文章信息
- 张毅, 周燕
- ZHANG Yi, ZHOU Yan
- 基于Riesz导数的分数阶Birkhoff系统的Noether对称性与守恒量
- Noether Symmetry and Conserved Quantity for Fractional Birkhoffian Systems in Terms of Riesz Derivatives
- 北京大学学报(自然科学版), 2016, 52(4): 658-668
- Acta Scientiarum Naturalium Universitatis Pekinensis, 2016, 52(4): 658-668
-
文章历史
- 收稿日期: 2015-10-07
- 修回日期: 2016-02-10
- 网络出版日期: 2016-07-14
2. 苏州市工业园区娄葑学校, 苏州 215021
2. Suzhou Industrial Park Loufeng School, Suzhou 215021
动力学系统对称性的研究一直是分析力学的重要发展方向。1918年, Noether[1]研究了Hamilton作用量在无限小变换下的不变性质, 揭示了力学系统的守恒量与其内在的动力学对称性之间的关系。Djukić等[2]将Noether定理推广到完整非保守系统, 李子平[3]、Bahar等[4]和Liu[5]进一步将Noether定理推广到非完整非保守系统。梅凤翔[6]用Pfaff作用量代替Hamilton作用量, 通过研究Pfaff作用量在无限小变换的广义准对称性, 建立了Birkhoff系统的Noether理论。近年来, 对Noether对称性的研究取得一些重要成果[6-10]。
分数阶微积分的概念最早出现在L’Hospital于1695年写给Leibniz的信中, 但是直到1974年, 第一本关于分数阶微积分理论的著作[11]才问世。近20余年来, 随着分数阶微积分应用领域的不断拓展, 分数阶微积分及其应用研究有了很大的发展。1996年, Riewe[12-13]首次将分数阶微积分应用于非保守系统动力学建模, 提出并初步研究了分数阶变分问题。之后, Agrawal[14-15]、Baleanu等[16-17]、Atanacković等[18-19]和El-Nabulsi等[20-22]对分数阶变分问题进行了深入研究。Frederico等[23-26]最早开展分数阶Noether对称性与守恒量的研究, 基于Riemann-Liouville分数阶导数定义[23-24]、Caputo分数阶导数定义[25]以及Riesz-Caputo分数阶导数定义[26], 分别考虑时间不变和时间变化的无限小变换作用, 得到分数阶Noether定理。此外, Frederico等[27-28]基于El-Nabulsi动力学模型研究了类分数阶作用变分的不变性问题。近年来, 约束力学系统基于分数阶模型的Noether对称性与守恒量的研究已经取得一些重要成果[29-35]。但是, 这些研究主要限于分数阶Lagrange系统和分数阶Hamilton系统。
本文基于Riesz分数阶导数的定义, 研究分数阶Birkhoff系统的分数阶Noether对称性。从分数阶Pfaff作用量在无限小变换下的不变性出发, 分别在时间不变和时间变化的无限小变换下, 研究分数阶Pfaff作用量的不变性, 建立分数阶Birkhoff系统的Noether定理。
1 分数阶导数本节列出研究中涉及的Riemann-Liouville分数阶导数、Caputo分数阶导数和Riesz分数阶导数的定义, 以及Riesz分数阶导数下的分部积分公式。具体的证明和讨论可参见文献[36-37]。
Riemann-Liouville分数阶左导数定义为
$ {}_{{t_1}}D_t^\alpha f(t) = \frac{1}{{\Gamma (m-\alpha )}}{\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)^m}\int_{{t_1}}^t {\frac{{f(\tau )}}{{{{(t-\tau )}^{\alpha-m + {\rm{1}}}}}}{\rm{d}}\tau } 。 $ | (1) |
Riemann-Liouville分数阶右导数为
$ {}_tD_{{t_2}}^\alpha f(t) = \frac{1}{{\Gamma (m-\alpha )}}{\left( {-\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)^m}\int_t^{{t_2}} {\frac{{f(\tau )}}{{{{(\tau-t)}^{\alpha - m + {\rm{1}}}}}}{\rm{d}}\tau } 。 $ | (2) |
Caputo分数阶左导数定义为
$ _{{t_1}}^{\rm{C}}D_t^\alpha f(t) = \frac{1}{{\Gamma (m-\alpha )}}\int_{{t_1}}^t {\frac{{{f^{(m)}}(\tau )}}{{{{(t-\tau )}^{\alpha-m + {\rm{1}}}}}}{\rm{d}}\tau } 。 $ | (3) |
Caputo分数阶右导数为
$ _t^{\rm{C}}D_{{t_{\rm{2}}}}^\alpha f(t) = \frac{{{{(-1)}^m}}}{{\Gamma (m-\alpha )}}\int_t^{{t_2}} {\frac{{{f^{(m)}}(\tau )}}{{{{(\tau-t)}^{\alpha - m + {\rm{1}}}}}}{\rm{d}}\tau }, $ | (4) |
其中Γ (*)是Euler Gamma函数, α是导数的阶, 且m-1≤α < m, m为正整数。如果α是整数, 上述分数阶导数成为整数阶导数, 有
$ \left\{ \begin{array}{l} {}_{{t_1}}D_t^\alpha f(t){\kern 1pt} {\kern 1pt} = {}_{{t_1}}^CD_t^\alpha f(t){\kern 1pt} {\kern 1pt} = {\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)^\alpha }f(t), \\ {}_tD_{{t_2}}^\alpha f(t){\kern 1pt} = {}_t^CD_{{t_2}}^\alpha f(t){\kern 1pt} {\kern 1pt} = {\left( {-\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)^\alpha }f(t)\; \end{array} \right. $ | (5) |
Riesz-Riemann-Liouville分数阶导数定义为
$ _{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha f(t) = \frac{1}{{2\Gamma (m-\alpha )}}{\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)^m}\int_{{t_1}}^{{t_2}} {\frac{{f(\tau )}}{{|t-\tau {|^{\alpha-m + {\rm{1}}}}}}{\rm{d}}\tau }, $ | (6) |
Riesz-Caputo分数阶导数定义为
$ {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha f(t) = \frac{1}{{{\rm{2}}\Gamma (m-\alpha )}}\int_{{t_1}}^{{t_2}} {\frac{{{f^{(m)}}(\tau )}}{{|t-\tau {|^{\alpha-m + {\rm{1}}}}}}{\rm{d}}\tau } 。 $ | (7) |
由上述定义可知, Riesz-Riemann-Liouville分数阶导数与Riemann-Liouville分数阶导数之间的关系为
$ {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha f(t) = \frac{1}{2}\left[{{}_{{t_1}}D_t^\alpha f(t) + {{(-1)}^m}{}_tD_{{t_2}}^\alpha f(t)} \right]; $ | (8) |
Riesz-Caputo分数阶导数与Caputo分数阶导数之间的关系为
$ {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha f(t) = \frac{1}{2}\left[{{}_{{t_1}}^{\rm{C}}D_t^\alpha f(t) + {{(-1)}^m}{}_t^{\rm{C}}D_{{t_2}}^\alpha f(t)} \right]。 $ | (9) |
Riesz-Riemann-Liouville分数阶导数下的分部积分公式[15]为
$ \begin{array}{l} \int_{{t_1}}^{{t_2}} {g(t)({}_{{t_1}}^RD_{{t_{\rm{2}}}}^\alpha f(t)){\rm{d}}t} {\kern 1pt} \\ = {(- 1)^m}\int_{t1}^{{t_2}} {f(t)({}_{{t_1}}^RD_{{t_2}}^\alpha g(t)){\rm{d}}t + } \\ \frac{1}{2}\sum\limits_{k = 0}^{m- 1} {{{\left. {\left[{{}_tD_{{t_2}}^{\alpha-k-1}g(t)\frac{{{{\rm{d}}^k}f(t)}}{{{\rm{d}}{t^k}}}} \right]} \right|}_{t = {t_2}}}} + \\ \frac{1}{2}\sum\limits_{k = 0}^{m - 1} {{{( - 1)}^{k + m}}{{\left. {\left[{{}_{{t_1}}D_t^{\alpha-k-1}g(t)\frac{{{{\rm{d}}^k}f(t)}}{{{\rm{d}}{t^k}}}} \right]} \right|}_{t = {t_1}}}} 。 \end{array} $ | (10) |
Riesz-Caputo分数阶导数下的分部积分公式[15]如下:
$ \begin{array}{l} \int_{t1}^{{t_2}} {g(t){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha f(t){\rm{d}}t} \\ = {(-1)^m}\int_{{t_1}}^{{t_2}} {f(t){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha g(t){\rm{d}}t + } \\ \left. {\sum\limits_{k = 1}^{m-1} {{{(-1)}^k}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^{\alpha + k - m}g(t)\frac{{{{\rm{d}}^{m - 1 - k}}}}{{{\rm{d}}{t^{m - 1 - k}}}}f(t)} } \right|_{{t_1}}^{{t_2}}。 \end{array} $ | (11) |
考虑由2n个Birkhoff变量
$ S({a^\mu }( \cdot )) = \int_{{t_1}}^{{t_2}} {\left\{ {\sum\limits_{\nu = 1}^{2n} {{R_\nu }(t, \;{a^\mu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }}-B(t, \;{a^\mu })} \right\}} {\rm{d}}t $ | (12) |
称为基于Riesz-Riemann-Liouville导数的分数阶Pfaff作用量。等时变分原理
$ \delta S = 0 $ | (13) |
带有交换关系
$ \begin{array}{l} \delta {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu } = {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha \delta {a^\nu }\\ (\nu = 1, 2, \;..., 2n) \end{array} $ | (14) |
以及端点条件
$ \begin{array}{l} {\left. {\delta {a^\nu }} \right|_{t = {t_1}}} = {\left. {\delta {a^\nu }} \right|_{t = {t_2}}} = 0\\ (\nu = 1, 2, \;..., 2n) \end{array} $ | (15) |
称为基于Riesz-Riemann-Liouville导数的分数阶Pfaff-Birkhoff原理。
由分数阶Pfaff-Birkhoff原理(13)~(15)容易导出如下方程[38]:
$ \begin{array}{l} \sum\limits_{\nu = 1}^{2n} {\left( {\frac{{\partial {R_\nu }}}{{\partial {a^\mu }}}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }} \right)-\frac{{\partial B}}{{\partial {a^\mu }}}-{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }} = {\rm{0}}\\ (\mu = 1, \;2, \;..., \;2n) \end{array} $ | (16) |
以及相应的横截性条件
$ \frac{1}{2}\sum\limits_{\nu = 1}^{2n} {{{\left. {[({}_tD_{{t_2}}^{\alpha-1}{R_\nu })\delta {a^\nu }]} \right|}_{t = {t_2}}}} - \frac{1}{2}\sum\limits_{\nu = 1}^{2n} {{{\left. {[({}_{{t_1}}D_t^{\alpha-1}{R_\nu })\delta {a^\nu }]} \right|}_{t = {t_1}}}} = 0。 $ | (17) |
由端点条件(15)易知横截性条件(17)恒成立。方程(16)称为Riesz-Riemann-Liouville导数下分数阶Birkhoff系统的分数阶Birkhoff方程。
当α→1时, 方程(16)成为
$ \begin{array}{l} \sum\limits_{\nu = 1}^{2n} {\left( {\frac{{\partial {R_\nu }}}{{\partial {a^\mu }}}-\frac{{\partial {R_\mu }}}{{\partial {a^\nu }}}} \right)} {{\dot a}^\nu }-\left( {\frac{{\partial B}}{{\partial {a^\mu }}} + \frac{{\partial {R_\mu }}}{{\partial t}}} \right) = 0\\ (\mu = 1, \;2, \;..., \;2n), \end{array} $ | (18) |
方程(18)是经典的Birkhoff方程。因此, 经典Birk-hoff方程是Riesz-Riemann-Liouville导数下的分数阶Birkhoff方程(16)的特例。
引进时间不变的单参数无限小变换群:
$ \begin{array}{l} {{\bar a}^\mu }(t) = {a^\mu }(t) + \varepsilon {\xi _\mu }(t, \;{a^\nu }) + o(\varepsilon )\\ (\mu = 1, \;2, \;..., \;2n), \end{array} $ | (19) |
下面, 定义Riesz-Riemann-Liouville导数下的分数阶Birkhoff方程(16)在无限小变换(19)下的Noether对称性, 并给出相应的分数阶守恒量。
定义1 如果分数阶Pfaff作用量(12)在无限小变换(19)作用下, 对于任意子区间
$ \begin{array}{l} \int_{{T_1}}^{T_2^{}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }}-B(t, \;{a^\nu })} \right\}{\rm{d}}t} \\ = \int_{{T_1}}^{{T_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{{\bar a}^\nu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {{\bar a}^\mu }}-B(t, \;{{\bar a}^\nu })} \right\}\;{\rm{d}}t} \end{array} $ | (20) |
始终成立, 则称这种不变性为Riesz-Riemann-Liouville导数下的分数阶Birkhoff系统(16)在时间不变的无限小变换下的Noether对称性。
定理1 对于Riesz-Riemann-Liouville导数下的分数阶Birkhoff系统(16), 如果时间不变的无限小变换(19)对应于定义1意义下的Noether对称性, 那么
$ \begin{array}{l} \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }(t, \;{a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, \;{a^\mu })} } \right){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }} + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {\xi _\mu }(t, \;{a^\nu })}-\\ \sum\limits_{\nu = 1}^{2n} {\frac{{\partial B(t, \;{a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, \;{a^\mu })} = 0 \end{array} $ | (21) |
成立。
证明 由积分区间[T1, T2]的任意性, 通过式(20)可得
$ \begin{array}{l} \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }}-B(t, \;{a^\nu })\\ = \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{{\bar a}^\nu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {{\bar a}^\mu }}-B(t, \;{{\bar a}^\nu }), \end{array} $ | (22) |
将式(22)两边对ε求导, 然后令ε=0, 有
$ \begin{array}{l} 0 = \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }(t, \;{a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, \;{a^\mu })} } \right){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }} + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu })\frac{{\rm{d}}}{{{\rm{d}}\varepsilon }}\left[{\frac{1}{{2\Gamma (m-\alpha )}}{{\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)}^m} \cdot } \right.} \\ \int_{{t_1}}^{{t_2}} {{{\left| {t-\tau } \right|}^{m-\alpha - 1}}{a^\mu }(\tau ){\rm{d}}\tau + } \\ {\left. {\left. {\frac{\varepsilon }{{2\Gamma (m - \alpha )}}{{\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)}^m}\int_{{t_1}}^{{t_2}} {{{\left| {t - \tau } \right|}^{m - \alpha - 1}}} {\xi _\mu }(\tau, \;{a^\nu })\;{\rm{d}}\tau } \right]} \right|_{\varepsilon = 0}} -\\ \sum\limits_{\nu = 1}^{2n} {\frac{{\partial B(t, \;{a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, \;{a^\mu })} \\ = \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }(t, \;{a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, \;{a^\mu })} } \right){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }} + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu })} \frac{1}{{2\Gamma (m -\alpha )}}{\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)^m} \cdot \\ \int_{{t_1}}^{{t_2}} {{{\left| {t -\tau } \right|}^{m - \alpha - 1}}{\xi _\mu }(\tau, \;{a^\nu }){\rm{d}}\tau } - \\ \sum\limits_{\nu = 1}^{2n} {\frac{{\partial B(t, \;{a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, \;{a^\mu })}, \end{array} $ | (23) |
显然式(23)即为式(21)。
下面引入Riesz-Riemann-Liouville导数下的分数阶守恒量的概念[23-25]。
定义2
$ I(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }) = \sum\limits_{i = 1}^r {I_i^1(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }) \cdot I_i^2(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu })}, $ | (24) |
其中r是任意整数, 对于每一组函数Ii1和Ii2(i=1, 2, …, r), 满足
$ {}^{\rm{R}}D_t^\alpha (I_i^1(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }), I_i^2(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu })) = 0 $ | (25) |
或
$ {}^{\rm{R}}D_t^\alpha (I_i^2(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }), I_i^1(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu })) = 0, $ | (26) |
其中, 算子
$ {}^{\rm{R}}D_t^\alpha (f, g) = g{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha f + f{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha g。 $ | (27) |
当α=1时, 式(27)给出
$ {}^{\rm{R}}D_t^1(f, g) = g{}_{{t_1}}^{\rm{R}}D_{{t_2}}^1f + f{}_{{t_1}}^{\rm{R}}D_{{t_2}}^1g = \dot fg + f\dot g = \frac{{\rm{d}}}{{{\rm{d}}t}}(fg)。 $ | (28) |
定理2 对于Riesz-Riemann-Liouville导数下的分数阶Birkhoff系统(16), 如果时间不变的无限小变换(19)对应于定义1意义下的Noether对称性, 那么
$ I(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }) = \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){\xi _\mu }(t, \;{a^\nu })} $ | (29) |
是系统的分数阶守恒量。
证明 由分数阶Birkhoff方程(16)可得
$ \frac{{\partial B}}{{\partial {a^\mu }}} = \sum\limits_{\nu = 1}^{2n} {\left( {\frac{{\partial {R_\nu }}}{{\partial {a^\mu }}}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }} \right)}-{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }, $ | (30) |
由于时间不变的无限小变换(19)相应于定义1意义下的Noether对称性, 故将式(30)代入式(21), 得
$ \begin{array}{l} \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }}}{{\partial {a^\nu }}}{\xi _\nu }} } \right){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }} + \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {\xi _\mu }}-\\ \sum\limits_{\mu = 1}^{2n} {\sum\limits_{\nu = 1}^{2n} {\left( {\frac{{\partial {R_\nu }}}{{\partial {a^\mu }}}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }} \right)} } \;{\xi _\mu } + \sum\limits_{\mu = 1}^{2n} {{\xi _\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }} = 0, \end{array} $ | (31) |
化简得
$ \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {\xi _\mu }} + \sum\limits_{\mu = 1}^{2n} {{\xi _\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }} = 0, $ | (32) |
即
$ \sum\limits_{\mu = 1}^{2n} {\{ {}^{\rm{R}}D_t^\alpha ({R_\mu }, \;{\xi _\mu })\} } = 0。 $ | (33) |
由定义2可知, (29)式是所论分数阶Birkhoff系统的分数阶守恒量。
下面, 考虑时间变化的单参数无限小变换群:
$ \left\{ \begin{array}{l} \bar t = t + \varepsilon \zeta (t, \;{a^\nu }) + o(\varepsilon ), \\ {{\bar a}^\mu }(t) = {a^\mu }(t) + \varepsilon {\xi _\mu }(t, {a^\nu }) + o(\varepsilon )\;\;\\ \;\;\;\;\;\;\;\;\;\;\;(\mu = 1, \;2, \;..., \;2n), \end{array} \right. $ | (34) |
定义分数阶Birkhoff系统(16)在无限小变换(34)下的Noether对称性, 并给出相应的分数阶守恒量。
定义3 如果分数阶Pfaff作用量(12)在无限小变换(34)作用下, 对于任意的子区间
$ \begin{array}{l} \int_{{T_1}}^{{T_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }}-B(t, \;{a^\nu })} \right\}} \;{\rm{d}}t\\ = \int_{{{\bar T}_1}}^{{{\bar T}_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(\bar t, \;{{\bar a}^\nu }){}_{{{\bar t}_1}}^{\rm{R}}D_{{{\bar t}_2}}^\alpha {{\bar a}^\mu }}-B(\bar t, \;{{\bar a}^\nu })} \right\}} \;{\rm{d}}\bar t \end{array} $ | (35) |
始终成立, 则称这种不变性为Riesz-Riemann-Liouville导数下的分数阶Birkhoff系统(16)在时间变化的无限小变换(34)下的Noether对称性。
定理3 对于Riesz-Riemann-Liouville导数下的分数阶Birkhoff系统(16), 如果时间变化的无限小变换(34)对应于定义3意义下的Noether对称性, 那么
$ I(t, {a^\nu }, {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\nu }) = \sum\limits_{\mu = 1}^{2n} {{R_\mu }{\xi _\mu }} + \left[{(1-\alpha )\sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }}-B} \right]\zeta $ | (36) |
是系统的分数阶守恒量。
证明 取关于时间t (t是独立变量)的李普希兹变换:
$ t \in \left[{{t_1}, \;\;{t_2}} \right] \mapsto \sigma f(\lambda ) \in \left[{{\sigma _1}, \;\;{\sigma _2}} \right], $ | (37) |
当λ=0时, 满足
$ {t'_\sigma } = \frac{{{\rm{d}}t(\sigma )}}{{{\rm{d}}\sigma }} = f(\lambda ) = 1。 $ |
在变换(37)作用下, 分数阶Pfaff作用量(12)成为
$ \begin{array}{l} \bar S(t( \cdot ), {a^\mu }( \cdot ))\\ = \int_{{\sigma _1}}^{{\sigma _2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma )))} } \right.} {}_{{\sigma _1}}^{\rm{R}}D_{{\sigma _2}}^\alpha {a^\mu }(t(\sigma ))-\\ \left. {B(t(\sigma ), {a^\nu }(t(\sigma )))} \right\}{{t'}_\sigma }{\rm{d}}\sigma, \end{array} $ | (38) |
其中, t (σ1)=t1, t (σ2)=t2,
$ \begin{array}{l} {}_{{\sigma _1}}^{\rm{R}}D_{{\sigma _2}}^\alpha {a^\mu }(t(\sigma )) = \frac{1}{{2\Gamma (m-\alpha )}}{\left( {\frac{{\rm{d}}}{{{\rm{d}}t(\sigma )}}} \right)^m} \cdot \\ \int_{\frac{{{t_1}}}{{f(\lambda )}}}^{\frac{{{t_2}}}{{f(\lambda )}}} {{{\left| {\sigma f(\lambda )-\tau } \right|}^{m-\alpha - 1}}{a^\mu }(\tau {f^{ - 1}}(\lambda ))} \;{\rm{d}}\tau \\ = \frac{{{{({{t'}_\sigma })}^{ - \alpha }}}}{{2\Gamma (m - \alpha )}}{\left( {\frac{{\rm{d}}}{{{\rm{d}}\sigma }}} \right)^m}\int_{\frac{{{t_1}}}{{{{({{t'}_\sigma })}^2}}}}^{\frac{{{t_2}}}{{{{({{t'}_\sigma })}^2}}}} {|\sigma - s{|^{m - \alpha - 1}}{a^\mu }(s)} \;{\rm{d}}s\\ = {({{t'}_\sigma })^{ - \alpha }}{}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma )。 \end{array} $ | (39) |
将式(39)代入式(38), 得
$ \begin{array}{l} = \int_{\sigma 1}^{{\sigma _2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma ))){{({{t'}_\sigma })}^{-\alpha }}} } \right.} \cdot \\ \left. {{}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma )-B(t(\sigma ), {a^\nu }(t(\sigma )))} \right\}{{t'}_\sigma }{\rm{d}}\sigma \\ \dot = \int_{\sigma 1}^{{\sigma _2}} {{{\bar B}_f}(t(\sigma ), {a^\nu }\left( {t(\sigma )} \right), {{t'}_\sigma }, {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma )){\rm{d}}\sigma } \\ = \int_{t1}^{{t_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }}-B(t, \;{a^\nu })} \right\}} \;{\rm{d}}t\\ = S({a^\mu }( \cdot ))。 \end{array} $ | (40) |
如果分数阶Pfaff作用量(12)在定义3意义下是不变的, 那么分数阶Pfaff作用量(38)在定义1意义下不变。由定理2可以得到
$ \begin{array}{l} {I_f}(t(\sigma ), {a^\nu }(t(\sigma )), \;{{t'}_\sigma }, {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma ))\\ = \sum\limits_{\mu = 1}^{2n} {\frac{{\partial {{\bar B}_f}}}{{\partial {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma )}}{\xi _\mu }} + \frac{{\partial {{\bar B}_f}}}{{\partial {{t'}_\sigma }}}\zeta, \end{array} $ | (41) |
式(41)是系统的分数阶守恒量。当λ=0时, 有
$ {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma ) = {}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }(t), $ | (42) |
因此, 可以得到
$ \frac{{\partial {{\bar B}_f}}}{{\partial {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma )}} = {R_\mu }(t, {a^\nu }(t)) $ | (43) |
以及
$ \begin{array}{l} \frac{{\partial {{\bar B}_f}}}{{\partial {{t'}_\sigma }}} = \frac{\partial }{{\partial {{t'}_\sigma }}}\left[{\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma )))\frac{{{{({{t'}_\sigma })}^{-\alpha }}}}{{2\Gamma (m-\alpha )}}{{\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)}^m}} } \right. \cdot \\ \left. {\int_{\frac{{{t_1}}}{{{{({{t'}_\sigma })}^2}}}}^{\frac{{{t_2}}}{{{{({{t'}_\sigma })}^2}}}} {|\sigma-s{|^{m - \alpha - 1}}{a^\mu }(s)} \;{\rm{d}}s} \right]{{t'}_\sigma } + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma ))){{({{t'}_\sigma })}^{ - \alpha }}} \cdot \\ {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{\rm{R}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma ) - B(t(\sigma ), {a^\nu }(t(\sigma )))\\ = \left[{-\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma )))\frac{{\alpha {{({{t'}_\sigma })}^{-\alpha-1}}}}{{2\Gamma (m - \alpha )}}{{\left( {\frac{{\rm{d}}}{{{\rm{d}}t}}} \right)}^m}} } \right. \cdot \\ \left. {\int_{\frac{{{t_1}}}{{{{({{t'}_\sigma })}^2}}}}^{\frac{{{t_2}}}{{{{({{t'}_\sigma })}^2}}}} {{{\left| {\sigma - s} \right|}^{m - \alpha - 1}}{a^\mu }(s)\;{\rm{d}}s} } \right]{{t'}_\sigma } + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu }(t)){}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }(t)} -B(t, {a^\nu }(t))\\ = -\alpha \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }} + \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^\mu }} -B。 \end{array} $ | (44) |
将式(44)和(43)代入式(41), 得到守恒量式(36)。
定理2和定理3称为Riesz-Riemann-Liouville导数下分数阶Birkhoff系统的分数阶Noether定理。显然, 当α=1时, 定理2和定理3给出经典Birkhoff系统的Noether定理。
3 Riesz-Caputo导数下分数阶Birk-hoff系统的Noether对称性积分
$ A({a^\mu }( \cdot )) = \int_{{t_1}}^{{t_2}} {\left\{ {\sum\limits_{\nu = 1}^{2n} {{R_\nu }(t, {a^\mu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }}-B(t, {a^\mu })} \right\}} \;{\rm{d}}t $ | (45) |
称为基于Riesz-Caputo导数的分数阶Pfaff作用量。等时变分原理
$ \delta A = 0 $ | (46) |
带有交换关系
$ \begin{array}{l} \delta {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu } = {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha \delta {a^\nu }\\ (\nu = 1, \;2, \;\;..., \;2n) \end{array} $ | (47) |
以及端点条件
$ \begin{array}{l} {\left. {\delta {a^\nu }} \right|_{t = {t_1}}} = {\left. {\delta {a^\nu }} \right|_{t = {t_2}}} = 0\\ (\nu = 1, \;2, \;..., \;2n) \end{array} $ | (48) |
称为基于Riesz-Caputo导数的分数阶Pfaff-Birkhoff原理。
设
$ \begin{array}{l} \sum\limits_{\nu = 1}^{2n} {\left( {\frac{{\partial {R_\nu }}}{{\partial {a^\mu }}}{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }} \right)-\frac{{\partial B}}{{\partial {a^\mu }}}-{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }} = {\rm{0}}\\ {\kern 1pt} {\kern 1pt} (\mu = 1, \;\;2, \;\;..., \;\;2n), \end{array} $ | (49) |
以及相应的横截性条件
$ \sum\limits_{\nu = 1}^{2n} {\left. {{}_{{t_1}}^{\rm{R}}D_{{t_2}}^{\alpha-1}{R_\nu }\delta {a^\nu }} \right|_{{t_1}}^{{t_2}}} = 0。 $ | (50) |
由端点条件(48)易知横截性条件(50)恒成立。方程(49)称为Riesz-Caputo导数下分数阶Birkhoff系统的分数阶Birkhoff方程。
当
下面定义Riesz-Caputo导数下的分数阶Birk-hoff方程(49)在无限小变换(19)下的Noether对称性, 并给出相应的分数阶守恒量。
定义4 如果分数阶Pfaff作用量(45)在无限小变换(19)作用下, 对于任意的子区间
$ \begin{array}{l} \int_{T1}^{{T_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }}-B(t, {a^\nu })} \right\}} \;{\rm{d}}t\\ = \int_{{T_1}}^{{T_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {{\bar a}^\nu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {{\bar a}^\mu }}-B(t, {{\bar a}^\nu })} \right\}} \;{\rm{d}}t, \end{array} $ | (51) |
则称这种不变性为Riesz-Caputo导数下的分数阶Birkhoff系统(49)在时间不变的无限小变换下的Noether对称性。
定理4 对于Riesz-Caputo导数下的分数阶Birkhoff方程(49), 如果时间不变的无限小变换(19)相应于定义4意义下的Noether对称性, 那么
$ \begin{array}{l} \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }(t, {a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, {a^\mu })} } \right){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }} + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {\xi _\mu }(t, {a^\nu })}-\\ \sum\limits_{\nu = 1}^{2n} {\frac{{\partial B\left( {t, {a^\nu }} \right)}}{{\partial {a^\nu }}}{\xi _\nu }(t, {a^\mu })} = 0 \end{array} $ | (52) |
成立。
证明 由积分区间[T1, T2]的任意性, 通过式(51)可得
$ \begin{array}{l} \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }}-B(t, {a^\nu })\\ = \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {{\bar a}^\nu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {{\bar a}^\mu }}-B(t, {{\bar a}^\nu }), \end{array} $ | (53) |
将式(53)两边对ε求导, 然后令ε=0, 有
$ \begin{array}{l} 0 = \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }(t, {a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, {a^\mu })} } \right){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }} + \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu }) \cdot } \\ \frac{{\rm{d}}}{{{\rm{d}}\varepsilon }}\left[{\frac{1}{{2\Gamma (m- \alpha )}}\int_{{t_1}}^{{t_2}} {{{\left| {t- \tau } \right|}^{m- \alpha - 1}}\frac{{{{\rm{d}}^m}}}{{{\rm{d}}{\tau ^m}}}\left[{{a^\mu }(\tau )} \right]} \;{\rm{d}}\tau } \right. + \\ {\left. {\left. {\frac{\varepsilon }{{2\Gamma (m - \alpha )}}\int_{{t_1}}^{{t_2}} {{{\left| {t - \tau } \right|}^{m - \alpha - 1}}\frac{{{{\rm{d}}^m}}}{{{\rm{d}}{\tau ^m}}}\left[{{\xi _\mu }(\tau, {a^\nu })} \right]} \;\;{\rm{d}}\tau } \right]} \right|_{\varepsilon = 0}} - \\ \sum\limits_{\nu = 1}^{2n} {\frac{{\partial B(t, {a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, {a^\mu })} \\ = \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }(t, {a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, {a^\mu })} } \right){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }} + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu })} \cdot \frac{1}{{2\Gamma (m - \alpha )}} \cdot \\ \int_{{t_1}}^{{t_2}} {{{\left| {t - \tau } \right|}^{m - \alpha - 1}}\frac{{{{\rm{d}}^m}}}{{{\rm{d}}{\tau ^m}}}[{\xi _\mu }(\tau, {a^\nu })]} \;{\rm{d}}\tau -\\ \sum\limits_{\nu = 1}^{2n} {\frac{{\partial B(t, {a^\nu })}}{{\partial {a^\nu }}}{\xi _\nu }(t, {a^\mu })}, \end{array} $ | (54) |
显然, 式(54)即为式(52)。
下面引入Riesz-Caputo导数下的分数阶守恒量的概念[25]。
定义5
$ = \sum\limits_{i = 1}^r {I_i^1(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }) \cdot } I_i^2(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }) $ | (55) |
其中r是任意整数, 对于每一组函数Ii1和Ii2(i=1, 2, …, r), 满足
$ {}^{{\rm{RC}}}D_t^\alpha (I_i^1(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }), I_i^2(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu })) = 0 $ | (56) |
或
$ {}^{{\rm{RC}}}D_t^\alpha (I_i^2(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }), \;I_i^1(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu })) = 0, $ | (57) |
其中, 算子
$ {}^{{\rm{RC}}}D_t^\alpha (f, g) = g{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha f + f{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha g。 $ | (58) |
当α=1时, 式(58)给出
$ {}^{{\rm{RC}}}D_t^1(f, g) = g{}_{{t_1}}^{\rm{R}}D_{{t_2}}^1f + f{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^1g = \dot fg + f\dot g = \frac{{\rm{d}}}{{{\rm{d}}t}}(fg) $ | (59) |
此时,
定理5 对于Riesz-Caputo导数下的分数阶Birkhoff系统(49), 如果时间不变的无限小变换(19)相应于定义4意义下的Noether对称性, 则
$ I(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }) = \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu }){\xi _\mu }(t, {a^\nu })} $ | (60) |
是系统的分数阶守恒量。
证明 由分数阶Birkhoff方程(49)可得
$ \frac{{\partial B}}{{\partial {a^\mu }}} = \sum\limits_{\nu = 1}^{2n} {\left( {\frac{{\partial {R_\nu }}}{{\partial {a^\mu }}}{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }} \right)}-{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }, $ | (61) |
由于时间不变的无限小变换(19)相应于定义4意义下的Noether对称性, 故将式(61)代入式(52), 得
$ \begin{array}{l} \sum\limits_{\mu = 1}^{2n} {\left( {\sum\limits_{\nu = 1}^{2n} {\frac{{\partial {R_\mu }}}{{\partial {a^\nu }}}{\xi _\nu }} } \right){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }} + \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {\xi _\mu }}-\\ \sum\limits_{\mu = 1}^{2n} {\sum\limits_{\nu = 1}^{2n} {\left( {\frac{{\partial {R_\nu }}}{{\partial {a^\mu }}}{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu }} \right)} } \;{\xi _\mu } + \sum\limits_{\mu = 1}^{2n} {{\xi _\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }} = 0, \end{array} $ | (62) |
化简得
$ \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {\xi _\mu }} + \sum\limits_{\mu = 1}^{2n} {{\xi _\mu }{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {R_\mu }} = 0, $ | (63) |
即
$ \sum\limits_{\mu = 1}^{2n} {\{ {}^{{\rm{RC}}}D_t^\alpha ({R_\mu }, \;{\xi _\mu })\} } = 0。 $ | (64) |
由定义5可知, 式(60)是所论分数阶Birkhoff系统(49)的分数阶守恒量。
下面, 定义Riesz-Caputo导数下的分数阶Birk-hoff方程(49)在时间变化的无限小变换(34)下的Noether对称性, 并给出相应的分数阶守恒量。
定义6 如果分数阶Pfaff作用量(45)在无限小变换(34)作用下, 对于任意子区间
$ \begin{array}{l} \int_{{T_1}}^{{T_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }}-B(t, \;{a^\nu })} \right\}} \;{\rm{d}}t\\ = \int_{{{\bar T}_1}}^{{{\bar T}_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(\bar t, \;{{\bar a}^\nu }){}_{{{\bar t}_1}}^{{\rm{RC}}}D_{{{\bar t}_2}}^\alpha {{\bar a}^\mu }}-B(\bar t, \;{{\bar a}^\nu })} \right\}} \;{\rm{d}}\bar t \end{array} $ | (65) |
成立, 则称这种不变性为Riesz-Caputo导数下的分数阶Birkhoff系统(49)在时间变化的无限小变换(34)下的Noether对称性。
定理6 对于Riesz-Caputo导数下的分数阶Birkhoff系统(49), 如果时间变化的无限小变换(34)相应于定义6意义下的Noether对称性, 则
$ \begin{array}{l} I(t, {a^\nu }, {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\nu })\\ = \sum\limits_{\mu = 1}^{2n} {{R_\mu }{\xi _\mu }} + \left[{(1-\alpha )\sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }}-B} \right]\zeta \end{array} $ | (66) |
是系统的分数阶守恒量。
证明 取关于时间t(t是独立变量)的李普希兹变换
$ t \in [{t_1}, \;{t_2}] \mapsto \sigma f(\lambda ) \in [{\sigma _1}, \;{\sigma _2}], $ | (67) |
当
$ {t'_\sigma } = \frac{{{\rm{d}}t(\sigma )}}{{{\rm{d}}\sigma }} = f(\lambda ) = 1。 $ |
在变换(67)作用下, 分数阶Pfaff作用量(45)成为
$ \begin{array}{l} \bar A(t( \cdot ), {a^\mu }( \cdot ))\\ = \int_{{\sigma _1}}^{{\sigma _2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), \;{a^\nu }(t(\sigma ))){}_{{\sigma _1}}^{{\rm{RC}}}D_{{\sigma _2}}^\alpha {a^\mu }(t(\sigma ))-} } \right.} \\ \left. {B(t(\sigma ), \;{a^\nu }(t(\sigma )))} \right\}{{t'}_\sigma }{\rm{d}}\sigma, \end{array} $ | (68) |
其中,
$ \begin{array}{l} {}_{{\sigma _1}}^{{\rm{RC}}}D_{{\sigma _2}}^\alpha {a^\mu }(t(\sigma ))\\ = \frac{1}{{2\Gamma (m- \alpha )}} \cdot \\ \int_{\frac{{{t_1}}}{{f(\lambda )}}}^{\frac{{{t_2}}}{{f(\lambda )}}} {{{\left| {\sigma f(\lambda )- \tau } \right|}^{m- \alpha - 1}}\frac{{{{\rm{d}}^m}}}{{{\rm{d}}{\tau ^m}}}[{a^\mu }(\tau {f^{-1}}(\lambda ))]} \;{\rm{d}}\tau \\ = \frac{{{{({{t'}_\sigma })}^{ - \alpha }}}}{{2\Gamma (m - \alpha )}}\int_{\frac{{{t_1}}}{{{{({{t'}_\sigma })}^2}}}}^{\frac{{{t_2}}}{{{{({{t'}_\sigma })}^2}}}} {|\sigma - s{|^{m - \alpha - 1}}\frac{{{{\rm{d}}^m}}}{{{\rm{d}}{s^m}}}[{a^\mu }(s)]} \;{\rm{d}}s\\ = {({{t'}_\sigma })^{ -\alpha }}{}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma ) \end{array} $ | (69) |
将式(69)代入式(68), 得
$ \begin{array}{l} \bar A(t( \cdot ), {a^\mu }( \cdot ))\\ = \int_{{\sigma _1}}^{{\sigma _2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ){a^\nu }(t(\sigma ))){{({{t'}_\sigma })}^{-\alpha }}} } \right.} \cdot \\ \left. {{}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma )-B(t(\sigma ), {a^\nu }(t(\sigma )))} \right\}{{t'}_\sigma }{\rm{d}}\sigma \\ \dot = \int_{{\sigma _1}}^{{\sigma _2}} {{{\bar B}_f}(t(\sigma ), {a^\nu }(t(\sigma )), {{t'}_\sigma }, {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma ))} \;{\rm{d}}\sigma \\ = \int_{{t_1}}^{{t_2}} {\left\{ {\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, \;{a^\nu }){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }}-B(t, \;{a^\nu })} \right\}} \;{\rm{d}}t\\ = A({a^\mu }( \cdot ))。 \end{array} $ | (70) |
如果分数阶Pfaff作用量(45)在定义6意义下是不变的, 那么分数阶Pfaff作用量(68)在定义4意义下不变。由定理5可以得到
$ \begin{array}{l} {I_f}(t(\sigma ), {a^\nu }(t(\sigma )), \;{{t'}_\sigma }, {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma ))\\ = \sum\limits_{\mu = 1}^{2n} {\frac{{\partial {{\bar B}_f}}}{{\partial {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma )}}{\xi _\mu }} + \frac{{\partial {{\bar B}_f}}}{{\partial {{t'}_\sigma }}}\zeta, \end{array} $ | (71) |
式(71)是系统(49)的分数阶守恒量。当
$ {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma ) = {}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }(t), $ | (72) |
因此, 可以得到
$ \frac{{\partial {{\bar B}_f}}}{{\partial {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\nu }(\sigma )}} = {R_\mu }(t, {a^\nu }(t)) $ | (73) |
以及
$ \begin{array}{l} \frac{{\partial {{\bar B}_f}}}{{\partial {{t'}_\sigma }}} = \frac{\partial }{{\partial {{t'}_\sigma }}}\left[{\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma )))\frac{{{{({{t'}_\sigma })}^{-\alpha }}}}{{2\Gamma (m-\alpha )}}} } \right. \cdot \\ \int_{\frac{{{t_1}}}{{{{({{t'}_\sigma })}^2}}}}^{\frac{{{t_2}}}{{{{({{t'}_\sigma })}^2}}}} {{{\left| {\sigma-s} \right|}^{m - \alpha - 1}}} \;\left. {\frac{{{{\rm{d}}^m}}}{{{\rm{d}}{s^m}}}{a^\mu }(s){\rm{d}}s} \right]{{t'}_\sigma } + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma ))){{({{t'}_\sigma })}^{ - \alpha }}} \cdot \\ {}_{{t_1}/{{({{t'}_\sigma })}^2}}^{{\rm{RC}}}D_{{t_2}/{{({{t'}_\sigma })}^2}}^\alpha {a^\mu }(\sigma ) - B(t(\sigma ), {a^\nu }(t(\sigma )))\\ = \left[{-\sum\limits_{\mu = 1}^{2n} {{R_\mu }(t(\sigma ), {a^\nu }(t(\sigma )))\frac{{\alpha {{({{t'}_\sigma })}^{-\alpha-1}}}}{{2\Gamma (m - \alpha )}}} } \right. \cdot \\ \left. {\int_{\frac{{{t_1}}}{{{{({{t'}_\sigma })}^2}}}}^{\frac{{{t_2}}}{{{{({{t'}_\sigma })}^2}}}} {{{\left| {\sigma - s} \right|}^{m - \alpha - 1}}\frac{{{{\rm{d}}^m}}}{{{\rm{d}}{s^m}}}{a^\mu }(s)} \;{\rm{d}}s} \right]{{t'}_\sigma } + \\ \sum\limits_{\mu = 1}^{2n} {{R_\mu }(t, {a^\nu }(t)){}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }(t)} -B(t, {a^\nu }(t))\\ = -\alpha \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }} + \sum\limits_{\mu = 1}^{2n} {{R_\mu }{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^\mu }} -B。 \end{array} $ | (74) |
将式(74)和(73)代入式(71), 得到守恒量式(66)。
4 算例例1 已知四阶分数阶Birkhoff系统在Riesz-Riemann-Liouville导数下的Pfaff作用量为
$ \begin{array}{l} S({a^\mu }( \cdot )) = \int_{{t_1}}^{{t_2}} {\left\{ {{a^2}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^1} + {a^{\rm{4}}}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^3}-} \right.} \\ \left. {\frac{1}{2}({{({a^4})}^2}-2{a^2}{a^3})} \right\}{\rm{d}}t, \end{array} $ | (75) |
试研究该系统的分数阶Noether对称性与分数阶守恒量。
从作用量(75)可知, 系统的Birkhoff函数和Birkhoff函数组为
$ \left\{ \begin{array}{l} B = \frac{1}{2}({({a^4})^2}-2{a^2}{a^3}), \\ {R_1} = {a^2}, \\ {R_2} = 0, \\ {R_3} = {a^4}, \\ {R_4} = 0, \end{array} \right. $ | (76) |
取无限小变换(34)的生成元为
$ \left\{ \begin{array}{l} \zeta = \frac{2}{3}t, \;\\ {\xi _1} = {a^1}, \;\\ {\xi _2} =-{a^2}, \;\\ {\xi _3} = \frac{1}{3}{a^3}, \;\\ {\xi _4} =-\frac{1}{3}{a^4}, \end{array} \right. $ | (77) |
由定义3, 生成元(77)对应于系统的Noether对称性。根据定理3, 得到
$ \begin{array}{l} I = {a^1}{a^2} + \frac{1}{3}{a^3}{a^4} + \frac{2}{3}t\left[{(1-\alpha ){a^2}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^1} + } \right.\\ \left. {(1-\alpha ){a^4}{}_{{t_1}}^{\rm{R}}D_{{t_2}}^\alpha {a^3}-\frac{1}{2}({{({a^4})}^2} - 2{a^2}{a^3})} \right], \end{array} $ | (78) |
式(78)是该系统的一个分数阶守恒量。
例2 已知四阶分数阶Birkhoff系统在Riesz-Caputo导数下的分数阶Pfaff作用量为
$ \begin{array}{l} A({a^\mu }( \cdot )) = \int_{{t_1}}^{{t_2}} {\left\{ {{a^3}{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^1} + {a^4}{}_{{t_1}}^{{\rm{RC}}}D_{{t_2}}^\alpha {a^2}- } \right.} \\ \left. {\frac{1}{2}[{{({a^3})}^2} + {{({a^4})}^2}]} \right\}{\rm{d}}t, \end{array} $ | (79) |
试研究该系统的分数阶Noether对称性与分数阶守恒量。
如取生成元为
$ \left\{ \begin{array}{l} \zeta = 0, {\kern 1pt} {\kern 1pt} \\ {\xi _1} = 1, {\kern 1pt} {\kern 1pt} \\ {\xi _2} = 0, {\kern 1pt} {\kern 1pt} \\ {\xi _3} = 0, {\kern 1pt} {\kern 1pt} \\ {\xi _4} = 0, \end{array} \right. $ | (80) |
由定义4, 生成元(80)相应于分数阶Birkhoff系统(79)的Noether对称性。因此, 由定理5得到
$ I = {a^3}, $ | (81) |
式(81)是该分数阶Birkhoff系统的一个守恒量。
5 结论Birkhoff力学是Hamilton力学的推广, 对Birk-hoff力学的研究是近代分析力学的一个重要发展方向。由于应用分数阶模型可以更准确地描述复杂系统的动力学行为, 因此对分数阶Birkhoff系统动力学的研究具有重要意义。本文提出并研究了分数阶Birkhoff系统在Riesz-Riemann-Liouville分数阶导数和Riesz-Caputo分数阶导数下的Noether对称性与守恒量问题, 建立了分数阶Noether定理。定理的证明分成两步:首先在时间不变的无限小变换下给出证明; 然后利用时间重新参数化技术, 得到一般情况下的分数阶Noether定理。分数阶Noether定理揭示了分数阶Noether对称性与分数阶守恒量之间的内在联系。由于求解Riemann-Liouville导数下的分数阶微分方程与求解Caputo导数下的分数阶微分方程所伴随的初始条件的形式不同, 后者仅涉及整数阶导数的初始条件, 因此, Riesz-Caputo导数下的结果更易于应用。当然, 两者都以经典Birkhoff系统的Noether定理作为其特例。因此, 本文研究的方法和结果具有普遍意义。
[1] | Noether A E. Nach-richten von der Gesellschaft der Wissenschaften zu G ttingen. Mathematisch-Physikalische Klasse , 1918, KI (Ⅱ) : 235–257 . |
[2] | Djukić D D S, Vujanović B D. Noether's theory in classical nonconservative mechanics. Acta Mecha-nica , 1975, 23 (1/2) : 17–27 . |
[3] | 李子平. 约束系统的变换性质. 物理学报 , 1981, 30 (12) : 1659–1671. |
[4] | Bahar L Y, Kwatny H G. Extension of Noether's theorem to constrained nonconservative dynamical systems. International Journal of Non-Linear Mecha-nics , 1987, 22 (2) : 125–138 DOI:10.1016/0020-7462(87)90015-1 . |
[5] | Liu D. Noether's theorem and its inverse of nonholonomic nonconservative dynamical systems. Science in China: Ser A , 1991, 34 (4) : 419–429 . |
[6] | 梅凤翔. 李群和李代数对约束力学系统的应用. 北京: 科学出版社, 1999 . |
[7] | Xia L L, Chen L Q. Noether conserved quantities and Lie point symmetries for difference nonholonomic Hamiltonian systems in irregular lattices. Chinese Physics B , 2012, 21 (7) : 070202 DOI:10.1088/1674-1056/21/7/070202 . |
[8] | Wang P, Xue Y, Liu Y L. Noether symmetry and conserved quantities of the analytical dynamics of a Cosserat thin elastic rod. Chinese Physics B , 2013, 22 (10) : 104503 DOI:10.1088/1674-1056/22/10/104503 . |
[9] | Cai P P, Fu J L, Guo Y X. Noether symmetries of the nonconservative and nonholonomic systems on time scales. Science China: Physics, Mechanics & Astronomy , 2013, 56 (5) : 1017–1028 . |
[10] | Zhai X H, Zhang Y. Noether symmetries and con-served quantities for Birkhoffian systems with time delay. Nonlinear Dynamics , 2014, 77 (1/2) : 73–86 . |
[11] | Oldham K B, Spanier J. The Fractional Calculus. San Diego: Academic Press, 1974 . |
[12] | Riewe F. Nonconservative Lagrangian and Hamil-tonian mechanics. Physical Review E , 1996, 53 (2) : 1890–1899 DOI:10.1103/PhysRevE.53.1890 . |
[13] | Riewe F. Mechanics with fractional derivatives. Physical Review E , 1997, 55 (3) : 3581–3592 DOI:10.1103/PhysRevE.55.3581 . |
[14] | Agrawal O P. Formulation of Euler-Lagrange equa-tions for fractional variational problems. Journal of Mathematical Analysis and Applications , 2002, 272 (1) : 368–379 DOI:10.1016/S0022-247X(02)00180-4 . |
[15] | Agrawal O P. Fractional variational calculus in terms of Riesz fractional derivatives. Journal of Physics A: Mathematical and Theoretical , 2007, 40 (24) : 6287–6303 DOI:10.1088/1751-8113/40/24/003 . |
[16] | Baleanu D, Avkar T. Lagrangians with linear velo-cities within Riemann-Liouville fractional derivatives. Nuovo Cimento B , 2003, 119 (1) : 73–79 . |
[17] | Baleanu D, Trujillo J I. A new method of finding the fractional Euler-Lagrange and Hamilton equations within Caputo fractional derivatives. Communications in Nonlinear Science and Numerical Simulation , 2010, 15 (5) : 1111–1115 DOI:10.1016/j.cnsns.2009.05.023 . |
[18] | Atanacković T M, Konjik S, Pilipović S. Variational problems with fractional derivatives: Euler-Lagrange equations. Journal of Physics A: Mathematical and Theoretical , 2008, 41 (9) : 095201 DOI:10.1088/1751-8113/41/9/095201 . |
[19] | Atanacković T M, Konjik S, Oparnica L, et al. Generalized Hamilton's principle with fractional derivatives. Journal of Physics A: Mathematical and Theoretical , 2010, 43 (25) : 255203 DOI:10.1088/1751-8113/43/25/255203 . |
[20] | El-Nabulsi A R. A fractional approach to noncon-servative Lagrangian dynamical systems. Fizika A , 2005, 14 (4) : 289–298 . |
[21] | El-Nabulsi A R. Fractional variational problems from extended exponentially fractional integral. Applied Mathematics and Computation , 2011, 217 (22) : 9492–9496 DOI:10.1016/j.amc.2011.04.007 . |
[22] | El-Nabulsi A R, Torres D F M. Non-standard fractional Lagrangians. Nonlinear Dynamics , 2013, 74 (1/2) : 381–394 . |
[23] | Frederico G S F, Torres D F M. A formulation of Noether's theorem for fractional problems of the calculus of variations. Journal of Mathematical Analysis and Applications , 2007, 334 (2) : 834–846 DOI:10.1016/j.jmaa.2007.01.013 . |
[24] | Frederico G S F, Torres D F M. Fractional isoperimetric Noether's theorem in the Riemann-Liouville sense. Reports on Mathematical Physics , 2013, 71 (3) : 291–304 DOI:10.1016/S0034-4877(13)60034-8 . |
[25] | Frederico G S F. Fractional optimal control in the sense of Caputo and the fractional Noether's theorem. International Mathematical Forum , 2008, 3 (10) : 479–493 . |
[26] | Frederico G S F, Torres D F M. Fractional Noether's theorem in the Riesz-Caputo sense. Applied Mathe-matics and Computation , 2010, 217 (3) : 1023–1033 DOI:10.1016/j.amc.2010.01.100 . |
[27] | Frederico G S F, Torres D F M. Constants of motion for fractional action-like variational problems. Inter-national Journal of Applied Mathematics , 2006, 19 (1) : 97–104 . |
[28] | Frederico G S F, Torres D F M. Non-conservative Noether's theorem for fractional action-like problems with intrinsic and observer times. International Journal of Ecological Economics & Statistics , 2007, 9 (F07) : 74–82 . |
[29] | Odzijewicz T, Malinowska A B, Torres D F M. Noether's theorem for fractional variational problems of variable order. Central European Journal of Physics , 2013, 11 (6) : 691–701 . |
[30] | Malinowska A B, Torres D F M. Introduction to the Fractional Calculus of Variations. London: Imperial College Press, 2012 . |
[31] | Zhang S H, Chen B Y, Fu J L. Hamilton formalism and Noether symmetry for mechanico electrical systems with fractional derivatives. Chinese Physics B , 2012, 21 (10) : 100202 DOI:10.1088/1674-1056/21/10/100202 . |
[32] | Zhang Y, Zhou Y. Symmetries and conserved quan-tities for fractional action-like Pfaffian variational problems. Nonlinear Dynamics , 2013, 73 (1/2) : 783–793 . |
[33] | Long Z X, Zhang Y. Noether's theorem for fractional variational problem from El-Nabulsi extended expo-nentially fractional integral in phase space. Acta Mechanica , 2014, 225 (1) : 77–90 DOI:10.1007/s00707-013-0956-5 . |
[34] | Long Z X, Zhang Y. Fractional Noether theorem based on extended exponentially fractional integral. International Journal of Theoretical Physics , 2014, 53 (3) : 841–855 DOI:10.1007/s10773-013-1873-z . |
[35] | Zhang Y, Zhai X H. Noether symmetries and con-served quantities for fractional Birkhoffian systems. Nonlinear Dynamics , 2015, 81 (1/2) : 469–480 . |
[36] | Podlubny I. Fractional differential equations. San Diego: Academic Press, 1999 . |
[37] | Kilbas A A, Srivastava H M, Trujillo J J. Theory and Applications of Fractional Differential Equations. Amsterdam: Elsevier, 2006 . |
[38] | Zhou Y, Zhang Y. Fractional Pfaff-Birkhoff principle and Birkhoff's equations in terms of Riesz fractional derivatives. Transactions of Nanjing University of Aeronautics and Astronautics , 2014, 31 (1) : 63–69 . |