In the present paper we study the nonlinear system ut + [ϕ(u)]x + v = 0, vt + ψ(u)vx = 0 as a model for the one-dimensional dynamics of dark matter. We prove that under certain conditions this system, such as the Gurevich-Zybin system, can also explain why the observed rotation speed (relative to the galactic center) of stars near galactic halos do not coincide with what it is expected in classical mechanics. The solutions are obtained in fully explicit formulas, in a convenient space of distributions, without using any result within the classical framework. For such purpose we use the α-solution concept which is defined within a product of distributions. Such a concept generalizes the classical solution concept and for evolution equations may also be seen as an extension of the weak solution concept to the nonlinear setting.

In the present paper we are concerned with the system
ut+[ϕ(u)]x+v=0,
(1)
vt+ψ(u)vx=0,
(2)
where xR is the space variable, t ∈ [0, +∞[ is the time variable, u = u(x, t) and v = v(x, t) represent the state variables and ϕ,ψ:CC are entire functions that take real values on the real axis. The variable u will be considered as an hydrodynamic speed and the variable v will be seen in such a way that vx(x, t) = ρ(x, t), being ρ a density of matter.
This way, (1) and (2) represents a family of systems indexed in ϕ, ψ that contains the nonlinear Gurevich-Zybin system
ut+u22x+v=0,
(3)
vt+uvx=0,
(4)
that rules the dynamic of dark matter. This system was presented for the first time in 2005 by Pavlov1 and was derived from a hydrodynamic system in a book by Jeans2 whose first edition goes back to 1928. See also, for example, Refs. 3–6, to account for several aspects under which this system was studied. A large number of papers and books have been published regarding the dark matter and the problem of galactic rotation. As an introduction we refer Refs. 7 and 8.
The goal of this work is to show that there exist a set of models that, such as the Gurevich-Zybin system, may explain the problem of galactic rotation that we will introduce in Sec. VII. For such purpose we subject the variables u, v of the system (1) and (2) to the conditions
u(x,0)=u1+(u2u1)H(x),
(5)
vx(x,0)=ρ(x,0)=mδ(x),
(6)
where u1,u2,mR, m > 0 and H, δ stand respectively for the Heaviside function and the Dirac delta measure supported at the origin. The solution (u, v) of problem (1), (2), (5), and (6) will be considered in a convenient space W of pairs of functions (u, v) defined by
u(x,t)=a(t)+b(t)H[xγ(t)],
v(x,t)=g(t)H[xγ(t)],
where a,b,g,γ:[0,+[R are C1-functions. In a sense that will be explained we will prove that, if ψ′(u1) ≠ 0, problem (1), (2), (5), and (6) has an unique solution in W, given by
u(x,t)=u1+(mt+u2u1)H[xγ(t)],
v(x,t)=mH[xγ(t)],
where
γ(t)=ψ(u1)t+ϕ(u1)2ψ(u1)0t[ψ(mτ+u2)ψ(u1)]dτ.
(7)
Therefore, the density of matter
ρ(x,t)=vx(x,t)=mδ[xγ(t)]
shows a particle of mass m > 0, that moves according to the law (7) in a one-dimensional universe. From a physical viewpoint this will be interpreted in Sec. VIII in the setting of the problem of galactic rotation.

Usually, the solutions of problems like this one are given by weak limits of sequences of continuous functions. Frequently, such limits cannot be substituted into the equations mainly owing to the difficulties of multiplying distributions and, sometimes, these processes do not yield to mathematically consistent solutions (see Ref. 9, Sec. II). These difficulties can be overcome, as we will explain.

In our framework, the product of two distributions is a distribution that depends on the choice of a function α encoding the indeterminacy inherent to such products. This indeterminacy generally is not avoidable and in many cases it has a physical meaning; concerning this point see Refs. 1013. As a consequence, the solutions of differential equations with such products may depend, or not, of α. We call such solutions α-solutions and, in each case, they must be defined in such a way that any classical solution is an α-solution for any α. Thus, the α-solutions can be seen as generalized classical solutions and, for evolution equations, they can also be considered as extensions of the concept of weak solution in the nonlinear setting (see Ref. 14, p. 868).

When the solutions depend on α the future behavior of the model cannot be fully predicted. This fact might be due to physical features omitted in the formulation of the model with the goal of simplifying it. See Refs. 15–17 concerning this point. In spite of the possible α-dependence, it is worthwhile to stress that for many problems (like our one) the α-solutions do not depend on α. Certainly this is one of the reasons for which several authors begun to adopt this method (see Refs. 1830). Another reason regards to important simplifications: the Rankine-Hugoniot shock conditions and their multiple generalizations are not necessary and, as much as we know, the final result is the same (see Refs. 21, 31, and 32). Thus, even in the cases where the classical results can be applied, it is certainly more easy to use α-solutions if the operations with distributions involved are perfectly defined.

Interestingly, in a recent paper,23 Pang et al. compared the exact α-solutions of a totally degenerate system of conservation laws, with the numerical solutions obtained by using the Nessyahu-Tadmor scheme: the α-solutions coincide with the corresponding numerical solutions. For more problems where the concept of α-solution was also applied the reader may see Refs. 33–36 and other works of the author.

The present section concerns some formulas from our theory of distributional products that are of interest in the sequel. To get a general view of our distributional products see Ref. 37, Secs. 2 and 3 and also Ref. 38, Secs. 2–4. Details are given in Ref. 39.

Let D be the space of indefinitely differentiable complex-valued functions defined on R, with compact support, and let D be the space of Schwartz distributions. In our theory, each function αD with +α=1 affords a general product Tα̇SD of T,SD.

Our general product is bilinear and is transformed as usual by translations, that is,
τa(Tα̇S)=(τaT)α̇(τaS),
(8)
where τa denotes the usual translation operator in distributional sense. In general, associativity and commutativity do not hold. Recall that in the setting of the classical products of distributions, the commutative property is a convention inherent to the definition of such products and the associative property does not hold in general (see the monograph of Schwartz40, pp. 117, 118, 121, where these products are defined). The usual differential rules are satisfied, including the Leibniz formula, which must be written in the form
D(Tα̇S)=(DT)α̇S+Tα̇(DS),
(9)
where D is the derivative operator in distributional sense. This general product is not consistent with the Schwartz products of distributions with functions but it is possible to define certain α-products in order to recover that consistency. This happens with the α-product
Tα̇S=Tβ+(Tα)f
(10)
for TDp and S=β+fCpDμ, where p ∈ {0, 1, 2, …, ∞}, Dp is the space of distributions of order p in the sense of Schwartz (D means D), Dμ is the space of distributions whose support has Lebesgue measure zero, is the usual Schwartz product of a Dp-distribution with a Cp-function and (Tα)f is the usual product of a C-function with a distribution. For instance, if β is a continuous function, we have for any α,
δα̇β=δα̇(β+0)=δβ+(δα)0=β(0)δ,
βα̇δ=βα̇(0+δ)=β0+(βα)δ=[(βα)(0)]δ,
δα̇δ=δα̇(0+δ)=δ0+(δα)δ=αδ=α(0)δ,
δα̇(Dδ)=(δα)Dδ=αDδ=α(0)Dδα(0)δ,
(Dδ)α̇δ=(Dδα)δ=αδ=α(0)δ,
Hα̇δ=(Hα)δ=+α(τ)H(τ)dτδ=0αδ,
(11)
Hα̇(Dδ)=(Hα)(Dδ)=0α(Dδ)α(0)δ.
The α-product (10) is consistent with all Schwartz products of Dp-distributions with Cp-functions if the Cp-functions are placed at the right-hand side. It also keeps the bilinearity and satisfies (8) and (9), this one clearly under certain natural conditions; for TDp we must suppose SCp+1Dμ.
From Leibniz formula (9) it is possible to define new α-products. The following formula was constructed this way (for details see Ref. 41, Sec. 2),
Tα̇S=Tw+(Tα)f,
(12)
for TD1 and S=w+fLloc1Dμ, where D1 denotes the space of distributions TD such that DTD0, and Tw is the usual pointwise product of TD1 with wLloc1. Thus, locally, T can be read as a function of bounded variation and D1 as the space of locally bounded variation functions. For instance, since HD1 and H=H+0Lloc1Dμ, we have
Hα̇H=HH+(Hα)0=H,
(13)
for any α. More generally, if TD1 and SLloc1 then Tα̇S=TS, for any α, because by (12) we can write
Tα̇S=Tα̇(S+0)=TS+(Tα)0=TS.
We also use another α-product that is computed by the formula
Tα̇S=D(Yα̇S)Yα̇(DS),
(14)
for TD0Dμ and S,DSLloc1Dc, where DcDμ is the space of distributions whose support is at most countable, and YD1 is such that DY = T. The products Yα̇S and Yα̇(DS) are supposed to be computed by (10) or (12). The value of Tα̇S given by (14) is independent of the choice of Y such that DY = T (see Ref. 41, p. 1004 and Ref. 42, Sec. 2, for details). For instance, using (11), (13), and (14), we have for any α,
δα̇H=D(Hα̇H)Hα̇(DH)=DHHα̇δ=δ0αδ=0+αδ,
(15)
so that
Hα̇δ+δα̇H=δ,
for any α.

The compatibility of (10), (12) and (14) is effective, that is, if the same α-product can be computed by two different formulas, they give the same result.

Also in general, supp(Tα̇S)suppS as it happens for the usual product of functions, but it may happen that supp(Tα̇S)suppT.

Let I be an interval of R with more that one point, and let F(I) be the space of continuously differentiable maps ũ:ID in the sense of the usual topology of D. For tI, the notation [ũ(t)](x) is sometimes used to emphasize that the distribution ũ(t) acts on functions ξD depending on x.

Let Σ(I) be the space of functions u:R×IR such that:

  • for each tI, u(x,t)Lloc1(R);

  • ũ:ID, defined by [ũ(t)](x)=u(x,t) belongs to F(I).

The natural injection uũ from Σ(I) into F(I) identifies any function in Σ(I) with a certain map in F(I). Since C1(R×I)Σ(I), we can write the inclusions
C1(R×I)Σ(I)F(I).
Thus, identifying u with ũ and v with ṽ, the system (1) and (2) can be read as follows:
d[ũ(t)]dt+D[ϕũ(t)]+ṽ(t)=0,
(16)
d[ṽ(t)]dt+[ψũ(t)]α̇[Dṽ(t)]=0.
(17)

Definition 1.

Given α, the pair (ũ,ṽ)F(I)×F(I) will be called an α-solution of the system (16) and (17) on I, if ϕũ(t), ψũ(t) and the α-product [ψũ(t)]α̇[Dṽ(t)] are well defined distributions and this system is satisfied for all tI.

We have the following results:

Theorem 2.

If (u, v) is a classical solution of (1) and (2) on R×I then, for any α, the pair (ũ,ṽ)F(I)×F(I) defined by [ũ(t)](x)=u(x,t), [ṽ(t)](x)=v(x,t) is an α-solution of (16) and (17) on I.

Note that, by a classical solution of (1) and (2) on R×I, we mean a pair of C1-functions (u, v) that satisfies (1) and (2) on R×I.

Theorem 3.

If u,v:R×IR are C1-functions and, for a certain α, the pair (ũ,ṽ)F(I)×F(I) defined by [ũ(t)](x)=u(x,t), and [ṽ(t)](x)=v(x,t) is an α-solution of (16) and (17) on I, then (u, v) is a classical solution of (1) and (2) on R×I.

For the proofs, it is enough to observe that any C1-function u can be read as a continuously differentiable function ũF(I) defined by [ũ(t)](x)=u(x,t) and to use the consistency of the α-products with the classical Schwartz products of distributions with functions.

Replacing [ψũ(t)]α̇[Dṽ(t)] by [Dṽ(t)]α̇[ψũ(t)] in Ref. 3, we get the equation
d[ṽ(t)]dt+[Dṽ(t)]α̇[ψũ(t)]=0,
(18)
which is not equivalent to Ref. 3 because our α-products are not, in general, commutative. However, all we said for the systems (1), (2), (16), (17) is also valid for the systems (1), (2), (16), and (18) including Definition 1. Thus, and taking advantage of this situation we introduce the following definition that further extends the concept of α-solution.

Definition 4.

Given α, any α-solution (ũ,ṽ) of the system (16) and (17) or of the system (16) and (18) on I will be called an α-solution of (1) and (2) on R×I.

As a consequence an α-solution (ũ,ṽ)F(I)×F(I) in this sense read, for each t, as a pair of distributions (u, v), affords a consistent extension of the concept of a classical solution for the system (1) and (2). Thus, and for short, we also call the pair (u, v) an α-solution of the system (1) and (2) on R×I.

Let us consider the problem (1), (2), (5), and (6) with (x,t)R×I, I = [0, +∞[ and (u, v) ∈ W. Remember that we assume ϕ, ψ, entire functions that take real values on the real axis and u1,u2,mR with m > 0. When we read this problem in F(I), the identification uũ allows us to replace the system (1) and (2) by the system (16) and (17) and conditions (5) and (6) respectively by the following ones
ũ(0)=u1+(u2u1)H,
(19)
{D[ṽ(t)]}t=0=mδ.
(20)
The following result gives us all α-solutions of the problem (16), (17), (19), and (20) in the space W̃F(I)×F(I) defined as follows: (ũ,ṽ)W̃, if and only if,
ũ(t)=a(t)+b(t)τγ(t)H,
(21)
ṽ(t)=g(t)τγ(t)H,
(22)
for certain C1-functions a,b,g,γ:IR and all tI. Clearly, W̃ is the set corresponding to W which was defined in the introduction.

Theorem 5.

Given α, the problem (16), (17), (19), and (20) with ψ′(u1) ≠ 0 has an α-solution (ũ,ṽ)W̃ if and only if the following two conditions are satisfied:

  • 0α=ϕ(u1)2ψ(u1),

  • ϕ(z)ϕ(u1)=(zu1)ψ(u1)+ϕ(u1)2ψ(u1)[ψ(z)ψ(u1)], for all z.

In this case, the α-solution is given by
ũ(t)=u1+(mt+u2u1)τγ(t)H,
(23)
ṽ(t)=mτγ(t)H,
(24)
where γ is given by
γ(t)=ψ(u1)t+ϕ(u1)2ψ(u1)0t[ψ(mτ+u2)ψ(u1)]dτ.
(25)
This α-solution is unique in W̃, does not depend on α and clearly
ρ̃(t)=Dṽ(t)=mτγ(t)δ.

Proof.
Suppose (ũ,ṽ)W̃. Then, from (19) and (21) we have
a(0)+b(0)τγ(0)H=u1+(u2u1)H.
(26)
From (22) we have,
D[ṽ(t)]=g(t)τγ(t)δ,
(27)
and from (20), it follows g(0)τγ(0)δ = , which means that γ(0) = 0 and g(0) = m. Then, from (26) we get a(0) + b(0)H = u1 + (u2u1)H and a(0) = u1 and b(0) = u2u1 follows.
From (21) and (22) we get respectively
dũdt(t)=a(t)+b(t)τγ(t)H+b(t)τγ(t)δ[γ(t)],
dṽdt(t)=g(t)τγ(t)H+g(t)τγ(t)δ[γ(t)],
and, since for each t, ũ(t) is the function defined by
[ũ(t)](x)=a(t)  if  x<γ(t)a(t)+b(t) if x>γ(t),
we have
[ϕũ(t)](x)=ϕ{[ũ(t)](x)}=ϕ[a(t)]  if  x<γ(t)ϕ[a(t)+b(t)] if x>γ(t).
Thus, we can write
ϕũ(t)=ϕ[a(t)]+{ϕ[a(t)+b(t)]ϕ[a(t)]}τγ(t)H,
and also
ψũ(t)=ψ[a(t)]+{ψ[a(t)+b(t)]ψ[a(t)]}τγ(t)H.
Then,
D[ϕũ(t)]={ϕ[a(t)+b(t)]ϕ[a(t)]}τγ(t)δ,
and using (11) and (27) we get
[ψũ(t)]α̇D[ṽ(t)]={ψ[a(t)]+{ψ[a(t)+b(t)]ψ[a(t)]}τγ(t)H}α̇[g(t)τγ(t)δ]
=ψ[a(t)]g(t)τγ(t)δ+{ψ[a(t)+b(t)]ψ[a(t)]}g(t)0ατγ(t)δ.
(28)
Hence, from (16) we get, for all t,
a(t)+[b(t)+g(t)]τγ(t)H+{b(t)γ(t)+ϕ[a(t)+b(t)]ϕ[a(t)]}τγ(t)δ=0,
and, using (28), from (17) we also get,
g(t)τγ(t)H+
g(t)γ(t)+ψ[a(t)]g(t)+{ψ[a(t)+b(t)]ψ[a(t)]}g(t)0ατγ(t)δ=0.
As a consequence, (ũ,ṽ) defined by (21) and (22) is an α-solution of (16) and (17) if and only if, for all t, the following five conditions hold:
  1. a′(t) = 0,

  2. b′(t) + g(t) = 0,

  3. b(t)γ′(t) + ϕ[a(t) + b(t)] − ϕ[a(t)] = 0,

  4. g′(t) = 0,

  5. g(t)γ(t)+ψ[a(t)]g(t)+{ψ[a(t)+b(t)]ψ[a(t)]}g(t)0α=0.

Therefore, from (1′), (4′) and (2′) we conclude that a(t) = u1, g(t) = m and b(t) = −mt + u2u1. From (3′) and (5′) we get respectively
(mt+u2u1)γ(t)+ϕ(mt+u2)ϕ(u1)=0,
(29)
γ(t)=ψ(u1)+[ψ(mt+u2)ψ(u1)]0α,
(30)
and (29) turns out to be
(mt+u2u1)ψ(u1)+[ψ(mt+u2)ψ(u1)]0α
+ϕ(mt+u2)ϕ(u1)=0.
Now, let η:CC be defined by
η(z)=(zu1)ψ(u1)+[ψ(z)ψ(u1)]0α+ϕ(z)ϕ(u1).
Then, η is an entire function and η(−mt + u2) = 0 for all t. Since the zeros of an entire function η ≠ 0 are always isolated points we conclude that η = 0, that is,
ϕ(z)ϕ(u1)=(zu1)ψ(u1)+[ψ(z)ψ(u1)]0α
for all z, and taking the second derivative of both sides we get
ϕ(z)=2ψ(z)0α+(zu1)ψ(z)0α.
Replacing z by u1 we conclude that
0α=ϕ(u1)2ψ(u1),
and (a) follows. From (30) we get
γ(t)=ψ(u1)+[ψ(mt+u2)ψ(u1)]ϕ(u1)2ψ(u1),
and (25) follows immediately. Also from (21) and (22) we get (23) and (24) and it is now easy to see that conditions (1′), (2′), (3′), (4′) and (5′) are all satisfied for all t. Clearly, this α-solution is unique in W̃ and does not depend on α. The value of ρ̃(t) follows from (22). The statement is proved.■

Taking in Theorem 5 the first derivative of both sides of (b) and replacing z by u1, we get ϕ′(u1) = ψ(u1) and a simple and practical necessary condition for the existence of an α-solution follows:

Corollary 6.

For the problem (16), (17), (19), and (20) with ψ′(u1) ≠ 0 to have an α-solution it is necessary that ϕ′(u1) = ψ(u1).

Since our α-products are not, in general, commutative we may analyze what happens to Theorem 5 if in (17) we replace [ψũ(t)]α̇[Dṽ(t)] by [Dṽ(t)]α̇[ψũ(t)]. Thus, and using (15), we must write instead of (28),
[Dṽ(t)]α̇[ψũ(t)]
=ψ[a(t)]g(t)τγ(t)δ+{ψ[a(t)+b(t)]ψ[a(t)]}g(t)0+ατγ(t)δ.
Therefore, for the problem (16), (18), (19), and (20), Theorem 5 must be replaced by another Theorem where the only difference is that 0α must be replaced by 0+α. As a consequence of Definition (4), the following result follows:

Theorem 7.

Given α, the problem (1), (2), (5), and (6), with ψ′(u1) ≠ 0, has an α-solution (u, v) ∈ W if and only if the following two conditions are satisfied:

  • 0α=ϕ(u1)2ψ(u1) or 0+α=ϕ(u1)2ψ(u1),

  • ϕ(z)ϕ(u1)=(zu1)ψ(u1)+ϕ(u1)2ψ(u1)[ψ(z)ψ(u1)], for all z.

In this case, the α-solution is given by
u(x,t)=u1+(mt+u2u1)H[xγ(t)],
v(x,t)=mH[xγ(t)],
where γ is given by
γ(t)=ψ(u1)t+ϕ(u1)2ψ(u1)0t[ψ(mτ+u2)ψ(u1)]dτ.
(31)
This α-solution is unique in W, does not depend on α and clearly
ρ(x,t)=mδ[xγ(t)].

Therefore, as we said in the introduction, the density of matter ρ(x, t) shows a particle of mass m > 0 that moves according to the law (31).

Remark 8.

Corollary 6 remains valid for the problem (1), (2), (5), and (6).

Example 9.

Let us consider the Gurevich-Zybin system (3) and (4) with the unknowns (u, v) subjected to conditions (5) and (6). Clearly, we have ϕ(z)=z22 and ψ(z) = z. For this particular case, we have ψ′(u1) = 1 ≠ 0 and since ϕ(u1)2ψ(u1)=12, (a) turns out to be simply 0α=12, because 0+α=10α=12. Also it is easy to see that (b) is satisfied for any z, and γ(t)=u1+u22t14mt2. As a consequence, we conclude:

Theorem 10.
Given α, the Gurevich-Zybin system (3) and (4) with the unknowns subjected to conditions (5) and (6) has an α-solution (u, v) ∈ W if and only if 0α=12. In this case, the α-solution is unique, it is independent of α and it is given by
u(x,t)=u1+(mt+u2u1)Hxu1+u22t+14mt2,
(32)
v(x,t)=mHxu1+u22t+14mt2.
(33)

Therefore, for the Gurevich-Zybin system we get, for the density of matter
ρ(x,t)=vx(x,t)=mδxu1+u22t+14mt2.
(34)

In Ref. 36, Theorem 6, p. 7, the expressions for u and v are not exactly the same. Unfortunately, in this paper, the author committed a mistake; in the Proof of Theorem 5, p. 5, line −3, instead of ũ(t)α̇ũ(t)=a2(t)+[b2(t)a2(t)]τγ(t)H, it must be ũ(t)α̇ũ(t)=a2(t)+[2a(t)b(t)+b2(t)]τγ(t)H. Thus, after the correction of some expressions, we get the corrected Theorem 5 of Ref. 36 where u and v are given exactly by (32) and (33) and the density of matter by (34). Fortunately, this mistake does not modify the essence of the physical interpretation we have done in Ref. 36, as we will see in Sec. VI, (I).

The speed of rotation of stars or gas within galaxies should decrease as a function of the distance of that matter to the center of such galaxies, as it is expected in classical mechanics. Instead, from astronomical observations regarding the “curves of galactic rotation” such speed of rotation remains approximately constant when going out to large distances from the galactic center. In 1937, Fritz Zwicky,8 applying the Virial Theorem of classical mechanics, was the first to discover that galaxies may contain much more mass then that which can be seen. Thus, to explain data, it was postulated that enormous halos exist involving galaxies made of unknown “dark matter” that does not interact with the usual baryonic matter or with the electromagnetic field (we cannot see the dark matter), but interact only with the gravitational field.

Astronomers were skeptical about dark matter until the appearing of the enormous work developed by Vera Rubin, that forced a large part of the astronomical community to take it seriously. For the interested reader we refer as an introduction.7,43 In what follows we will interpret our results.

  • First we will give an explanation of our results for ϕ(u)=u22 and ψ(u) = u in (1) and (2), that is, for the Gurevich-Zibin system (3) and (4).

We begin to chose a star located at a point P near the halo of a galaxy so that the hydrodynamic speed at P in the instant t = 0 be u(x, 0) = 0. This means that u1 = u2 = 0 in (5). For the x-axis of our model we will chose a straight line belonging to the galactic plane, passing by P and being ortogonal to the position vector of P relative to the galactic center. This way, P will be the origin of our x-axis. Then, from (32) and (33) the hydrodynamic speed u and the density of mass ρ = vx are given, by
u(x,t)=mtHx+14mt2,
ρ(x,t)=mδx+14mt2,
and the expression of ρ can be interpreted as the motion of a “dark particle” of mass m > 0 that moves on the x-axis according to the law γ(t)=14mt2, that is, with constant acceleration γ(t)=m2. This acceleration may be seen, as the result of a constant “dark force” F, of gravitational nature, that acts on the baryonic matter in negative sense, that is, in the same sense of the hydrodynamic speed u(0, t), for each t > 0. As a consequence, the gravitational pull F tends to increase (in absolute value) the speed of the matter at P, that is, near the galactic halo, as observed in the curves of galactic rotation!
  • For our general model (1), (2), (5), and (6) we will try to follow closely what we did for the Gurevich-Zybin model.

From Theorem 7 our general problem (1), (2), (5), and (6) has an α-solution (u, v) ∈ W, with u1 = u2 = 0, if and only if the following two conditions are satisfied:

  • 0α=ϕ(0)2ψ(0) or 0+α=ϕ(0)2ψ(0);

  • ϕ(z)ϕ(0)=zψ(0)+ϕ(0)2ψ(0)[ψ(z)ψ(0)] for all z.

The hydrodynamic speed u and the density ρ = vx in this case are given by
u(x,t)=mtH[xγ(t)],
ρ(x,t)=mδ[xγ(t)],
where γ(t) is given by
γ(t)=ψ(0)t+ϕ(0)2ψ(0)0t[ψ(mτ)ψ(0)]dτ.
(35)
Then,
γ(t)=mϕ(0)2ψ(0)ψ(mt),
and the gravitational pull F(t) spanned by this acceleration acts to increase (in absolute value) the speed of the matter at P if there exists an interval J ⊂ ]0, +∞[, with more than one point, such that for all tJ, F(t) has the same sense of the hydrodynamic speed u(0, t), which means that u(0, t)γ″(t) > 0 for all tJ, that is, if
ϕ(0)ψ(0)ψ(mt)H[γ(t)]>0,
(36)
for all tJ, where γ(t) is given by (35). Summing up, we have the following result:

Theorem 11.

The problem (1), (2), (5), and (6) with u1 = u2 = 0 and ψ′(0) ≠ 0 has an α-solution (u, v) ∈ W if and only if conditions (a′) and (b′) are satisfied. In such case, if there exists an interval of time J, with more than one point, such that for all tJ (36) is satisfied, the gravitational pull F(t) spanned by the acceleration γ″(t), has, the same sense of the hydrodynamic speed u(0, t), and thedark forceF(t) acts to increase the speed of the matter at P during the interval of time J.

Example 12.
Let us consider the problem
ut+(u4u2)x+v=0,
(37)
vt+(uu3)vx=0,
(38)
with the unknowns u, v subjected to the conditions
u(x,0)=0,
(39)
vx(x,0)=ρ(x,0)=mδ(x),
(40)
with m > 0. Regarding the general problem, we have ϕ(z) = z4z2, ψ(z) = zz3, u1 = u2 = 0, and ψ′(0) = 1 ≠ 0. Hence, this problem has an α-solution (u, v) ∈ W if and only if conditions (a′) and (b′) of Theorem 11 are satisfied. Since (b′) is trivially satisfied, we conclude that for all α satisfying (a′), that is, for all α such that
0α=1  or  0+α=1,
the problem (37)(40) has an α-solution (u, v) ∈ W given by
u(x,t)=mtH[xγ(t)],
v(x,t)=mH[xγ(t)],
where γ(t) is given by (35), that is,
γ(t)=0t(mτ+m3τ3)dτ=mt221m2t22.
(41)
Hence, assuming that the system (37) and (38) rules the dynamic of dark matter (which from a theoretical viewpoint can be true only approximately), the density of matter ρ(x, t) = vx(x, t) = [xγ(t)] shows a star of mass m that moves according to the law (41). Thus, (36) turns out to be
(3m2t21)Hmt22m2t221>0,
and in the interval of time J=]2m,+[, or any subinterval with more than one point, (36) is satisfied. Therefore, from the instant t=2m on, the gravitational pull F(t) = ″(t) acts in order to increase the speed of the star.

Clearly, for the Gurevich-Zybin system (3) and (4), condition (36) turns out to be the trivial condition 1 > 0 with
γ(t)=14mt2.
during all the interval of time J = ]0, +∞[ and the dark force F(t) spanned by the acceleration γ(t)=m2 is constant. Therefore, the possibility that the curves of galactic rotation do not adjust to this model during all this interval of time is real. Possibly, in the general model (1) and (2), a convenient choice of the functions ϕ, ψ can adjust this model to the curves of galactic rotation and better predict future events.

The author is grateful to the referee for the suggestions which clearly improved the text.

The present research was supported by FCT, UID/MAT/04561/2019.

The author has no conflicts to disclose.

C. O. R. Sarrico: Conceptualization (equal); Investigation (equal).

Data sharing is not applicable to this article as no new data were created or analyzed in this study.

1.
M. V.
Pavlov
, “
The Gurevich-Zybin system
,”
J. Phys. A: Math. Gen.
38
,
3823
3840
(
2005
).
2.
J. H.
Jeans
,
Astronomy and Cosmology
, 2nd ed. (
Cambridge University Press
,
1929/1969
).
3.
M. V.
Pavlov
and
A. K.
Prykarpatski
, “
A generalized hydrodynamical Gurevich-Zybin equation of Riemann type and its Lax type integrability
,”
Condens. Matter Phys.
13
(
4
),
43002
(
2010
).
4.
Y. A.
Prykarpatski
,
D.
Blackmore
,
J.
Golenia
, and
A. K.
Prykarpatski
, “
A vertex operator representation of solutions to the Gurevich-Zybin hidrodynamical equation
,”
Opusc. Math.
33
(
1
),
139
149
(
2013
).
5.
R. C.
Thompson
, “
The Cauchy problem for the 1-D Gurevich-Zybin system
,”
J. Math. Phys.
60
,
051509
(
2019
).
6.
L.
Wei
, “
Wave breaking, global existence and persistent decay for the Gurevich-Zybin system
,”
J. Math. Fluid Mech.
22
,
47
(
2020
).
7.
V. C.
Rubin
, “
Dark matter in spiral galaxies
,”
Sci. Am.
248
(
6
),
96
108
(
1983
).
8.
F.
Zwicky
, “
On the masses of nebulae and of clusters of nebulae
,”
Astrophys. J.
86
,
217
246
(
1937
).
9.
C. O. R.
Sarrico
and
A.
Paiva
, “
Newton’s second law and the multiplication of distributions
,”
J. Math. Phys.
59
,
013505
(
2018
).
10.
A.
Bressan
and
F.
Rampazzo
, “
On differential systems with vector valued impulsive controls
,”
Bull. Un. Mat. Ital. B
7
(
2
),
641
656
(
1988
).
11.
J. F.
Colombeau
and
A.
Le Roux
, “
Multiplications of distributions in elasticity and hydrodynamics
,”
J. Math. Phys.
29
,
315
319
(
1988
).
12.
G.
Dal Maso
,
P.
LeFloch
, and
F.
Murat
, “
Definitions and weak stability of nonconservative products
,”
J. Math. Pure Appl.
74
,
483
548
(
1995
).
13.
C. O. R.
Sarrico
, “
Distributional products and global solutions for nonconservative inviscid Burgers equation
,”
J. Math. Anal. Appl.
281
,
641
656
(
2003
).
14.
A.
Paiva
and
C.
Sarrico
, “
New distributional travelling waves for the nonlinear Klein-Gordon equation
,”
Differ. Integral Equ.
30
(
11/12
),
853
878
(
2017
).
15.
S. T.
Hilden
,
H. M.
Nilsen
, and
X.
Raynaud
, “
Study of the well-posedness of models for the inaccessible pore volume in polymer flooding
,”
Transp. Porous Media
114
,
65
86
(
2016
).
16.
H.
Kalisch
and
D.
Mitrovic
, “
Singular solutions of a fully nonlinear 2 × 2 system of conservation laws
,”
Proc. Edinburgh Math. Soc.
55
(
3
),
711
729
(
2012
).
17.
M.
Mazzoti
,
C. J.
Tarafder
,
F.
Gritti
, and
G.
Guiochon
, “
Experimental evidence of a delta shock in nonlinear cromatography
,”
J. Cromatogr. A
1217
,
2002
2012
(
2010
).
18.
A. R.
Domingos
, “
Multiplication of distributions and singular waves in several physical models
,”
Eur. J. Math.
10
,
12
(
2024
).
19.
Z.
Ouyang
, “
Singular travelling wave solutions for a generalized Camassa Holm equation
,”
J. Appl. Math. Comput.
6
(
3
),
380
389
(
2022
).
20.
A.
Paiva
, “
Formation of δ-shock waves in isentropic fluids
,”
Z. Angew. Math. Phys.
71
(
4
),
110
(
2020
).
21.
A.
Paiva
, “
New δ-shock waves in the p-system: A distributional product approach
,”
Math. Mech. Solids
25
(
3
),
619
629
(
2020
).
22.
A.
Paiva
, “
Interaction of Dirac δ-waves in the nonlinear Kline-Gordon equation
,”
J. Differ. Equ.
270
,
1196
1217
(
2021
).
23.
Y.
Pang
,
L.
Shao
,
Y.
Wen
, and
J.
Ge
, “
The δ′ wave solution to a totally degenerate system of conservation laws
,”
Chaos, Solitons Fractals
161
,
112302
(
2022
).
24.
Y.
Pang
,
Y.
Wen
, and
C.
Xu
, “
The Riemann problem for pressureless compressible fluid system with time- and space-dependent external force
,”
Results Phys.
54
,
107115
(
2023
).
25.
Y.
Pang
and
C.
Xu
, “
The propagation and collision behaviour of δ′ waves in a model of three partial differential equations
,”
Z. Angew. Math. Phys.
75
,
14
(
2024
).
26.
A.
Sen
and
T.
Raja Sekhar
, “
The multiplication of distributions in the study of delta shock waves for zero-pressure gasdynamics system with energy conservation laws
,”
Ric. Mat.
72
,
653
(
2023
).
27.
C.
Shen
, “
The multiplication of distributions in the one-dimensional Eulerian droplet model
,”
Appl. Math. Lett.
112
,
106796
(
2021
).
28.
C.
Shen
and
M.
Sun
, “
A distributional product approach to the delta shock wave solution for the one-dimensional zero-pressure gas dynamics system
,”
Int. J. Nonlinear Mech.
105
,
105
112
(
2018
).
29.
C.
Shen
and
M.
Sun
, “
The multiplication of distributions in the study of delta shock wave for a hyperbolic system of Temple class
,”
Int. J. Math.
32
(
14
),
2150103
(
2021
).
30.
M.
Sun
, “
The multiplication of distributions in the study of delta shock wave for the nonlinear chromatography system
,”
Appl. Math. Lett.
96
,
61
68
(
2019
).
31.
C. O. R.
Sarrico
, “
New solutions for the one-dimensional nonconservative inviscid Burgers equation
,”
J. Math. Anal. Appl.
317
,
496
509
(
2006
).
32.
C. O. R.
Sarrico
, “
Products of distributions and singular travelling waves as solutions of advection-reaction equations
,”
Russ. J. Math. Phys.
19
(
2
),
244
255
(
2012
).
33.
C. O. R.
Sarrico
and
A.
Paiva
, “
Distributions as initial values in a triangular hyperbolic system of conservation laws
,”
Proc. R. Soc. Edinburgh, Sect. A: Math.
150
,
2757
(
2019
).
34.
C. O. R.
Sarrico
, “
Distributions as travelling waves in a nonlinear model from elastodynamics
,”
Physica D
403
,
132328
(
2020
).
35.
C. O. R.
Sarrico
, “
New singular travelling waves for convection–diffusion–reaction equations
,”
J. Phys. A: Math. Theor.
53
,
155202
(
2020
).
36.
C. O. R.
Sarrico
, “
The movement of a particle according to the Gurevich–Zybin dark matter model
,”
J. Math. Phys.
62
,
051509
(
2021
).
37.
C. O. R.
Sarrico
, “
Travelling waves for the Brio system
,”
J. Nonlinear Sci.
31
,
69
(
2021
).
38.
C. O. R.
Sarrico
and
A.
Paiva
, “
The multiplication of distributions in the study of a Riemann problem in fluid dynamics
,”
J. Nonlinear Math. Phys.
24
,
328
345
(
2021
).
39.
C. O. R.
Sarrico
, “
About a family of distributional products important in the applications
,”
Port. Math.
45
,
295
316
(
1988
).
40.
L.
Schwartz
,
Théorie des Distributions
(
Hermann
,
Paris
,
1965
).
41.
C. O. R.
Sarrico
, “
The multiplication of distributions and the Tsodyks model of synapses dynamics
,”
Int. J. Math. Anal.
6
(
21
),
999
1014
(
2012
).
42.
C. O. R.
Sarrico
, “
Emergence of δ′-waves in the zero pressure gas dynamic system
,”
Int. J. Math.
31
,
2050031
(
2020
).
43.
V. C.
Rubin
, “
One hundred years of rotating galaxies
,”
Publ. Astron. Soc. Pac.
112
,
747
750
(
2000
).