In work, it is constructed a discrete mathematical model of motion of a perfect fluid. The fluid is represented as an ensemble of identical so-called liquid particles, which are in the form of extended geometrical objects: circles and spheres for two-dimensional and three-dimensional cases, respectively. The mechanism of interaction between the liquid particles on a binary level and on the level of the n-cluster is formulated. This mechanism has previously been found by the author as part of the mathematical modeling of turbulent fluid motion. In the turbulence model was derived and investigated the potential interaction of pairs of liquid particles, which contained a singularity of the branch point. Exactly, this is possible to build in this article discrete stochastic-deterministic model of an ideal fluid. The results of computational experiment to simulate various kinds of flows in two-dimensional and three-dimensional ensembles of liquid particles are presented. Modeling was carried out in the areas of quadratic or cubic form. On boundary of a region satisfies the condition of elastic reflection liquid particles. The flows with spontaneous separation of particles in a region, various kinds of eddy streams, with the quite unexpected statistical properties of an ensemble of particles characteristic for the Fermi-Pasta-Ulam effect were found. We build and study the flow in which the velocity of the particles is calibrated. It was possible using the appropriate flows of liquid particles of the ensemble to demonstrate the possibility to reproduce any prescribed image by manipulating the parameters of the interaction. Calculations of the flows were performed with using MATLAB software package according to the algorithms presented in this article.
Previously in several papers [
In the previously constructed model from the equations of inviscid incompressible fluid, it was deduced the potential used as an interaction potential between individual marked points of a continuous medium. The presence of the mathematical feature of the branching type is showed by the study of this potential in the framework of the two-body problem. This feature was considered as one of the most important characteristics of the turbulence phenomenon.
Presented in this paper, mathematical model of the motion of a perfect fluid is constructed from such point of view, when individual marked points of fluid are loaded with a weight and turn into discrete liquid particles. After specification of the mechanism of interaction of fluid pair particles model takes complete features and, in particular, allows performing a set of computational experiments on modeling different kinds of various flows. At all stages of constructing this model the effects associated with viscosity and thermal conductivity, are not taken into account. For this reason, this model is presented as a discrete model of a perfect fluid.
Overall, the model constructed below allowed us to separate deterministic and stochastic components in the description of turbulent flows of an ideal fluid. This separation is provided on an analytical level access to the mechanism of generation of stochastics and, consequently, to the possibility of constructing a typical representatives of the various classes of flows.
Note that the stochastic component in the dynamics of liquid particles in this model is a consequence of their interaction with each other and reflects the presence of the branching mechanism, previously studied on the example of the description of the dynamics of pairs of marked points of liquid. This type of stochastic is different, for example, from the stochastics specific for the class of lattice models of the Boltzmann equation (Lattice Boltzmann Method), used to describe the hydrodynamic flows [
There is the well known method of direct numerical simulation of turbulent flows by solving the Navier- Stokes equations [
Various possible interpretations of a discrete object of a turbulent fluid can be divided into two groups. The first of these will take the submission of the discrete object as a particle. In this case, a particle refers to an object that is defined in the sense of classical mechanics, i.e. it has a fixed mass, position in space, velocity, etc. The second group will take the representation of the discrete object in the form of a wave. Under the wave is understood generally a geometrical structure through which the mass, energy, and other physical parameters flow. As an example, the representative of the second group is a vortex. The division into two groups in the interpretation of discrete liquid object is conditional, because in fact due to the turbulent mixing is not possible to draw a clear distinction between them. Thereby, the discrete object will be interpreted as both particle and wave, i.e. within the framework of a kind of “wave-particle” or “wave-corpuscle” dualism.
We note a number of works in the field of computational fluid dynamics, which may be interpreted from the standpoint of wave-particle duality. In the works of S. Ulam and coauthors [
In general, the Eulerian and Lagrangian approaches of the continuous medium description are illustrative in connection with the dualism of particles and waves. For example, in the Eulerian approach being in a fixed point in space, the observer monitors the motion of the fluid that expressly assumes the wave interpretation of the cell in which it is located. In the Lagrange approach observer is located on the moving fluid, i.e. follows the movement of fixed particle. In the theoretical hydrodynamics Euler and Lagrange approaches are considered equivalent, but in practice the Euler approach prevails. Such preference Euler approach is due to the fact that in the Lagrangian approach for turbulent mixing to determine the particle it is impossible once and for all.
In this article considered in detail two-dimensional, the most studied case. Managed to build and explore using the computational experiment a number of interesting flows. Some regimes that provide separation of the particles in the ensemble were detected. Different kinds of vortex flows were simulated. Some interesting statistical properties of certain flows were detected. The flows, in which there is a concentration of kinetic energy in a small number of particles of the ensemble were detected. Discovered and studied also the flows, in which the kinetic energy of the particles is calibrated. In addition, it was possible using the appropriate flows of liquid particles of the ensemble to demonstrate the possibility to reproduce any prescribed image by manipulating the parameters of the interaction.
Briefly recall the procedure of entering into the consideration a discrete liquid particle under the model of turbulence developed earlier by the author. Write the equations of inviscid incompressible fluid:
where i, j = 1, 2, 3; t―time, r = (x1, x2, x3)―radius-vector, v = v (t, r), P = P (t, r)―the velocity and the pressure fields, r = const > 0―density of the fluid (on repeated indices is implied summation).
Applies to the second equation in (1) the operation ¶/¶xi and take into account the condition of continuity, then
Formally, using the Green’s function, invert in (2) the Laplace operator. Performing twice the integration by parts assuming no motion at infinity and differentiating the resulting expression with respect to xi after some transformations and substitution in (1), we find
where v¢ = v(t, r¢), dr¢―element of volume in r¢-space.
At the Lagrangian approach in the description of the liquid is introduced vector function r = r (t, a), which represents the trajectory of the point with index
where
We will approximate the integral in (3) in the sum by breaking the area of integration for simplicity on the same small amounts w, then
where g =w /4p; n, m―numbers of small volumes w.
Thus, using the hypothesis of a discrete structure of turbulent fluid, reduce the study of the integro-differential Equation (3) to a system of ordinary differential Equation (4). At this stage, under the discrete object we will understand a point. Rewrite the system of Equation (4), introducing a binary interaction potential fn,m of the pair of points with numbers n and m:
Suppose that for discretization of the space of initial conditions the fluid motion is approximated not by the motion of points, but by the motion of particles, then the following difficulty arises in the interpretation. The presence of turbulent mixing, as is known, leads ultimately to the fact that the particle, initially well defined in the context of the ensemble of particles will eventually fill the whole of the observed region of space. Thus, the particle’s individuality will be completely “erased” and it as though will “dissolve” into other particles. This is one of the main reasons why the equations of continuous medium in the form of Lagrange are so difficult to use in practice. That is why discrete objects at the level of Equation (5) are interpreted as points because it is natural to assume that the point remains the point throughout the time of movement.
Equation (5) were “prepared” in the form of a dynamic system the behavior of which is described by a set of radius vectors r(n) and the velocity V(n). The potential fn,m was first obtained in [
In the next section, we investigate the problem of motion of a pair of points. After that the points will be “loaded” weight and become more similar to liquid particles.
We rewrite (5), denoting by r1, r2; v1, v2―the radius and velocity vectors of a pair of fixed points of the liquid, then
where v12 = v1 - v2, r12 = r1 - r2.
The system of Equation (6) has an integral of motion: v1 + v2 = const, which can be regarded as an analogue of the total momentum of the system. The existence of this integral fully characterizes the movement of the pair of points as a whole in the form of a rectilinear and uniform motion.
Let us turn to the study of the relative motion of a pair of points. For this purpose we introduce the notation r = r12, v = v12 and after the obvious manipulation in (6) we find
By a direct calculation one can verify that the system of Equation (7) allows the following integrals of motion:
where
The integrals K and E can be considered as analogues of the conservation laws of angular momentum and energy. Going over to the coordinate system in which the vector K is directed along the axis x3, then the vectors r and v are in the plane of the variables x1, x2. We introduce in plane polar coordinates for the vectors r and v according to
where y =b - a, while
The system of Equation (8) can be reduced to a pair of equations
where y = ctgy. Dividing the first equation in (9) to the second one and introducing the dimensionless distance x = g-1/3r, we obtain
Equation (10) has been numerically investigated earlier [
In
The term “quasipair” was introduced in previous papers to designate to a pair of points that in the process of interaction fall to a common center, and stick together. It can be shown that the formation and decay of quasipair
take place in a finite time, and from its formation to the breakup it “remembers” on integrals of motion K and E. This can be seen by studying the second Equation (9) in the neighborhood of a singular point A2, then, assuming
that y ® ?1/2, find,
in terms of time, we can make a change of variables
Let us consider the process of disintegration of the quasipair. Assume that it is formed by moving the pair of points in accordance with one of the trajectories belonging A2 (
In the vicinity of r = 0, i.e., at the time of the fall (decay) of the points to each (from) other, according to (8), we have
Thus, to describe the process of formation and decay of quasipair it is not enough Equation (7). It is necessary to define additionally two mechanisms: the transition of quasipair from state with y = ?1/2 to a state with y = +1/2 and equiprobable selecting of one of a set of the trajectories which are permitted to occur in the plane perpendicular to the vector K. Thus we can speak about the presence of the branching point singularity in Equation (7).
From the standpoint of the problem of describing of a pair of points motion, so formed quasipair can exist infinitely long due to the fact that any later date is not highlighted. The uncertainty in the lifetime of the quasipair will be eliminated with using the liquid particles ensemble constructed below.
The presence of branching in a system consisting of a pair of points, suggests that this phenomenon is typical for Equations (5), which describe the motion of an ensemble of points of the liquid. Now, a natural question arises. How to relate to each other the results of the study of the motion of an ensemble of points of the fluid with the original equations of inviscid incompressible fluid?
Note that the transition from continuous to discrete in the form in which it was done above leads to the fact that talk about the condition of continuity in a discrete aspect is meaningless. For this reason, the system of Equation (5) has only indirect relevance to nonviscous incompressible fluid equations. The transition to a discrete representation and the study of it from the point of view of the movement points of a liquid allows us to take a new look at the main hydrodynamic nonlinearity, i.e. quadratic in the velocity term in the equations of motion of a continuous medium. In this case, Equation (5) can be considered as a special tool that allows you to reveal the mechanism of branching, due to the quadratic in the velocity hydrodynamic nonlinearity.
It is interesting to compare our point of view with the position of the dynamic chaos [
of stochastics for resolution of the question of choosing a certain trajectory from the set of permissible ones. In this case, the stochastic mechanism is external with respect to the equations describing the motion of fluid points. In the model of dynamic chaos, in the strict sense, there is no stochasticity. In this case, the term “stochasticity” has the meaning, such as random choice from the set of feasible outcomes. In this regard, the following questions arise. Have generally any sense the branching themselves? Whether branching is a product of the inadequacy of our way of describing the continuum? It seems that the discovered mechanism of branching is an extreme form of manifestation of turbulence in the fluid motion representation logic describing the set of points movement.
In sections 1, 2, fluid flow was considered as a movement marked points. In this case, it was found that a pair of points can fall on each other over a finite time to form an intermediate point―quasipair. Quasipair after a certain period of time is divided into two points, and there are many ways of break up, satisfying the integrals of motion. Possible decay trajectories are characterized by the presence of features such as branching in the mechanism of formation and decay of quasipair.
Now put the following task. We will construct an ensemble of discrete liquid particles again using only a few necessary results mentioned in Sections 1, 2, in particular those related to the mechanism of formation and decay of quasipair. For this we will represent a liquid as a set of separate liquid particles. Following the analogy between turbulence and ordinary gas atoms, the liquid will be presented initially in the form of rarefied gas of liquid particles that move between collisions rectilinearly and uniformly. We assume that the chosen discrete liquid particles are of circular shape for two-dimensional case and the spherical form in three-dimensional space.
Let a pair of liquid particles moving closer to a distance equal to or less than the sum of their radii, then assume that the pair coalescence occurs with the formation of an intermediate particle, i.e. quasipair. The quasipair is a metastable object and after some time splits into a pair of liquid particles. Formation and decay of quasipair stepwise interprets the collision of a pair of liquid particles, in this case should be the laws of conservation of mass, momentum, energy and angular momentum.
We assume that the quasipair exists during the interaction of a pair of particles, i.e., quasipair is another name of the mechanism of interaction. By particle collisions, the size and weight of which, generally speaking, are unequal to each other, the quasipairs are formed, which then decays into a pairs of identical particles. The last condition is justified by the fact that quasipair has one preferred axis passing through its center along the angular momentum. Thus, if the original liquid was represented by a set of liquid particles with a corresponding distribution of volumes and masses, the further evolution of the system will lead to a state where the volumes and masses of the particles are the same. Hereinafter we assume that the masses of all liquid particles are equal and have the value of m. It should be noted that the decay of the quasipair into two identical liquid particles was previously interpreted by the author as an elementary act of turbulent mixing.
If we consider the mass of the particles as a marker of individuality then due to the mechanism of formation and decay of quasipair and mass redistribution between the particles, it follows that the individuality of particles is not saved―they constantly acquire new masses and lose old ones. In other words, they can be regarded as the objects of the wave nature. In the intervals between collisions, discrete objects may be considered as liquid particles, in the moment of collision and during the quasipair’s lifetime the wave nature of fluid particles is emerged. In this it manifests itself wave-particle duality of the discrete object, interpreted in this model as a liquid particle.
At first, we study an ensemble of two-dimensional liquid particles. Without loss of generality, we place an ensemble of particles into the region shape a square with a side of L. For simplicity, the square-shaped area we will call a box. We assume that the liquid particles are elastically reflected from the walls of the box.
Let us choose some size L of the original box. Fluid particle we will represent as a circle with a radius r0. Position of the particle will be characterized by the radius vector r = (x, y), where x and y―abscissa and ordinate of the center of the liquid particle in the form of a circle. The velocity of the particle will be denoted by the radius vector v = (vx, vy), where vx and vy―projections of the velocity vector v on the axis of abscissa and ordinate.
Let at some point of time t the position and the velocity of the liquid particle are r and v respectively. We
define the time step t, and the position and velocity of the fluid particle in the time moment t + t and we denote by the symbols
In the case where
In the case where
Conditions (11)-(13) provide the motion of the center of the liquid particle within the box, in this case the particle is elastically reflected from the walls.
Let us consider in the area of the square form a pair of liquid particles, which interact with each other and with the walls of the box. Interaction with the walls describe according to the algorithm (11)-(13). Separately focus on describing the interaction of pair liquid particles.
Denote the current position and velocity of the particles before the interaction radius vectors r1, r2 and v1, v2, respectively. After the interaction, the position and velocity of particles is denoted the radius vectors
Define the known in mechanics two pairs of variables r, R and v, V to describe the interaction of two particles of equal mass:
The vector R defines the position of the center of mass of the pair of particles, and V―velocity vector of the center of mass. The vector r describes the relative radius vector of the position a pair of particles, and v―rel- ative velocity pair of the particle.
From (14) we can write the position and velocity of each of the particles through pair new variables:
Let at the interaction of pair liquid particles the law of conservation of momentum is satisfied, then
After substitution of the second pair of Equation (15) to the left and similarly to the right side of Equation (16), we obtain
Ensure compliance with the law of conservation of energy, then
After substitution of the second pair of Equation (15) to the left and similarly the right side of (18), we obtain
Taking into account (17), we find
where v = |v|, v¢ = |v¢|.
At last, we write the condition for the fulfillment of the law of conservation of angular momentum, i.e.
Taking into account (15) Equation (21) can be rewritten as:
In connection with the implementation of the law of conservation of angular momentum, we assume that the interaction does not change the position of the center of mass of the pair of fluid particles, i.e.
Taking into account (17), (23) in (22), we find
Let a and b the angular deviation of the vectors r and v from the x-axis. Analogously angles a¢ and
where r = |r|, v = |v|; r¢ = |r¢|, v¢ = |v¢|.
Substituting the polar representation of the relative position and velocities in vector cross products (24), we find
Along with the (24) we make another assumption about the mechanism of interaction in connection with the law of conservation of angular momentum, namely
Assumptions (23) and (26) clarify the geometry of the interaction of a pair of fluid particles and may be considered as postulates.
We introduce notation for the relative angles between the radius vectors of positions and velocities: y = b ? a and
Equation (27) has two classes of solutions:
1)
where
Formulate a criterion of interaction of pair of liquid particles in the form both of the following two conditions:
where (r, v)―scalar product of two vectors. According to the criterion (29), a pair of liquid particles interact under two conditions: firstly, the distance between the particles become less or equal to two radii of particles and, secondly, the projection of the relative speed on the relative distance is negative, i.e. centers of particles approach one another.
The criterion of interaction (29) can be rewritten as:
Taking into account (30) let us assume that the interaction is terminated when condition
is satisfied.
It is easy to see that condition (31) is satisfied on the second class of solutions (28). Indeed, since
Let us write
can be represented as:
where
where sign(K)―signum-function of magnitude K.
Substituting (33) in the second class of solutions (28), we find
Taking into account that
Equations (34), (35) are the key issue in the study of the interaction of a pair of liquid particles. According to Equation (35) angular deviation from the abscissa a¢,
As a consequence of Equation (35) occurs an additional freedom of choice of angles in each act of interaction of pair of particles. For example, suppose that in any way the angle a¢ is determined, then the angle
To complete the description of the mechanism of interaction of a pair of liquid particles it is necessary to supplement a new configuration of a pair of particles with criterion that it is fitted in the selected box.
Positions and velocities of particles in a moment of time t before and after the interaction denoted by the symbols: r1 = (x1, y1), r2 = (x2, y2), v1 = (v1,x, v1,y), v2 = (v2,x, v2,y) and
If any of the new positions
The procedure includes the following checks and transformation: if
then
cases, it is considered that
As a result, we can formulate the following algorithm, which simulates the movement of the pair of liquid particles interacting with each other and with the walls of the box. Suppose that at time t the positions and velocities of particles are r1, r2 and v1, v2, respectively. Find at time t + t new positions
We introduce a second constraint along with the constraint
To complete the procedure of modeling the motion of a pair of particles in the area of the square form is necessary for each act of interaction between pairs of particles to clarify the option of angles a¢,
Consider the example where the angle a¢ is chosen uniformly randomly from the interval [0, 2π]. Taking into account (35) we write
where x―uniformly distributed on [0,1] random variable.
Let us examine with how many particles simultaneously interact with each other according to the criterion (29), (30). In
From geometrical considerations it is clear that with increasing fluid particle radius r0 the number of binary relationships with a random arrangement of positions and velocities must increase, while the single binary relationships will be less. In other words, the particles will interact with many other particles. According to
Let us study question about the number of 2-clusters. Denote the number of 2-clusters in the ensemble of the liquid particles through Cl2. The fraction of 2-clusters in the set of all binary relations we denote by S2. Under share 2-clusters refers to the ratio of the number of 2-clusters Cl2 to the total number of binary relations Nb, i.e.
The curve on the left graph
relation
of the liquid particles. When
As in the study of the interaction of pairs of particles, we define the position R and the velocity V of the center of mass of some n-cluster, i.e.
where
A set of n particles takes the form n-cluster when any particle from the set
where ri,j = ri ? rj, vi,j = vi ? vj.
Taking into account (37) we write a representation for the positions and velocities of the particles of n-cluster, namely
where
According to (40) it is easy to verify that
Let us consider the law of conservation of momentum in the n-cluster. In this case we assume that the equality of such form is true:
where a dash will denote the corresponding characteristics of the particles after interaction in the n-cluster, i.e.
Taking into account (37), (39), the law of conservation of momentum can be rewritten in the form:
According to the law of conservation of momentum in the form of (42) velocity of the center of mass of the n-cluster after the interaction does not change.
Let us consider energy conservation law, more precisely the kinetic energy in the n-cluster, namely
Substituting in (43) representation for the particle velocity from (39), we find
According to the law of conservation of energy in the form of (44) the sum of squares of the velocities
Let us consider law of conservation of angular momentum in the n-cluster. We write it as:
where
Substituting the representation (39) in (45), we find
In connection with the law of conservation of angular momentum in the form of (46) will make a number of statements that can be considered as additional postulates concerning the mechanism of interaction in the n-cluster. We assume that after the interaction position of the center of mass of the cluster particles does not change, i.e.
Taking into account (42), (47) in (46), we rewrite the condition of conservation of angular momentum in the form:
At this stage, being limited two-dimensional case, we can write the following representation for the positions and velocities of the particles of n-cluster before and after the interaction:
where
Substituting in (48) presentation (49), we find
where
Will have two statements regarding the mechanism of interaction in the n-cluster in connection with the implementation of the law of conservation of angular momentum, namely
According to (51), it is believed that the modulus of distances and velocities relative to the center of mass of particles in n-cluster remain unchanged after the interaction. In particular, the second condition ensures that the law of conservation in the form of (44).
Taking into account (51), we rewrite Equation (50), then we find
Equation (52) is the generalization to the n-cluster Equations (27), constructed for 2-cluster in connection with the implementation of the law of conservation of angular momentum.
Following (28), will choose as the solutions of Equation (52), the following two sets of values of the angles yi and
1)
where
To choose as one of the solutions in the two sets of (53), let us consider in more details the criterion of interaction (38) in terms of the sign of the scalar product
According to the criterion of interaction in the form (38), if a pair of particles interact, then
Taking into account (51) write the expression for the scalar product before and after the interaction, then
In addition to Equations (54), (55) it is still need to take into account the requirement that
In summary, we conclude that it is necessary to choose the angles
where
As a result, according to (54)-(57), succeeded to find such conditions under which an arbitrary n-cluster ceases to exist after the interaction. In other words, the binary relationships that ensure the existence of the n-cluster cease to exist after the interaction according to the scheme (54)-(57).
For example, consider the experiment of the formation and disappearance of 4-cluster with a special choice of the initial data. Suppose that from the four corners of the original box with the same velocities directed along the diagonals begin to move 4 identical particles. After some time, they are found in the center region, forming a 4-cluster. It is necessary to observe the further dynamics.
After studying the interaction of particles in n-cluster has became clear a possible algorithm for calculation the dynamics of the ensemble. Let at some time t are found all the clusters which in general can be divided into 2-clusters, 3-cluster,
Let expand the specified set of clusters formally with the 1-cluster, when the particle is included in that 1-cluster does not interact with other particles of the ensemble. In this case at any point in time there must be true equality of the form:
Consider further the algorithm of interaction in an ensemble of particles, when the basic is the binary mechanism of interaction of particles. In this case, a special role is played by 2-clusters, and for clusters of higher orders interactions will be described according to the schemes (53)-(57).
Consider an ensemble of N liquid particles placed in the square box with a side L. We assume that the particles interact with each other and with the walls in accordance with the procedures defined in the previous two sections.
For the numerical simulation of an ensemble of interacting liquid particles it is necessary to clarify the angular deviations of the radius vectors of relative positions and velocities of interacting particles.
Taking into account section 4, collect the necessary provisions for the binary interaction mechanism. Let the criterion of interaction (29) is satisfied for i-th and j-th particles, i.e.
where ri,j = ri ? rj, vi,j = vi ? vj, here it is assumed that the i-th and j-th particles interact only with each other. After the interaction the new values of the angular deviations
where
Taking into account section 5, we will collect the necessary provisions for the description of the interaction in the k-cluster, where k > 2. Suppose that some k-cluster includes particles with indices
where
Let the ensemble of particles is simulated numerically in time increments t. We denote the number of time steps with symbol T. We define during the simulation time the number of k-clusters, where k > 2, with the symbol Clk > 2, similar to the number of 2-cluster―with the symbol of Cl2. Fraction of k―k-clusters, k > 2 in the total set of clusters for all the simulation time we calculate by the formula:
We construct the characteristic types of motion of fluid particles ensemble, taking into account the possibilities, which are provided with the choice of angles
Example №1. Discuss the typical types of flows when changing the particles size, i.e. varying the radius r0.
To determine the binary interaction mechanism we choose the following angles:
, (62)
where xi,j―a random number uniformly distributed on the interval [0, 1].
To determine the interaction in the k-clusters, k > 2 we use the formula (60), in which the parameter μ we define according to the condition:
where h―a random number uniformly distributed on the interval [0, 1].
Formulas (62) and (63) are sufficient to carry out the computational experiment on modeling of motion of fluid particles ensemble.
Under the density r of an ensemble of particles in a given box we will mean the ratio of the total area of the particles to the area of the region of localization of the ensemble, i.e.
In all further calculations the initial positions and velocities of particles were chosen randomly according to the algorithm: xi = ?L/2 + Lxi, yi = ?L/2 + Lhi; similarly for velocities vi,x = ?Vmax + 2Vmaxli, vi,y = ?Vmax + 2Vmaxmi; where xi, hi, li, mi―uniformly distributed on [0, 1] random variables,
The same conclusions are confirmed by counting the proportion k―k-clusters, k > 2 in the total number of clusters, starting with the binary. It appeared that this parameter is equal 39% and 100% for
Example №2. We construct the flow, leading to separation of the original uniformly chaotic motion of particles in a box.
We choose the angles to describe the binary interactions in the following form:
, (65)
To describe the interactions in k-clusters, k > 2, we assume that the average angle of the relative deviations of
particles of k-cluster does not change after the interaction, i.e.
it possible to determine the choice of the parameter μ, therefore, let μ = 0.
According to
In the following examples, the flows will be addressed in more complicated versions of the range of angles
Example №3. Construct a flow in which the relative radius-vector of the provisions of a pair of interacting
particles is directed along (against) the radius-vector of center of mass
Determine the angle
particles is directed along the radius-vector of center of mass, we write
The choice of the angle
Taking into account (66), we write the expression for the calculation of the angle
Angle μ, is present in the formula (60) to describe the interaction in k-clusters, k > 2, we construct by analogy with (66), i.e. believe that
where
We define the average per the number of time steps T, the function of the particle density
where nx, ny―the number of nodes in the grid, and hx, hy―the steps of the meshes along the axes of abscissa and ordinate.
Define accumulated during a determined time t the density of particles at point
where [...]―integer part of number, the index i runs over all particles of the ensemble, i.e.
Instant distribution of N = 500 centers of the particles of the ensemble in the last computational step T = 5000 in
The following example is the supplement to the previous one.
Example №4. We construct a flow in which the relative radius-vector of the positions of pair of interacting
particles is directed perpendicular to the radius-vector of center of mass
Determine the angle
particles is directed perpendicular to the radius-vector of center of mass, we obtain
The angle
where
Example №5. Construct a flow in which the relative radius-vector of velocity of the pair of interacting particles directed along (against) the radius-vector of center of mass
This example is the supplement to the Example №3, since in this case it is assumed that the angle
where, as above
The parameter μ we define by formula (68).
The typical form of the flow calculated according to the algorithm (68), (74), (75) is presented in
To obtain time-averaged velocity fields, the values of the velocities
The following example is a supplement to the previous one.
Example №6. We construct a flow in which the relative radius-vector of the velocity of a pair of interacting
particles is directed perpendicular to the radius-vector of the center of mass
Determine the angle
directed perpendicular to the radius vector of the center of mass, we obtain
Taking into account (76), we find the angle
where
In
Example №7. We construct a flow in which the relative radius-vector of velocities of a pair of interacting
particles is directed along (against) the velocity vector of the center of mass
Determine the angle
vector
Knowing, according to (78),
interactions in the k-clusters, k > 2, suppose that angle
horizontal axis of the velocity vector of the center of mass of the cluster particles V = (Vx, Vy), wherein
Taking into account the second equation in (60), we find
where f = arctg|Vy/Vx|, and
Under review in the example the class of flows proved to be extremely interesting in terms of velocity distribution of fluid particles during the evolution of the dynamics of the ensemble. It turned out that the kinetic energy, initially uniformly randomly distributed between the particles, over time is redistributed in favor of a small group of particles the velocities of which are noticeably larger than the others. The flow with such properties is similar to well known in the theory of solids the effect of the Fermi-Pasta-Ulam. To characterize the resulting velocity distribution, we compare it with the one that follows from the Maxwell distribution.
For particle velocities Maxwell density distribution f (vx, vy) on the plane can be written as follows:
where
Modeling the dynamics of an ensemble of liquid particles, we construct an empirical density distribution as a histogram and compare the resulting density with the density (81). This will allow us to understand how the resulting distribution of the modules of liquid particles velocities deviates from the Maxwell distribution.
In addition to describe the obtained flows we introduce the function N90, which characterizes in descending order the most energetic particles, the total energy of which amounts to 90% of the initial kinetic energy. Numerically construct the N90, as a function of time t, i.e. find N90 = N90(t).
In the left graph
corresponds to the density of Maxwell for the module speed. One can see that the histogram and the density (81) have nothing in common. According to the histogram, the majority of particles (over 80%) are practically at rest. Finally, on the right chart of
Thus, using a choice of angles according to the scheme (75), (78), (79) it was succeeded on the basis of mainly the binary mechanism of interaction (e.g., for computational experiment in
The following example is a supplement to the previous one.
Example №8. We construct a flow in which the relative velocity vector of a pair of interacting particles is
directed perpendicularly to the velocity vector of the center of mass
Taking into account (78), determine the angle
Knowing according to (82),
To describe the interaction in the k clusters, k > 2, suppose that the angle
perpendicular to the velocity vector of the center of mass of particles in the cluster V = (Vx, Vy). Taking into account the second equation in (60), we find an equation to determine the angle μ, which is exactly the same as Equation (79), where it is believed that f = arctg|Vx/Vy|, and
In
In the left chart
The standard deviation was calculated according to the known formula in statistics:
Graph of dynamics vstd = vstd(t) in
The following example will demonstrate the fundamental possibility of reproduction of any images by using the dynamics of an ensemble of interacting particles. To achieve the desired, will find appropriate angles in the algorithm of particles interaction (58)-(60).
Example №9. Using a suitable algorithm of interaction in an ensemble of liquid particles will play the picture in the form of an abbreviation of the Moscow State University, i.e. MSU.
The picture with the letters MSU will prepare so that it is in pixels had dimensions of 100 ´ 100 and was purely black and white with no shades of gray. This condition allows to clearly define the indicator function MSU = MSU(x, y), which returns a 1 if the pair (x, y) indicates the location of one of the letters MSU and vice versa, ―0, if the dot indicates the background.
We define the corresponding angles to clarify the type of interaction between the liquid particles in pairs and in k-cluster, k > 2.
To describe the pair interaction we will take as a basis the examples №1 and №2. We assume that in the field of localization of the image background the angles are selected, according to the algorithm of example №1, and in places of localization of the letters MSU―according with some modification of example №2. As a result, we assume that
when MSU(Xi,j, Yi,j) = 0, Xi,j = (xi + xj)/2, Yi,j = (yi + yj)/2;
when MSU(Xi,j, Yi,j) = 1, t―the number of the time step, and T―the total number of time steps. The expression for
By analogy with (84), (85) we define the angles for the description of k-cluster, k > 2, which includes particles
when
To identify the original picture in terms of the characteristics of an ensemble of particles was calculated and averaged over the number of time steps T the function of particles density
From the analysis of
In the following example, consider the combination of flows investigated in examples №7, №8. In the end, manage to build a flow in which a particle ensemble in a box split.
Example №10. We construct a flow in which the relative radius vector of the velocities of a pair of interacting
particles is directed along the velocity vector of the center of mass
half of the box, and vice versa―in the right half of the box it is perpendicular to the velocity vector
Describe the procedure of angles choice for binary interactions. For this, we introduce the notation
As well as
Knowing, according to (87), (88),
To describe the interaction in the k-cluster, k > 2, with numbers of particles
mass of the k-cluster. Following the reasoning of example №7, determine the unknown angle μ, included in the formula (60), as follows:
where
And X is the abscissa of the vector of the center of mass of the k-cluster.
On the left graph
Visual analysis of the plots in
This example allows you to return to example №9, which was reproduced using the flow of an ensemble of particles, the image in the form of letters MSU. If in the example №9 this picture was identified after the time averaging of the particles density, then using the algorithm of example №10 it can be done directly without the procedure of averaging.
Example №11. Using the algorithm of interaction of the particles of the previous example, reproduce with the help of an ensemble of particles the abbreviation for Moscow State University, i.e. MSU.
In the example №9 the indicator function MSU = MSU(x, y) is built. This function returns 1 if the pair (x, y) indicates the location of one of the letters MSU and vice versa―0 if the point indicates the background. We assume that in the field of localization of the letters MSU is implemented the scheme of choice of angles (87), and for points of background is realized the scheme (88) for the binary mechanism of the interaction of particles. Similarly to (89) we determine the angle μ to describe the interaction in the k-cluster, k > 2.
As a criterion for the interaction of pairs of particles in three-dimensional space we choose, as in the two-dim- ensional case, the criterion (29), where the relevant vectors we mean vectors in three-dimensional space.
To describe the binary interaction of the pair of liquid particles in three-dimensional space it is necessary to express concerns and to carry out all the calculations, as in section 4 up to (24), which appears in connection with the fulfillment of the law of conservation of angular momentum. Rewrite (24) as:
where the prime denotes the new relative position vectors r¢ and speed v¢ of the pair of particles after the interaction.
The unit vectors of the original coordinate system we denote as a set of vectors
Associate the unit vectors in the new coordinate system using the unit vectors in the old coordinate system
according to the equation,
pair of coordinate systems. In this case, the coordinates of an arbitrary vector q in different coordinate systems are related according to the equation:
Taking into account the designations of the two-dimensional case, we consider Equation (90) in the new coordinate system. Represent vectors
where y―the angle between the vectors r and v. Taking into account the conservation law of angular momentum in the form (90) after interaction of the pair of particles the vectors
Taking into account (92), (93), the expression (90) we can rewrite in the form of (27) and then in the form of (28). Now considering that there is an interaction between pairs of particles, i.e., the interaction criterion (29), (30) is met, we choose the second solution in (28), then we get a solution
where
assume that sign (K) = 1 because by construction the unit vector
Discuss one of the possible algorithms describing the interactions in the n-cluster, n > 2. Taking into account formulas (37)-(48), which are devoted to the description of the n-cluster in a two-dimensional case, we write our basic equations that must be satisfied in the context of the implementation of the laws of conservation of energy and angular momentum:
, (94)
As well as the normalization conditions:
As a solution of the system of Equations (94) and (95) is the set of relative positions
We write the representation:
where the parameters Qx, Qy, Qz and Ux, Uy, Uz are considered yet unknown.
The sets px,i, py,i, pz,i, sx,i, sy,i, sz,i,
Similarly for the coordinates y and z.
Played sets px,i, py,i, pz,i, sx,i, sy,i, sz,i,
Similarly for the coordinates y and z, where hx,i, zx,i,,
To ensure compliance with the terms of the normalization (95) subtract from the sets (98) the corresponding mean values, i.e. implement the conversion:
where with the three dots the corresponding transformation to the coordinates y and z are denote.
We will satisfy the first equation in (94). Taking into account the (97)-(99), we substitute the representation (96) in the part of the relative velocities in the first equation in (94), then get
For the solution of Equation (100) with respect to the three unknowns Ux, Uy, Uz will raffle off three random numbers xx, xy, xz, uniformly distributed on the interval [0, 1]. In this case, the solution of Equation (100) can be written as:
It remains to write down the equations for determining the unknown quantities Qx, Qy, Qz. Write out the second equation in (94) in connection with the notation of (96), then
In the system of Equation (102) everything is known except for the values of Qx, Qy, Qz. Solving Equation (102) relative to Qx, Qy, Qz, we find the set of locations of the particles after the interaction
Obtained after solving the system (102) the values Qx, Qy, Qz allows to find the set of positions of the particles after the interaction
where
To illustrate the algorithm discussed above (90)-(103) of three-dimensional modeling of motion within the space we build the flow in which a part of the ensemble of particles appears within the sphere inscribed in the original box. For modeling the desired flow we take as a prototype the example №10.
Example №12. To build the ensemble of the liquid particles, a part of particles of the ensemble were within the sphere inscribed in the original box cubic shape.
Recall that in the example №10 succeeded by adjusting the corresponding angles in the scheme of interaction of particles to build the flow, in which most of the particles concentrated in the left half of the box of square shape. For the three-dimensional case it is not fully possible to make due to the fact that the radius-vector of the velocity of the center of mass Vi,j = (vi + vj)/2 of the pair of particles does not lie in a plane perpendicular to the angular momentum of the pair, which took place in the two-dimensional case. However, some of the particles is possible to put into the desired area.
We project the velocity vector of the center of mass onto the plane perpendicular to the angular momentum K. For this we consider the previously constructed coordinate system with unit vectors
We introduce the notation:
As well as
where Ri,j = (Xi,j, Yi,j, Zi,j)―the radius vector of the center of mass of the pair of particles.
Knowing, according to (104), (105),
n-clusters with n > 2, by the algorithm (90)-(103). As the parameters of the experiment were selected the following values: N = 2500, T = 2500, t = 10?3, L = 1, Vmax = 5, r0 = 0,015, k = 0.03. In the left graph
On the right graph
particles, i.e. such particles the kinetic energy which is smaller than the amount
number of such particles was varied in the range from 4 in the initial time to 933 in one of the final moments of time. Exactly ultra cold particles are positioned distinctly within the sphere inscribed to the original region of localization of the ensemble.
In this paper, a mathematical model of motion of a perfect fluid is constructed. The model had a back story in the form of the mathematical model of turbulent fluid motion, built earlier by the author. Unlike previous models, this model considered only the discrete approach.
A perfect fluid is presented in the form of a finite ensemble of so-called liquid particles, which are extended geometric objects in the form of identical circles in two-dimensional and in the form of identical spheres in three-dimensional cases. When approaching such particles at a distance smaller of the two radii, they stick together. It is believed that the resulting intermediate particle instantly disintegrates, while ensuring the preservation of all the conservation laws: momentum, energy and angular momentum.
Considered in detail two-dimensional, the most studied case. Managed to build and explore using the computational experiment a number of interesting flows. Some regimes that provide separation of the particles in the ensemble were detected. Different kinds of vortex flows were simulated. Some interesting statistical properties of certain flows were detected. The flows, in which there is a concentration of kinetic energy in a small number of particles of the ensemble were detected. Discovered and studied also the flows, in which the kinetic energy of the particles is calibrated. In addition, it was possible using the appropriate flows of liquid particles of the ensemble to demonstrate the possibility to reproduce any prescribed image by manipulating the parameters of the interaction.
The model includes the development of mechanism of interaction, as in the binary level, acting as the main, and in the n-cluster, n > 2. Analysis of the mechanisms of interaction at the binary level and in the n-cluster revealed additional degrees of freedom, taking into account that led to the construction of a huge family of new, and in some cases unexpected flows. Note that the identification of additional degrees of freedom in the mechanism of interaction between the liquid particles allowed directly, i.e. “man-made” to control the stochastic, which, we believe, is typical for turbulent flows of continuous medium. In the future, it is expected that the management of the stochastics using the Monte Carlo method will allow to build an ensemble of specific flows and to calculate the average flow, as well as all other statistical characteristics in the framework of sampling statistics.
The ensemble of particles in three-dimensional space is considered in the final part of the model. Here, the results are more modest, although in general managed to develop suitable algorithms that describe both the binary level of interaction and the level of n-cluster.
The author is grateful for help in translating article V.N. Nikolenko.
Konstantin Eduardovich Plokhotnikov, (2016) About One Discrete Mathematical Model of Perfect Fluid. Open Journal of Modelling and Simulation,04,129-167. doi: 10.4236/ojmsi.2016.43012