In a previous note (1), we argued that maximization of Shannon’s entropy -P(x)ln(P(x)) subject to the periodic (complex) constraint i p x P(x) yields the free particle two-dimensional probability exp(ipx). Similarly one may obtain exp(-iEt) (with a minus associated with E to allow for -Et+px to remain invariant under t->t+constant/E and x->x+constant/p), also for a free particle. We argued that multiplying the two together exp(-iEt+ipx) yields a phase which should be invariant when seen from frames moving at constant average velocity v=x/t.. This means that not only -Et+px, but -EE+pp= momo (c=1) is a Lorentz invariant. The latter, however, represents conservation of energy for no potential. We argue that conservation of energy for a free particle is linked to both the maximization of Shannon’s entropy for both E and p, i.e. it is an equilibrium in both time and space. We have argued before that in the presence of a potential V(x), one may replace E with E-V(x) in the Klein-Gordon equation. Here we wish to examine the statistical features of this situation in more detail. Finally, we note that the third statistical feature is present, associated with the absorption or emission of a photon. The photon wavefunction exp(-iEt+ipx) is subject to both E and p maximization of Shannon’s entropy and it is associated with an equilibrium E1 or E2 energy level of say an electron in an atom. The difference is that this absorption or emission seems to occur stochastically without being directly linked to the momentum impulse hits between the electron and V(x). In describing photon emission/absorption, one may maximize Shannon’s entropy using only conservation of energy, i..e. use the Maxwell-Boltzmann factor. The x and t notion of maximum entropy are still present as the photon wavefunction is exp(-iEt+px). The photon momentum, however, is conserved with the total momentum of the atom and not the impulse hits of the electron with V(x). The two stochastic features occur together.
In a previous note (1), we argued that maximization of Shannon’s entropy -P(x)ln(P(x)) subject to the periodic (complex) constraint i p x P(x) yields the free particle two-dimensional probability exp(ipx). Similarly one may obtain exp(-iEt) (with a minus associated with E to allow for -Et+px to remain invariant under t->t+constant/E and x->x+constant/p), also for a free particle. We argued that multiplying the two together exp(-iEt+ipx) yields a phase which should be invariant when seen from frames moving at constant average velocity v=x/t.. This means that not only -Et+px, but -EE+pp= momo (c=1) is a Lorentz invariant. The latter, however, represents conservation of energy for no potential. We argue that conservation of energy for a free particle is linked to both the maximization of Shannon’s entropy for both E and p, i.e. it is an equilibrium in both time and space. We have argued before that in the presence of a potential V(x), one may replace E with E-V(x) in the Klein-Gordon equation. Here we wish to examine the statistical features of this situation in more detail. Finally, we note that the third statistical feature is present, associated with the absorption or emission of a photon. The photon wavefunction exp(-iEt+ipx) is subject to both E and p maximization of Shannon’s entropy and it is associated with an equilibrium E1 or E2 energy level of say an electron in an atom. The difference is that this absorption or emission seems to occur stochastically without being directly linked to the momentum impulse hits between the electron and V(x). In describing photon emission/absorption, one may maximize Shannon’s entropy using only conservation of energy, i..e. use the Maxwell-Boltzmann factor. The x and t notion of maximum entropy are still present as the photon wavefunction is exp(-iEt+px). The photon momentum, however, is conserved with the total momentum of the atom and not the impulse hits of the electron with V(x). The two stochastic features occur together.
The Klein-Gordon equation is often created by taking the deterministic relativistic invariant EE = pp+ momo (c=1), replacing E with E-V(x), E=id/dt partial and p with -id/dx partial and having the derivatives act on a function W(x,t). This seems to suggest that the starting point is deterministic classical mechanics (special relativistic form) E,p equation with stochasticity somehow introduced by having average E and p follow from the action of the operators id/dt partial and -id/dx partial on a stochastic function W(x,t). It also makes it somewhat difficult to understand, from first principles, why one would create a linear equation in E,p (compatible with EE=pp+momo), i.e. the Dirac equation. We suggest that the entire approach to Klein-Gordon equation is stochastic, with the EE=pp+momo equation appearing as an invariant expression of average values which describes motion as seen from different frames moving with constant average velocity v. In such a case, one is not limited to the form EE=pp+momo and there is nothing unusual about trying to create an equation linear in E and p. In particular, we argue that the free particle form exp(-iEt) exp(ipx) follows from two separate maximizations of Shannon’s entropy subject to complex constraints (for periodicity). As a result, free motion with a constant E and p is already stochastic as seen from a photon moving in a region with an index of refraction n or a particle moving in a region with constant potential. Changing to a potential V(x) leads to the notion of an average kinetic energy because a constant p is associated with a distribution in x i.e. exp(ipx). One may think in terms of these average values in order to consider frames moving with constant v. This leads to both -Et+px -EE+pp=momo being invariants. If a V(x) is present, one may replace E with E-V(x), but then needs a statistical average Sum over p a(p)exp(ipx) to describe pp.. The time portion may still be exp(-iEt) if one has a bound system characterized by a single average energy. This yields the Klein-Gordon equation.There is, however, no particular reason why one could not use an equation linear in E and p (consistent with -EE+pp=momo), because the latter follows form the an average v (deterministic view) of the stochastic exp(-iEt+ipx) two-dimensional probability.
In previous notes (1), we argued that stochasticity (in x and t for a fixed E energy and p momentum) may occur at a single particle level based on the Lorentz invariant A = -Et+px ((1)). This invariant is also the relativistic/nonrelativistic classical action for v=x/t. The argument made in (1) was that a fluctuation in t of constant/E could cancel one in x of constant/p. We further argued one could create eigenvalue equations using id/dt and -id/dx and obtain the two dimensional probabilities exp(-iEt) and exp(ipx), i.e obtain a description of free particle quantum mechanics. Experimentally, both exp(-iEt) and exp(ipx) have been verified through interference patterns. Theoretically, however, ((1)) was known about 1905 and no single particle stochasticity was associated with it. In this note, we try to argue why there should be an associated stochasticity associated with ((1)) based on motion through a constant potential for a particle with rest mass and motion through a medium with an index of refraction for a photon. We in fact argue that one cannot solve certain photon problems such as one dimensional reflection/refraction without explicitly considering the stochasticity which is usually left out when describing p2=pn2 and c2=c/n2 where n2 is the index of refraction in the second medium and p and c are values in the first medium with n1=1. We further consider exp(ipx) and exp(-iEt) as arising from the maximization of Shannon’s entropy subject to certain constraints. The product of these two two-dimensional periodic probabilities is exp(-iEt+ipx) which we argue must hold as seen from observers moving with constant velocity. Thus we argue that the Lorentz invariant -Et+px follows from statistical considerations alone. Given this expression (a subset of the whole picture), one may try to create a deterministic viewpoint by setting x/t=v=constant. This allows one to consider transformations which define E and p in terms of mo and v and yields the invariant EE = pp+momo (c=1). Thus we argue that the conservation of energy equation ultimately follows from a deterministic view of the argument of a statistical probability. In other words, it is only a subset of the full picture i.e the quantum mechanical statistical one.
A quantum particle (or photon) is described by the two dimensional probabilistic exp(-iEt+ipx), with t and x being treated as independent. exp(-iEt) suggests a stochasticity in time associated with energy E, but none in space. Similarly, there is a stochasticity in x associated with p but none in time. For -Et+px to be invariant as seen in different frames moving with constant average velocities, -EE+pp must also be an invariant i.e. equal momo (c=1). This leads to a contradiction. exp(ipx) suggests a stochasticity in x for a given p, but -EE+pp=momom converts a given p into an E, so there exists stochasticity in x for E as well (which contracts the notion that exp(-iEt) is only stochastic in time). The link between E and p, however, which breaks x-t independence is the notion of average velocity v=x/t, which is physical (as is exp(ipx)). How does one deal with this contradiction? We suggest that p and E are associated with different physical interaction processes. P is related to impulse hits, while E may be associated with loss of energy over a distance if a potential V(x) exists. Even though exp(ipx) suggests stochasticity in x, at the same time one considers fixed objects (targets, slits) and even potentials V(x) at well-defined deterministic x points. This seems to mean that if one wishes to make energy deterministic at each x, one must create an average of exp(ipx)s. We argue that there seem to be two important ideas associated with exp(ipx). First, it is a two-dimensional probability linked to a flux because it is due to p which is due to a v. Probability flux does not get created or destroyed (unless there are sinks or sources). This leads to issues when dealing with a bound state which should be localized. A flux probability is not localized, nor is an average in general (except in the case of an infinite potential). A bound state problem is really a flux problem consisting of a particle trapped in a potential together with it tunneling out or in i.e. one has a dynamic equilibrium picture somewhat like evaporation in equilibrium. (To see the clear difference between a state based probability and a flux-based one, consider the problem of a photon moving in the x direction and reflecting/refracting from an n1-n2 (index of refraction) y-plane at x=0. State probabilities (P(reflect), P(refract)) are not sufficient to solve the problem, but a flux-based one exp(ipx) which allows a single particle to appear as if in a steady stream scenario is.) Secondly, unlike evaporation, the guiding statistical condition for exp(ipx) is that its average makes E deterministic in x space. Thus E is like T (temperature) in that it is a statistical parameter, but it is not based on maximization of entropy (equilibration), but rather in creating E as an average which is then treated as deterministic (ignoring in a sense exp(-iEt) value characterizing an equilibrium system (not based on equilibration). This deterministic E system is linked with discrete values (due to the exp(ipx) equilibrium), but given that E is treated as deterministic, a stochastic process based on maximization of entropy is introduced to describe the absorption and emission of photons (which allow for jumps from one average-deterministic En to another). We argue that the flux-like probability exp(ipx) exists in order to deal with sharp interfaces (e.g. n1-n2 index of refraction) at a deterministic x=0 which probabilistically leads to different p values. It uses the reverse approach i.e. a deterministic p and a stochastic x. In such a way, one may use continuity (and first derivative continuity) at the sharp x with flux-probabilities containing the various allowed p values (due to the impulses). The continuity follows from no sinks or sources existing for the flux-like probability.
Classical statistical equilibrium, for a Maxwell-Boltzmann gas for example, is based on localizing the gas to a certain volume and having no average changes in time. In other words, maximization of entropy subject to the average energy constraint sum over i ei P(ei) is linked with these two ideas. Ei (kinetic energy for no V(x)) is a conserved quantity in reactions and overall energy represents a quantity which does not change in time consistent with a center-of-mass which does not change and average values which do not change in time. This approach is associated with a specific kind of statistics, namely that of the constituents of the system. As a result, one obtains the Maxwell-Boltzmann distribution exp(-ei/T) where ei is an extensive property of a particle (or subset of the system) and T an intensive property linked with the whole system. Entire systems, however, may interact. For example, a N particle system with T1 may achieve equilibrium with a N system with T2 yielding a new average (T1+T2)/2. The probabilities exp(-ei/T), however, are for subcomponents of a system. One could argue that momentum is an additive (vector) quantity which is conserved in collisions, but the average momentum of the system is 0 which does not allow one to describe increases in average p and -p which may be described using energy, i.e. through T temperature. This begs the question whether momentum may be associated with maximization of entropy subject to a constraint. In (1), we argued that there were flow probabilities in quantum mechanics linked with exp(-iEt) and exp(ipx), which may be obtained formally through maximization of entropy. Here we argue that while an MB gas describes a localized volume with average values not changing in time, x and t state values change, even if p and pp/2m remain constant for a free particle and it may be possible to have a maximum entropy situation to describe the combination of px and -Et. In other words, we suggest localized versus nonlocalized equilibria. The latter represents a quantum free particle, but one may then try to localize such a particle in a potential. What happens in such a case? One does not switch to a localized maximization of entropy as in the MB gas case, because it is possible to have the potential leave the flow unbroken as argued in (1). In other words, the nonlocalized free particle equilibrium is akin to a particle bouncing back and forth in a box, which in classical mechanics is not considered a statistical process. Here we suggest that the underlying picture is statistical meaning that the classical mechanics approach seems to be an approximation. We consider both a quantum particle trapped in an infinite potential well and the ground state of an oscillator to examine ideas of maximizing entropy beyond exp(ipx). A key common feature of both types of equilibrium is the product rule P(ei)P(ej) = P(ei+ej) or exp(ip1x)exp(ip2x) = exp(i (p1+p2)x). We consider this feature as well.
Classical statistical mechanics often utilizes the idea of maximizing entropy subject to a physical constraint For example, for a Maxwell-Boltzmann gas, this constraint is Sum over i ei p(ei) where ei=pp/2m (no V(x)). We have argued previously that there may be many possible constraints, i.e. an average of any power of ei could be a physical constraint. We stated that a conserved quantity associated with interactions present (creating equilibrium) should be used. Here we examine the constraint in terms of mechanical resistance i.e. pressure. We further consider the free particle quantum wavefunction exp(-iEt+ipx) in terms of resistances in time and space. We argue that the form exp(-ipx)exp(ipx) = 1 removes the stochasticity associated with motion due to the resistance (impulse) p Thus 1 or 1/L (L=length) seems to be a smoothed over picture associated with a constant flux and overlooks the inherent stochasticity associated with motion. This stochasticity is crucial when describing impulse hit reactions (n-slits, index of refraction interfaces, interactions with V(x)), but if a particle does not interact in space, then the W*(x)W(x) picture seems to be used to describe the v(average)=x/t scenario. The issue is that for a quantum bound state, there is also an rms velocity based on a wavefunction and one may create an average W(x)W(x) (W(x) is real for a bound state), but like exp(-ipx)exp(ipx), this bypasses some of the stochasticity in the problem, we argue. It only demonstrates stochasticity associated with differences in momentum.
Quantum free particles are represented by exp(ipx) which we argue is a dynamical probability which maps into a static probability exp(-ipx)exp(ipx) i.e. the usual spatial density. Shannon’s entropy = - Sum over i P(i) ln(P(i)) is often associated with the static probability P(x)=W*(x)W(x) (1), but in (2) we argue that one may introduce a different probability C1 Real(W(x)). In this note we consider both in different examples. In particular we argue that the two different probability definitions are ultimately linked with momentum conservation. The form W*(x)W(x) introduces local nonconstant spatial densities if there are mixed exp(ip1x) and exp(ip2x)’s within W(x). If there are not, this form W*(x)W(x) does not distinguish space i.e. exp(ipx)exp(ipx+ib) is still a constant. Nevertheless interference may still occur which may be linked with entropy. Thus we first examine the case of exp(ipx) + exp(ipx + ib) where b is a constant phase shift. W*(x)W(x)=constant which does not lead to a structure different from exp(-ipx)exp(ipx), yet W(x)=C2{exp(ipx) + exp(ipx+ib)} interfere. On the other hand |cos(px)| and | cos(px)(1+cos(b)) - sin(px)sin(b) | are not the same distributions (one is a parity eigenstate, the other is not) and so we argue there is a different local entropy density. Thus introducing extra information i.e. b changes local density and entropy density if one adjusts the definition of probability used. We next consider W(x)= exp(ip1x)+exp(ip2x). One may also use the probability C3 Real|W(x)|, but here we are interested in an association with conservation of momentum. For a given exp(ipx), exp(-ipx)exp(ipx)=1. We argue that this represents a steady state picture i.e. a kind of equilibrium in which all x points are treated the same. W*(x)W(x) breaks this equilibrium scenario by introducing a term 2 cos( (p1-p2)x ). Thus both local spatial density and local entropy (based on W*W) are different and do not represent the “equilibrium” scenario. We further argue that momentum conservation may be linked with this equilibrium type of density and entropy density at least in the case of the transition amplitude <p1 V(x) p2>. V(x)exp(ipx) yields a sum of exp(ipx)’s, but only exp(ip1x) survives the integral over dx. The others create spatial densities which are not constant and so represent local entropy densities which do not treat all x points the same i.e .is not an equilibrium one. Thus we try to use Shannon’s entropy with two different probability definitions to analyse different quantum interference scenarios and argue that momentum conservation (and possibly others) may be linked with local entropy density in some situations.