­º­¶ | ·j´M

¡D­ì¸ü©ó¼Æ¾Ç¶Ç¼½²Ä¤C¨÷²Ä¤T´Á
¡D§@ªÌ·í®É¥ô±Ð©ó¥æ¤j¼Æ¾Ç¨t
 

½Í¬ã¨s¦~ÄÖµ²ºc¤§¼Æ¾Ç¼Ò«¬
Leslie's Model

³\¥@¾À

 
 

¥»¤å³Ì¥D­nªº¥Øªº¬O¤¶²Ð¦p¦ó¬ã¨s¤H¤f°Ê¤O¾Ç (Population Dynamics) ùتº¤@¨Ç¦³Ãö¦~ÄÖµ²ºc (Age Strucure) ¤§°ÝÃD¡CÁ|­Ó¨Ò¨Ó»¡©ú¡G ¥Ø«e¥xÆW¤H¤f¤¤ 0¡ã5 ·³¦³ x1 ¤H¡A6¡ã10 ·³¦³ x2¤H¡A¡K¡K¡A76¡ã80 ·³¦³ x16 ¤H¡C¸Õ°Ý20¦~©Î100¦~«á¡A³o¨Ç¤H¤fªºÅܤƦp¦ó¡H ÁÙ¦³¨C¤@­Ó¦~ÄÖ¤ÀÃþ (Age class) ¦bÁ`¤H¤f¤Wªº¤ñ¨Ò·|¤£·|«Üí©w¦aÁͦV¬Y¤@­Ó©T©w­È¡H¦pªG¬Oªº¸Ü¡A¦h§Ö¡H´NªÀ·|¾Ç¡B¸gÀپǦӨ¥¡A³o¬O¤@­Ó«Ü¹ê»Ú¦Ó¥B­È±o¬ã¨sªº­«­n°ÝÃD¡C ¤U­±§Ú­Ì´N­n¾É¥X¦³Ãö³o­Ó°ÝÃD¤§¼Æ¾Ç¼Ò«¬¢w¢wLeslie's Model¡C ¥¦¦P®É¤]¥i¥HÀ³¥Î¨ì¨ä¥L¥Íª«¡A¦p³½Ãþ¤Î©øÂC

­º¥ý¡A°²³]±q²{¦³ªº²Î­p¼Æ¾Ú¡A§Ú­Ì¯à¿ï¾Ü¥X¤@­Ó¾A·íªº³æ¦ì®É¶¡ T¡A ¦Ó«á±N¤H¤f¤À¦¨ A Ãþ¡C¥O¦V¶q $\overrightarrow{V}_N$ ¥Nªí¦b²Ä N ´Á®É¡]®É¶¡¬° NT¡^¤H¤fùتº¤k©Ê¦~ÄÖµ²ºc¡]¦b¦¹§Ú­Ì°²³]¨k©Ê¡A¤k©Ê¤H¤f¼Æ¥Ø¬Ûµ¥¡^¡A²¦Ó¨¥¤§¥O

\begin{displaymath}
\overrightarrow{V}_N =
\left[
\begin{array}{c}
V_{1,N} \\
\vdots \\
V_{A,N}
\end{array}\right]
\end{displaymath}

¦b¦¹¡A¤À¶q Vk,N¡Ak= 1,¡K,A ¥Nªí¦~ÄÖ¤¶©ó (k-1)T ¤Î kT ¤¤¶¡¤§¤k©ÊÁ`¤H¼Æ¡CÄ´¦p»¡À³¥Î¨ì¹ê»Ú¤H¤f®É¡A§Ú­Ì³q±`¨ú T=5 ¦~¡A¦Ó¥B±N¤H¤f¤À¦¨ 16 Ãþ¡A§Y A=16¡A

V1,N = ¦b²Ä N ´Á®É¡A¤¶©ó 0¡ã5 ·³¤§¤k©ÊÁ`¤H¼Æ¡C
V2,N = ¦b²Ä N ´Á®É¡A¤¶©ó 6¡ã10 ·³¤§¤k©ÊÁ`¤H¼Æ¡C
$\vdots$
V16,N = ¦b²Ä N ´Á®É¡A¤¶©ó 76¡ã80 ·³¤§¤k©ÊÁ`¤H¼Æ¡C

¦pªG¦~ÄÖ¶W¹L80·³®É¡A«h§Ú­Ì¤£¤©°Q½×¡C

¤U¤@¨B§Ú­Ì­n°µªº¤u§@¬O¦p¦ó§ä¥X²Ä N+1 ´Áªº¦~ÄÖµ²ºc¦V¶q (Age structure vector) $\overrightarrow{V}_{N+1}$¡A »P¦b²Ä N ´Á¤§¦~ÄÖµ²ºc¦V¶q $\overrightarrow{V}_N$ ¤§Ãö«Y¡C °²³]¤U¦Cªº bk ¤Î mk,k=1,¡K,A ¬°¤wª¾¡A

\begin{eqnarray*}
b_k &=& \mbox{{\fontfamily{cwM0}\fontseries{m}\selectfont \cha...
...inus0.1pt{\fontfamily{cwM1}\fontseries{m}\selectfont \char 170}}
\end{eqnarray*}


§Q¥Î bn¡Amk ¤Î Vk,N ¤§©w¸q¡A§Ú­Ì¥i¥H¾É¥X¤U¦CÃö«Y¦¡¡G

\begin{eqnarray*}
(1) V_{k,N+1} &=& m_{k-1} V_{k-1,N} , \\
k &=& 2,3,\cdots, A \\
(2) V_{1,N+1} &=& b_1 V_{1,N} + \cdots + b_m V_{A,N}
\end{eqnarray*}


¨ä¤¤(2)¦¡»¡©ú¤F±q²Ä N ´Á¨ì N+1 ´Á©Ò¥X¥Í¤§¤k«ÄÁ`¼Æ¡C ©Ò¥H¡A¦pªG§Ú­Ì°²³] $b_1,\cdots,b_A$ ¤Î $m_1,\cdots,m_A$ ³o¨Ç«D­t¤§¹ê¼Æ§¡¥i¥Ñ¤H¤f¤§²Î­p¸ê®Æ±o¨ì¡A«h§Ú­Ì¦³¤U¦C¦¡¤l

\begin{displaymath}
(3)
\begin{array}{ccl}
V_{1,N+1} &=& b_1 V_{1,N} + \cdots + ...
...N} \\
\vdots &&\\
V_{A,N+1} &=& m_{A-1} V_{A-1,N}
\end{array}\end{displaymath}

©ÎªÌ¥Î¯x°}¤§ªí¥Üªk¡A(3)¦¡¥i§ï¼g¬°

\begin{displaymath}
\begin{eqalign}
(4)& \quad \overrightarrow{V}_{N+1} = M \ove...
...
\end{array}\right] (\mbox{Reproduction Matrix})
\end{eqalign}\end{displaymath}

¦pªG¡A§Ú­Ì°²³]²{¦bªº¦~ÄÖµ²ºc¦V¶q $\overrightarrow{V}_0$¡C ¬°¤wª¾¡A«h¥Ñ(4)¦¡¡A§Ú­Ì±o¨ì¡G

\begin{displaymath}
(6) \quad \overrightarrow{V}_N = M^N \overrightarrow{V}_0
\end{displaymath}

©Ò¥H¡A§Ú­Ì±N¦~ÄÖµ²ºcªº°ÝÃDÅܦ¨¤@­Ó½u©Ê¥N¼Æªº°ÝÃD¡G

·í N «Ü¤j®É¡A¦V¶q $M^N \overrightarrow{V}_0$ ¦p¦óÅܤơH

¬°¤F¸Ñ¨M³o­Ó°ÝÃD¡A¥²¶·§Q¥Î½u©Ê¥N¼Æ¤¤¦³Ãö©T¦³­È (eigenvalue) ¤Î©T¦³¦V¶q (eigenvector) ¤§Æ[©À¤Î¨ä­«­n©w²z Primary decomposition Theorem¡]°Ñ¦Ò[1]¡^¡C ­º¥ý§Ú­Ì¦Ò¼{ A x A ¯x°} M ¤§©T¦³­È £f¡C ±q©T¦³­È¤§©w¸q¡A£f ¬° M ¤§¯S¼x¦h¶µ¦¡ (characteristic polynomial) $f(\lambda) = \mbox{det} (M-\lambda I) =0$ ¤§®Ú¡C ³q±`¯S¼x¦h¶µ¦¡«ÜÃøºâ¡A¦ý¦b³oùدx°} M ¦³¨ä¯S®í§Î¦¡ (5)¡A ©Ò¥H§Q¥Î­°¶¥®i¶}¦æ¦C¦¡ det ($M-\lambda I$)¡A±o¨ì

\begin{eqnarray*}
(7) \quad f(\lambda) &=& \lambda^A - b_1\lambda^{A-1} - b_2 m_1 \lambda^{A-2} - \cdots - (b_A m_1 \cdots m_{A-1}) \\
&=& 0
\end{eqnarray*}


¦]¬° f(0) < 0 ¦Ó¥B·í $\lambda >0$ °÷¤j®É $f(\lambda) > 0$¡A©Ò¥H $f(\lambda)=0$ ¥²¦³¤@¥¿¹ê®Ú¡C ¨Æ¹ê¤W¡A¦]¬°¯x°} M ùؤ§«Y¼Æ¬Ò¤j©ó©Îµ¥©ó¹s¡A®Ú¾Ú¦³¦Wªº Frobenius©w²z¡]°Ñ¦Ò[2]¡^§Ú­Ì±oª¾¦s¦b¤@©T¦³­È $\lambda_0 > 0$¡A ¦Ó¥B¨ä¥L A-1 ­Ó©T¦³­È £f¡Aº¡¨¬ $\vert\lambda\vert \leq \lambda_0$¡C ²{¦b¡A§Ú­Ì°²³]¯x°} M º¡¨¬¤U¦C©Ê½è¡G

(H) ¦s¦b¤@©T¦³­È $\lambda_0 > 0$ ¦Ó¥B¨ä¥L A-1 ­Ó©T¦³­È £f¡Aº¡¨¬ $\vert\lambda\vert < \lambda_0$¡C

¦b°²³](H)¤U¡A§Ú­Ì¥i¥H¥Î¼Æ­È¤ÀªR¤§¤èªk Power Method ¹ê»Ú¦aºâ¥X $\lambda_0$¡]°Ñ¦Ò[3]¡^¡C¦³¤F $\lambda_0$¡A¦]¬° $\lambda_0$ º¡¨¬(7)¦¡¡A§Ú­Ì¥i¥HÀˬd¤@¤U¤U¦C $\overrightarrow{\psi}_0$ ¬°¤@¹ïÀ³©ó £f ¤§©T¦³¦V¶q¡F

\begin{displaymath}
\begin{eqalign}
\overrightarrow{\psi}_0 =
\left[
\begin{arra...
...rrow{\psi}_0 &= \lambda_0 \overrightarrow{\psi}_0
\end{eqalign}\end{displaymath}

¥O E0 ¬°¥Ñ¦V¶q $\overrightarrow{\psi}_0$ ©Ò²£¥Í¤§¤@ºû¤lªÅ¶¡¡A $E_0 \subset \mathbf{R}^A$¡C ¥Ñ½u©Ê¥N¼Æªº Primary decomposition Theorem¡A§Ú­Ì¥i±N RA¼g¦¨ RA $= E_0 \oplus E_1$¡A¨ä¤¤ E1 ¬O¥H¹ï©ó©T¦³­È $\lambda \neq \lambda_0$ ¤§©T¦³¦V¶q (eigenvectors) ©Î¼s¸q©T¦³¦V¶q (generalized eigenvectors) ¬°¨ä°ò©³ (basis) ©Ò²Õ¦¨ªº A-1 ºû¤lªÅ¶¡¡C ¬°¤F°Q½×¤è«K°_¨£¡A§Ú­Ì´N°²³]¯x°} M ¤§©T¦³­È¬° $\lambda_0$¡A $\lambda_1$,¡K,$\lambda_{A-1}$, $\lambda_i \neq \lambda_j$ ·í $i \neq j$ ¦Ó¥B¥O $\overrightarrow{u}_i$ ¬°¹ïÀ³©ó $\lambda_i$ ¤§©T¦³¦V¶q¡C ©Ò¥H¥ô¦ó $\overrightarrow{x}$ $\in$ ${\bf R^A}$¡A¥i°ß¤@ªí¬°

\begin{displaymath}
\overrightarrow{x} = c_0 \overrightarrow{\psi}_0
+ c_1 \overrightarrow{u}_1
+ \cdots + c_{A-1} \overrightarrow{u}_{A-1}
\end{displaymath}

²{¥O P ¬° E0 ¤W¤§§ë¼vºâ¤l (Projection operator on E0)¡A§Y

\begin{displaymath}
P(\overrightarrow{x}) = c_0\overrightarrow{\psi}_0
\end{displaymath}

¥O $I(\overrightarrow{x}) = \overrightarrow{x}, \quad Q= I-P $
«h $Q(\overrightarrow{x}) = c_1 \overrightarrow{u}_1 + \cdots + c_{A-1} \overrightarrow{u}_{A-1} $

©Ò¥H¦pªG±N $\overrightarrow{V_0}$ ªí¬°

\begin{displaymath}
\overrightarrow{V}_0 = c_0 \overrightarrow{\psi}_0 + c_1 \overrightarrow{u}_1
+ \cdots + c_{A-1} \overrightarrow{u}_{A-1}
\end{displaymath}

«h

\begin{eqnarray*}
M^N \overrightarrow{V}_0 &=& I M^NI\overrightarrow{V}_0 \\
&=...
...\\
&& {}+ QM^NP\overrightarrow{V}_0 + QM^NQ\overrightarrow{V}_0
\end{eqnarray*}


±q©T¦³­È¡A©T¦³¦V¶q¤Î P,Q ¤§©Ê½è¡A¥i±o

\begin{eqnarray*}
PM^NP\overrightarrow{V}_0 &=& PM^N(c_0\overrightarrow{\psi}_0)...
...{u}_1 +\cdots + c_{A-1} \lambda^N_{A-1} \overrightarrow{u}_{A-1}
\end{eqnarray*}


©Ò¥H¡A

\begin{eqnarray*}
M^n \overrightarrow{V}_0 &=& c_0 \lambda^N_0 \overrightarrow{\...
...{A-1} \frac{\lambda_{A-1}}{\lambda_0}^N \overrightarrow{u}_{A-1}
\end{eqnarray*}


¦]¬°§Ú­Ì°²³] $\vert\lambda_i\vert < \lambda_0$¡Ai=1,¡K,A-1¡A ©Ò¥H·í N «Ü¤j®É

\begin{displaymath}
(8) \quad M^N \overrightarrow{V}_0 \approx \lambda^N_0 c_0 \overrightarrow{\psi}_0
\end{displaymath}

±q(8)¦¡¡A§Ú­Ì±o¨ì¨â­Óµ²½×
(I) ¦~ÄÖµ²ºc (Age distribution) ¬O¥H $\lambda_0$ ¤§³t²v¦¨ªø
(II)¨C­Ó¦~ÄÖ¤ÀÃþ (Age class) ¹ïÁ`¤H¤f¤§¤ñ¨Ò¬Oí©w¦aÁͦV¤@©T©w­È

¦]¬°¥Ñ(8)¦¡

\begin{eqnarray*}
\lefteqn{ \frac{\mbox{{\fontfamily{cwM0}\fontseries{m}\selectf...
...^A_{j=1}\psi_{0,j}}
= \frac{\psi_{0,k}}{\sum^A_{j=1}\psi_{0,j}}
\end{eqnarray*}


[1] Smale & Hirsch :¡mDifferential Equations,Dynamical System and Linear algebra¡n,Academic Press 1974.
[2] S, Karlin & H. M. Taylor :¡mA first course in Stochastic Process¡n,Academic Press 1975.
[3] K. Atkinson: ¡mAn introduction to Numerical Analysis¡n,1978. John Wiley son.
[4] F.C. Hoppensteadt: ¡mMathematical method of Population Biology¡n.Courant Institute Lecture Note 1976.

 
¹ï¥~·j´MÃöÁä¦r¡G
¡D¤H¤f°Ê¤O¾Ç
¡D©T¦³­È
¡D©T¦³¦V¶q
¡DPrimary decomposition Theorem
¡D¯S¼x¦h¶µ¦¡
¡DFrobenius
¡Dbasis
¡DProjection operator

¦^­¶­º
 
¡]­Y¦³«ü¥¿¡BºÃ°Ý¡K¡K¡A¥i¥H¦b¦¹ ¯d¨¥ ©Î ¼g«H µ¹§Ú­Ì¡C¡^
EpisteMath

EpisteMath (c) 2000 ¤¤¥¡¬ã¨s°|¼Æ¾Ç©Ò¡B¥x¤j¼Æ¾Ç¨t
¦Uºô­¶¤å³¹¤º®e¤§µÛ§@Åv¬°­ìµÛ§@¤H©Ò¦³


½s¿è¡G±d©ú°a ³Ì«á­×§ï¤é´Á¡G4/26/2002