1.1 Introduction
In this paper we consider a projectile motion in view of Riemann-Liouville fractional calculus. The projectile motion is one of the simplest problems whose analogs are ubiquitous in physics. The purpose of this paper is to extend the Caputo approach of [1] to the Riemann-Liouville case. We obtain some new formulas, in particular, the trajectory using the Riemann-Liouville fractional derivative is different. We compare both approaches and indicate new directions of research.
The fractional calculus is an extension of the ordinary calculus and has a history of over 300 years old. It represents a generalization of the ordinary differentiation and integration to arbitrary order and fractional calculus has applications in various fields, e.g. physics, engineering or biology [2–4]. Differential equations of fractional order have assumed a relevant role in the most diverse areas of science and engineering. Some physical considerations in favor of the use of fractional models are given in [5] and fractional mechanics is presented in [6, 7].
Many times the authors replace the usual integer derivative by another derivative of fractional order. However, from the physical point of view that is not totally correct [8] and some dimensional correction in the new equation is necessary; for example, substituting a first order derivative \(D^{1} :=\frac{d}{dt}\) by \(\frac{1}{{\sigma }^{1-\alpha}}D^{\alpha}\) where σ has an appropriate dimension [9].
For some new directions in fractional calculus and fractional differential equations we refer the reader for example to [10–13].
1.2 Definitions and preliminaries
We recall some definitions of fractional calculus.
The fractional integral of order
\(\alpha>0\) of a function \(f:[0,T]\rightarrow\mathbb{R}\) is defined by
$$I_{0^{+}}^{\alpha} f(t)=\int_{0}^{t} \frac{(t-s)^{\alpha-1}}{\Gamma (\alpha)}f(s)\,ds, $$
provided the right-hand side integral exists for almost every \(t\in [0,T]\). Here Γ is the classical gamma function. This fractional integral is well defined if, for example, \(f\in L^{1}(0,T)\).
Let \(\alpha>0\), \(n-1<\alpha<n\), \(n\in\mathbb{N}\), \(n\ge1\).
1.2.1 Caputo fractional derivative
Consider the space \(AC^{n}[0,T]\) of functions with absolutely continuous derivatives up to order \(n-1\) and with absolutely continuous n-derivative.
The Caputo fractional derivative of a function \(f\in AC^{n}[0,T]\), \(T>0\) is defined by [4]
$$ {}^{c}D^{\alpha}f(t)=\frac{d^{\alpha}f}{dt^{\alpha}}= \frac{1}{\Gamma (n-\alpha)}\int_{0}^{t}{(t- \tau)^{n-\alpha-1}}f^{(n)}(\tau)\, d\tau,\quad t\in [0,T]. $$
(1.1)
The Laplace transform of a function
\(f:[0,\infty)\rightarrow \mathbb{R}\) is the function \(F(s)\),
$$F(s)=\mathcal{L}\bigl\{ f(t)\bigr\} =\int_{0}^{\infty} e^{-st}f(t)\, dt, $$
provided it is well defined.
If we apply the Laplace transform to (1.1) we get [4]
$$ \mathcal{L} \bigl\{ {}^{c}D^{\alpha}f(t) \bigr\} =s^{\alpha}F(s)-\sum_{m=0}^{n-1}s^{\alpha-m-1}f^{(m)}(0). $$
(1.2)
1.2.2 Riemann-Liouville fractional derivative
The Riemann-Liouville fractional derivative of a function
f is defined as
$$ D^{\alpha}f(t)=\frac{d^{\alpha}f}{dt^{\alpha}}=\frac{1}{\Gamma(n-\alpha )} \biggl( \frac{d}{dt} \biggr)^{n}\int_{0}^{t}{(t- \tau)^{-\alpha+n-1}}f(\tau)\, d\tau, $$
(1.3)
provided the left-hand side is defined for almost every \(t>0\). If we apply the Laplace transform, we get
$$ \mathcal{L} \bigl\{ D^{\alpha}f(t) \bigr\} =s^{\alpha}F(s)- \sum_{m=0}^{n-1}s^{m}f^{(\alpha-m-1)}(0). $$
(1.4)
We recall that the Caputo derivative of a constant function is zero, i.e., \({}^{c}D^{\alpha} (1)=0\). However, for the Riemann-Liouville derivative:
$$D^{\alpha} 1=\frac{1}{\Gamma(1-\alpha)} t^{-\alpha}. $$
Also \(D^{\alpha}t^{\alpha-j}=0\) for \(j=1,2,\ldots,[\alpha]+1\).
A useful formula is the following relation:
$$D^{\alpha} t^{\gamma}=\frac{\Gamma(\gamma+1)}{\Gamma(\gamma+1-\alpha)} t^{\gamma-\alpha},\quad \alpha>0, \gamma-1>0, t>0, $$
and for \(\gamma=0\) we obtain the Riemann-Liouville derivative of a constant.
1.2.3 Mittag-Leffler function
A two-parameter function of the Mittag-Leffler type is defined by the series expansion [4]:
$$ E_{\alpha,\beta}(z)=\sum_{m=0}^{\infty} \frac{z^{m}}{\Gamma(\alpha m+\beta )}\quad (\alpha>0,\beta>0). $$
(1.5)
The Laplace transform for the Mittag-Leffler function is very useful in solving fractional differential equations:
$$ \int_{0}^{\infty} e^{-st}t^{\alpha m+\beta-1}E_{\alpha,\beta}^{(m)} \bigl(\pm at^{\alpha}\bigr)\, dt=\frac{m!s^{\alpha-\beta}}{(s^{\alpha}\pm a)^{m+1}}\quad \bigl( \operatorname{Re}(s)>a^{\frac{1}{\alpha}}\bigr). $$
(1.6)
Hence
$$ \mathcal{L}^{-1} \biggl[\frac{m!s^{\alpha-\beta}}{(s^{\alpha}\pm a)^{m+1}} \biggr]=t^{\alpha m+\beta-1}E_{\alpha,\beta}^{(m)}\bigl(\pm at^{\alpha} \bigr). $$
(1.7)
The general solution of the following simple fractional differential equation:
is given by
$$f(t)=c_{0}+c_{1} t+\cdots+c_{n-1} t^{n-1} $$
with \(c_{0},c_{1},\ldots,c_{n-1}\) arbitrary constants.
Thus for \(0<\alpha<1\), the general solution of \({}^{c}D^{\alpha}f=0\) is a constant.
However, for the Riemann-Liouville derivative we find that the general solution of
is given by
$$ f(t)=c_{0} t^{\alpha-n}+c_{1} t^{\alpha-n+1}+\cdots+c_{n-1} t^{\alpha-1}. $$
(1.8)
Thus for \(0<\alpha<1\), the general solution of
$$D^{\alpha}f=0\mbox{ is }f(t)=c t^{\alpha-1},\quad c\in\mathbb{R}. $$
This is a crucial difference since \(t^{\alpha-1}\) has a singularity at \(t=0^{+}\).
1.3 Classical problem formulation of projectile motion
A familiar basic physics problem involves the determination of the motion of an object which is projected into a spatial medium and subject to a uniform gravitational field. In this section, we consider the introductory version of this problem in which the medium usually does not offer resistance to the projectile motion. The projectile is treated as a particle of mass m under an uniform gravitational force and no drag force is considered. Under these conditions, the classical equations of motion for the particle, in the x-y plane, are given by
$$ m\frac{d^{2}x}{dt^{2}}=0,\qquad m\frac{d^{2}y}{dt^{2}}=-mg, $$
(1.9)
by the classic Newton law.
The corresponding initial conditions are
$$ \begin{aligned} &x(0)=0,\qquad \dot{x}(0)=v_{0}\cos\phi, \\ &y(0)=0,\qquad \dot{y}(0)=v_{0}\sin\phi; \end{aligned} $$
(1.10)
namely, the projectile starts from rest, with an initial force of module \(v_{0}\) and an angle ϕ.
The trajectory is given by a parabola:
$$y=\tan\phi x-\frac{g}{2v_{0}^{2}}\sec^{2}\phi x^{2}. $$
The range is the horizontal distance traveled by the projectile from the time it is fired until it lands. The maximum altitude is the height of the highest point in the trajectory. The time of flight is the amount of time the projectile spends in the air between when it is fired and when it lands.
The range is
$$\frac{2v_{0}^{2}}{g}\sin\phi\cos\phi, $$
and the corresponding flight time is
$$\frac{2v_{0}\sin\phi}{g}. $$
The range is maximum for \(\phi=\frac{\pi}{4}\). Finally the maximum height is equal to
$$\frac{v_{0}^{2}\sin^{2}\phi}{2g}. $$
1.4 Caputo fractional problem formulation
Now consider the above problem in view of the fractional calculus. Before doing so, we consider the question of how to formulate the acceleration of a particle in the fractional approach. It seems reasonable to consider instead of the second derivative a fractional order derivative [1]. The fractional differential equations for the projectile problem is then
$$ \frac{d^{\alpha}x}{dt^{\alpha}}=0,\qquad \frac{d^{\alpha}y}{dt^{\alpha}}=-g, $$
(1.11)
where \(1<\alpha\le2\).
For \(\alpha=2\) we recover, of course, the classical case (1.9).
Physically, we can interpret the fractional derivatives of x and y, respectively, as the accelerations of the projectile in the horizontal and vertical directions, which reduce to the acceleration of the classical mechanics at \(\alpha\to2^{-}\).
First we recall the solutions using the Caputo derivative [1]. The general solution of (1.11) using the Caputo derivative is
$$x(t)=c_{1}+c_{2} t, \qquad y(t)=-g \frac{t^{\alpha}}{\Gamma(\alpha+1)}+d_{1}+d_{2} t. $$
Implementing the initial conditions (1.10) we get
$$ x(t)=v_{0} \cos\phi t, \qquad y(t)=-g \frac{t^{\alpha}}{\Gamma(\alpha+1)}+v_{0}\sin \phi t. $$
(1.12)
For the Caputo derivative the solution of (1.11) can also be obtained by means of Laplace transform (1.2) as in [1]:
$$ \begin{aligned} &s^{\alpha}X(s)-s^{\alpha-1}x(0)-s^{\alpha-2} \dot{x}(0)=0, \\ &s^{\alpha}Y(s)-s^{\alpha-1}y(0)-s^{\alpha-2}\dot{y}(0)=- \frac{g}{s}. \end{aligned} $$
(1.13)
Using the initial conditions (1.10), we get
$$ X(s)=\frac{v_{0}\cos\phi}{s^{2}},\qquad Y(s)=-\frac{g}{s^{\alpha+1}}+ \frac{v_{0}\sin \phi}{s^{2}}, $$
(1.14)
and then x, y are given by (1.12).
1.5 Riemann-Liouville fractional problem formulation
For the Riemann-Liouville derivative the corresponding and adequate initial conditions are [2]
$$ \begin{aligned} & D^{\alpha-2}x(0)=0,\qquad D^{\alpha-1}x(0)=v_{0} \cos\phi, \\ &D^{\alpha-2}y(0)=0,\qquad D^{\alpha-1}y(0)=v_{0}\sin\phi, \end{aligned} $$
(1.15)
which, of course, coincide with the initial conditions (1.10) for \(\alpha=2\).
The general solution of (1.11) for the Riemann-Liouville derivative is given by
$$\begin{aligned}& x(t)=c_{0} t^{\alpha-2}+c_{1} t^{\alpha-1}, \\& y(t)=d_{0} t^{\alpha-2}+d_{1} t^{\alpha-1}- \frac{g}{\Gamma(\alpha+1)} t^{\alpha}, \end{aligned}$$
in view of (1.8).
However,
$$D^{\alpha-1}x(t)=c_{0} D^{\alpha-1}\bigl(t^{\alpha-2} \bigr)+c_{1} D^{\alpha -1}\bigl(t^{\alpha-1}\bigr). $$
We have
$$D^{\alpha-1}\bigl(t^{\alpha-2}\bigr)=0 $$
and
$$D^{\alpha-1}\bigl(t^{\alpha-1}\bigr)=\Gamma(\alpha). $$
Therefore \(D^{\alpha-1}x(t)=c_{1} \Gamma(\alpha)\).
Using the initial condition \(D^{\alpha-1}x(0)=v_{0}\cos\phi\), we can conclude that \(c_{1}=\frac{v_{0}\cos\phi}{\Gamma(\alpha)}\), so that
$$x(t)=c_{0} t^{\alpha-2}+\frac{v_{0}\cos\phi}{\Gamma(\alpha)} t^{\alpha-1}. $$
Using the other initial condition we obtain with \(c_{0}\),
$$D^{\alpha-2}x(t)=c_{0} \Gamma(\alpha-1)+v_{0}\cos\phi t, $$
since \(D^{\alpha-2}(t^{\alpha-2})=\Gamma(\alpha-1)\) and \(D^{\alpha -2}(t^{\alpha-1})=\Gamma(\alpha) t\).
Therefore \(D^{\alpha-2}x(0)=0\) implies \(c_{0}=0\) and we get
$$ x(t)=\frac{v_{0}\cos\phi}{\Gamma(\alpha)} t^{\alpha-1}. $$
(1.16)
In the same way
$$\begin{aligned}& D^{\alpha-1}y(t)=-g t+d_{1} \Gamma(\alpha), \\& D^{\alpha-2}y(t)=\frac{-1}{2} g t^{2}+d_{0} \Gamma(\alpha-1)+v_{0}\sin\phi t, \end{aligned}$$
where we obtain \(d_{1}=\frac{v_{0}\sin\phi}{\Gamma(\alpha)}\), \(d_{0}=0\), i.e.,
$$ y(t)=\frac{-g}{\Gamma(\alpha+1)} t^{\alpha}+\frac{v_{0}\sin\phi}{\Gamma (\alpha)} t^{\alpha-1}. $$
(1.17)
We point out that the solution \(x(t)\), \(y(t)\) given by (1.16), (1.17) is qualitatively different from the solution (1.12). With this Riemann-Liouville approach we get new fractional trajectories.
For the Riemann-Liouville derivative, the solution of (1.11) can also be reached by means of the Laplace transform (1.4) as follows:
$$ \begin{aligned} &s^{\alpha}X(s)-D^{\alpha-1}x(0)-sD^{\alpha-2} x(0)=0, \\ &s^{\alpha}Y(s)-D^{\alpha-1}y(0)-sD^{\alpha-2} y(0)=- \frac{g}{s}. \end{aligned} $$
(1.18)
Using the initial conditions (1.15), we get
$$ X(s)=\frac{v_{0}\cos\phi}{s^{\alpha}},\qquad Y(s)=-\frac{g}{s^{\alpha+1}}+ \frac {v_{0}\sin\phi}{s^{\alpha}}. $$
(1.19)
From this it follows that \(x(t)\), \(y(t)\) are given by (1.16), (1.17), respectively.
1.6 Features of projectile motion in the fractional calculus
As we have recalled before, three quantities are particularly relevant for identifying, distinguishing, and analyzing trajectories in our setting: the range, the maximum altitude, and the time of flight.
We recall each of these quantities using the Caputo derivative [1] and compute them for the Riemann-Liouville derivative.
1.6.1 Trajectory
Caputo: By eliminating t from (1.12), we obtain the trajectory of the fractional projectile for arbitrary α as:
$$ y=\tan\phi x-\frac{g x^{\alpha}}{\Gamma(\alpha+1)(v_{0}\cos\phi)^{\alpha}}. $$
(1.20)
This was obtained in [1].
As \(\alpha\to2^{-}\), (1.20) gives the classical trajectory equation:
$$ y=\tan\phi x-\frac{g}{2v_{0}^{2}}\sec^{2}\phi x^{2}. $$
(1.21)
Riemann-Liouville: By eliminating t from (1.16), (1.17) we now obtain the trajectory of the fractional projectile from (1.16), (1.17):
$$ y=\tan\phi x-\frac{g}{\Gamma(\alpha+1)} \biggl(\frac{x \Gamma(\alpha )}{v_{0}\cos\phi} \biggr)^{\frac{\alpha}{\alpha-1}}. $$
(1.22)
As a new result, as \(\alpha\to2^{-}\), from (1.22) we obtain the classical trajectory equation (1.21).
Observe that (1.22) is different from (1.20).
1.6.2 Range
The fractional projectile range is defined as the value of x at the impact point.
Caputo [1]: Thus, \(y=0\) at \(x=R_{\mathrm{F}}\). Hence, \(R_{\mathrm{F}}\) is given as
$$ R_{\mathrm{F}}=(v_{0})^{\frac{\alpha}{\alpha-1}} \biggl[ \frac{\Gamma(\alpha+1)}{g} \biggr]^{\frac{1}{\alpha-1}}(\sin\phi)^{\frac{1}{\alpha-1}}\cos\phi. $$
(1.23)
Also as \(\alpha\to2^{-}\), (1.23) leads to the range of the classical projectile:
$$ R_{\mathrm{C}}=\frac{2 v_{0}^{2}}{g}\sin\phi\cos\phi. $$
(1.24)
See Figure 1 for \(v_{0}=2\).
Riemann-Liouville: Doing again \(y=0\) at \(x=R_{\mathrm{F}}\), we obtain in this case for \(R_{\mathrm{F}}\),
$$ R_{\mathrm{F}}=\frac{v_{0}\cos\phi}{\Gamma(\alpha)} \biggl(\frac{\alpha v_{0}\sin\phi }{g} \biggr)^{\alpha-1}, $$
(1.25)
which is different and, surprisingly, simpler than (1.23). See Figure 2.
As \(\alpha\to2^{-}\), from (1.25) we obtain the range of the classical projectile (1.24).
We compare both ranges in Figure 3.
1.6.3 Flight time
The fractional time of flight
\(t_{\mathrm{F}\text{-flight}}\) is defined as the value of t at which the projectile hits the ground.
Caputo: Thus, \(y=0\) at \(t=t_{\mathrm{F}\text{-flight}}\), hence
$$ t_{\mathrm{F}\text{-flight}}= \biggl[\frac{\Gamma(\alpha+1) v_{0} \sin\phi}{g} \biggr]^{\frac {1}{\alpha-1}}. $$
(1.26)
Here, it should be noted also that the classical flight time \(t_{\mathrm{C}\text{-flight}}\) can be deduced from (1.26) at \(\alpha\to2^{-}\) as
$$ t_{\mathrm{C}\text{-flight}}=\frac{2 v_{0}\sin\phi}{g}. $$
(1.27)
Riemann-Liouville: Doing \(y=0\) at \(t=t_{\mathrm{F}\text{-flight}}\), we obtain
$$ t_{\mathrm{F}\text{-flight}}=\frac{\alpha v_{0}\sin\phi}{g}. $$
(1.28)
This has a simpler expression than the Caputo analogous (1.26).
Again, it is verified that the classical flight time \(t_{\mathrm{C}\text{-flight}}\) can be obtained from (1.28) at \(\alpha\to2^{-}\) to get (1.27).
1.6.4 Maximum height
Caputo: The projectile reaches its maximum height when its vertical component vanishes, i.e., \(\dot{y}=0\). By solving this equation for t, we get
$$ t_{\text{height-}\mathrm{max}}= \biggl[\frac{\Gamma(\alpha) v_{0}\sin\phi}{g} \biggr]^{\frac {1}{\alpha-1}}. $$
(1.29)
Substituting (1.29) into \(y(t)\) in (1.12), we obtain the Caputo fractional maximum height \(H_{\mathrm{F}}\):
$$ H_{\mathrm{F}}= \biggl(1-\frac{1}{\alpha} \biggr) \biggl[ \frac{\Gamma(\alpha)}{g} \biggr]^{\frac{1}{\alpha-1}}(v_{0} \sin \phi)^{\frac{\alpha}{\alpha-1}}. $$
(1.30)
Here, we can also obtain the maximum height of the classical projectile when \(\alpha\to2^{-}\):
$$ H_{\mathrm{C}}=\frac{v_{0}^{2} \sin^{2}\phi}{2g}. $$
(1.31)
Riemann-Liouville: Solving for t the equation \(\dot{y}=0\), we obtain
$$ t_{\text{height-}\mathrm{max}}=\frac{(\alpha-1)v_{0}\sin\phi}{g}. $$
(1.32)
Substituting (1.32) into \(y(t)\) in (1.17), we obtain the Riemann-Liouville fractional maximum height \(H_{\mathrm{F}}\):
$$ H_{\mathrm{F}}=\frac{-((\alpha-1) v_{0}\sin\phi)^{\alpha}}{\Gamma(\alpha+1)g^{\alpha -1}}+\frac{(v_{0}\sin\phi)^{\alpha}}{\Gamma(\alpha)} \biggl( \frac{\alpha -1}{g} \biggr)^{\alpha-1}. $$
(1.33)
If \(\alpha\to2^{-}\), we obtain again (1.31). We have
$$ H_{\mathrm{C}}=\frac{v_{0}^{2}\sin^{2}\phi}{2g}. $$
(1.34)