 # En clo sed mass ( c umu l ati ve sum )

159 M_enc = np . cum su m ( E_bin )

160

161 # Ro tat ion velo city : v ^2 = G M ( r) / r

162 v_rot = np . sqrt ( M_enc / ( r_c + 1e -10) )

163

164 return r_c , rho , M_enc , v_rot , valid , core_r , N_eff , r_perp , inte rio r

165

166 def f it_ p owe r _la w (r , y , valid , r_min , r _max ) :

167 mask = val id & ( r > r_m in ) & ( r < r_max ) & ( y > 0)

168 if np . sum ( mask ) < 5:

169 return None , None

170 lr = np . log ( r [ mask ])

171 ly = np . log ( y [ mask ])

172 coeffs = np . p ol yfit (lr , ly , 1)

173 return coe ff s [0] , coeffs [1]

174

175 # = = = = = = = = ====== = = = = = = ====== = = = = = = = ====== = = = = = = ====== = = = = = = = ======

176 # 5. PLO TTI NG

177 # = = = = = = = = ====== = = = = = = ====== = = = = = = = ====== = = = = = = ====== = = = = = = = ======

178

179 def c rea t e_f i gur e ( r_c , rho , M_enc , v_rot , valid , n_rho , A_rho ,

180 n_M , A_M , r_ perp_int , E_node_i nt , b , core_r , N_ ef f ):

181 plt . r cPa ram s . upda te ({

182 ’font . family ’: ’ se ri f ’ ,

183 ’font . size ’: 11 ,

184 ’axes . li new idt h ’ : 0.8 ,

185 ’ xt ick . di rec tio n ’: ’ in ’ ,

186 ’ yt ick . di rec tio n ’: ’ in ’ ,

187 })

188

189 fig , axes = plt . sub plo ts (2 , 2, fi gs ize =(12 , 10) )

190 fig . patch . set_ face colo r ( ’ white ’)

191

192 c_data = ’ #1 f77b4 ’

193 c_fit = ’# d 62 72 8 ’

194 c_ the ory = ’ #2 ca02 c ’

195 c_kep = ’ # 94 67 bd ’

196 r_th = np . l ins pac e ( r_c [ valid ][0] , r_c [ v al id ][ -1] , 200)

197

198 # (a) Ener gy densit y

199 ax = axes [0 , 0]

200 vp = valid & ( rho > 0)

201 ax . l og log ( r_c [ vp ] , rho [ vp ], ’o ’ , col or = c_data , ms =4.5 , mew =0 ,

202 alpha =0.85 , la be l = ’ FCC la tt ice ( m eas ured ) ’)

203 if n_rho is not None :

204 ax . loglog ( r_th , np . exp ( A_rho ) * r_th ** n_rho , ’ - ’ , c olor = c_fit ,

205 lw =2 , label =r ’ Best fit : $ \ rho \ propto r ^{%. 2 f}$ ’ % n_rho )

206 C_ ref = rho [ vp ][ len ( rho [ vp ]) // 3] * r_c [ vp ][ len ( r_c [ vp ]) // 3 ]* *2

207 ax . loglog ( r_th , C _ref / r_th **2 , ’ -- ’ , col or = c_theory , lw =1.5 ,

208 alpha =0.6 , label =r ’ A nal y ti c al : $\ rho \ pro pto r ^{ -2} $ ’)

209 ax . set_ x lab el (r ’ Dist anc e from d islo cat ion core $r / a$ ’)

210 ax . set_ y lab el (r ’ Strain e ne rg y d en sity $ \ rho (r ) $ ’)

211 ax . set_ tit le (r ’(a) Energy Den sity : $\ rho \ pro pt o r ^{%.2 f}$ ’ % n_rho , fon tsi ze =12)

212 ax . l eg end ( fonts ize =9 , f ram eal p ha =0.9)

213 ax . grid ( True , alp ha =0.2 , wh ich = ’ both ’)

214 ax . text (0.03 , 0.03 , r ’ T heory : $ \ rho = \ frac {\ mu b ^2}{8\ pi ^2 r ^2} $ ’ ,

215 tra nsf orm = ax . trans Axes , f ont size =10 , va = ’ bot to m ’ ,

216 bbox = dict ( box sty le = ’ round , pad =0.3 ’, f ace col or = ’ whe at ’, alpha =0 .5 ) )

217

218 # (b) En clo sed mass

219 ax = axes [0 , 1]

220 fit_M = v al id & ( M_enc > 0)

221 ax . l og log ( r_c [ f it_M ] , M_en c [ fit _M ] , ’o ’, color = c_data , ms =4.5 ,

222 mew =0 , alph a =0.85 , l ab el = r ’$M_ {\ m at hr m { halo }}( r ) $ ’)

223 if n_M is not None :

224 ax . loglog ( r_th , np . exp ( A_M ) * r_th ** n_M , ’- ’ , co lor = c_fit ,

225 lw =2 , label =r ’ Best fit : $M \ prop to r ^{%.2 f } $ ’ % n_M )

226 M_ lin = np . exp ( A_M ) * r_th [ len ( r_th ) // 2] ** n_M * ( r_th / r_ th [ len ( r_th ) //2])

227 ax . loglog ( r_th , M_lin , ’ -- ’ , color = c_theory , lw =1.5 , alpha =0.6 ,

228 label = r ’ Theo ry : $M \ propt o r$ ’)

229 ax . set_ x lab el (r ’ Radius $r / a$ ’)

230 ax . set_ y lab el (r ’ Encl ose d dark mass $M_ {\ math rm { halo }}( r) $ ’)

231 ax . set_ tit le (r ’(b) En clo se d Mass : $M \ pr op to r ^{%.2 f } $ ’ % n_M , fon tsi ze =12)

232 ax . l eg end ( fonts ize =9 , f ram eal p ha =0.9)

233 ax . grid ( True , alp ha =0.2 , wh ich = ’ both ’)

234

235 # (c) Ro tat ion c ur ve

236 ax = axes [1 , 0]

237 v_max = np . max ( v_rot [ valid ])

238 v_norm = v _r ot / v _max

239 vv = valid & ( v_n orm > 0)

240 ax . plot ( r_c [ vv ] , v _norm [ vv ], ’o - ’ , col or = c_data , ms =4 , lw =1.5 ,

241 label =r ’$v ( r ) $ from t ors ion al halo ’)

242 v_ ou ter = v_norm [ vv ][ len ( v_ no rm [ vv ]) // 3:]

243 v_fl = np . mean ( v_ outer )

244 fl atn ess = np . std ( v_ outer ) / v_fl * 100

245 ax . a xhlin e ( v_fl , c ol or = c_fit , ls = ’ -- ’, lw =1.5 , a lpha =0.7 ,

246 label = r ’ Flat : $v / v_ {\ mathrm { max }} = %.2 f$ ’ % v_fl )

247 r_kep = r_c [ vv ]

248 v_kep = v_norm [ vv ][0] * np . sqrt ( r_c [ vv ][0] / r_kep )

249 ax . plot ( r_kep , v_kep , ’ : ’ , color = c_kep , lw =2 , alpha =0.6 ,

250 label = ’ Kepl eri an ( no dark halo ) ’)

251 ax . set_ x lab el (r ’ Radius $r / a$ ’)

252 ax . set_ y lab el (r ’ $v ( r ) / v_ {\ mathrm { max }} $ ’)

253 ax . set_ tit le (r ’(c) Ro tat io n Curve ( $ \ si gma /\ mu = %.1 f \%% $) ’ % flatness , f ont siz e =12 )

254 ax . set_ yl im (0 , 1.4)

255 ax . l eg end ( fonts ize =9 , f ram eal p ha =0.9)

256 ax . grid ( True , alp ha =0.2)

257

258 # (d) Per - node s catte r

259 ax = axes [1 , 1]

260 sca tter _ma s k = (( r _pe rp_ int > 2 * b ) & ( r_pe rp_ int < N_eff ) & ( E _ nod e_i nt > 0) )

261 ax . s catte r ( r_p erp _in t [ s cat t er_ m ask ] , E_ nod e_i n t [ sca t ter _mas k ] ,

262 s =0.5 , alpha =0.08 , color = c_data , ras ter i zed = True )

263 r_sort = np . l ins pa ce (2 * b, N_eff , 300)

264 me d_ val = np . media n ( E_ n ode _in t [ scat ter _ mas k ] * r_ per p_i nt [ sca t ter _ma s k ]**2 )

265 ax . plot ( r_sort , me d_ val / r _sort **2 , ’ - ’ , color = c_fit , lw =2 ,

266 label =r ’$ \ v are psi lon \ propto 1/ r ^2 $ ’ )

267 ax . set_ x lab el (r ’ Dist anc e from core $r / a$ ’)

268 ax . set_ y lab el (r ’ Energy per no de $ \ v are p sil o n_i $ ’)

269 ax . set_ tit le ( ’(d) Per - Node Strai n Energy ’ , fo nts ize =1 2)

270 ax . set_ y sca le ( ’ log ’)

271 ax . set_ x sca le ( ’ log ’)

272 ax . l eg end ( fonts ize =9 , f ram eal p ha =0.9)

273 ax . grid ( True , alp ha =0.2 , wh ich = ’ both ’)

274

275 plt . ti ght _lay out ( rect =[0 , 0.02 , 1, 0.94])

276 fig . s upt itl e (

277 ’ Sc rew Disl oca tion on FCC ( $K =12 $ ) Latti ce : ’

278 ’ Tor sio nal Dark M atter Halo \ n ’

279 r’ $ \ rho_ \ theta \ propto 1/ r ^2 \;\ R i ght arr ow \; ’

280 r’ M_ {\ m athrm { halo }} \ p ro pt o r \;\ R igh tar row \; ’

281 r’ v (r) = \ ma th rm { cons t }$ ’ ,

282 fon tsi ze =13 , fo ntw eig ht = ’ bold ’, y = 0. 98 )

283

284 return fig , f lat ness

285

286 # = = = = = = = = ====== = = = = = = ====== = = = = = = = ====== = = = = = = ====== = = = = = = = ======

287 # 6. MAIN

288 # = = = = = = = = ====== = = = = = = ====== = = = = = = = ====== = = = = = = ====== = = = = = = = ======

289

290 def main () :

291 a = 1.0

292 N = 15

293 mu = 1.0

294 b = a / np . sqrt (2)

295

296 nodes = bui l d_f c c_s l ab (N , a, nz =2)

297 pairs = fin d _ne i ghb o rs ( nodes , a)

298 u = a pply _ s crew_ d i sloc a t ion ( nodes , a)

299 E_bond , E_nod e = c o m put e _ bond _ e nerg i e s ( nodes , u , pairs , a , mu )

300

301 ( r_c , rho , M_enc , v_rot , valid ,

302 core_r , N_eff , r_perp , inte rio r ) = m e a sure _ r adia l _ prof i l e ( nodes , E_node , pairs , a)

303

304 fi t_ min = 3 * b

305 fi t_ max = N _eff - 2 * a

306 n_rho , A _rho = fi t _po w er_ l aw ( r_c , rho , valid , fit_min , f it_ma x )

307 n_M , A_M = fit _pow er_l aw ( r_c , M_enc , v alid & ( M_enc > 0) , fit_min , fit_ max )

308

309 r_p erp _in t = r _p erp [ inter ior ]

310 E_n ode _in t = E _n ode [ inter ior ]

311

312 fig , fl at nes s = cr eate _fig ure (

313 r_c , rho , M_enc , v_rot , valid ,

314 n_rho , A_rho , n_M , A_M ,

315 r_ perp_int , E_node_i nt ,

316 b , core_r , N_e ff )

317

318 ou tp ath = ’ fc c _ d ark_ m a t ter_ s i m ulat i o n . jpg ’

319 fig . s avefi g ( outpath , dpi =300 , bbo x_i nche s = ’ tight ’ , fa cec olo r = ’ white ’)

320

321 if __n ame __ == " __ mai n_ _ " :

322 main ()