Introducción al curso

IE3041 - Sistemas de Control 2

¿Por qué otro curso de control?

consideremos una aplicación contemporánea en control...

consideremos una aplicación contemporánea en control...

¿Cómo logramos esto con control clásico?

consideremos una aplicación contemporánea en control...

¿Cómo logramos esto con control clásico?

Resulta que esta no es la pregunta adecuada

¿Qué debe hacer cada servo actuador en las juntas del robot?

Capa de nivel bajo

\(\to\) Teoría de control clásica

¿Cómo coordina el robot el movimiento de sus patas y su cuerpo?

Capa de nivel medio

\(\to\) Teoría de control moderna

¿Cómo navega el robot en el entorno? Capa de nivel alto

  • Planificación de movimiento.
  • Optimización.
  • Inteligencia artificial.

En resumen

  1. El control clásico presenta limitantes (escalabilidad y computación).
  2. Herramientas adecuadas a los niveles de abstracción.
  3. Las aplicaciones contemporáneas requieren de métodos computacionales.

Un approach bottom-up:

  • Control clásico
  • Control moderno
    • Modelado y simulación.
    • Análisis.
    • Diseño de control.
  • Optimización
    • Fundamentos.
    • Generación de trayectorias y control.

¿Cómo lo haremos?

  • Desarrollo de competencias teóricas - 60%
    • Ejercicios en clase* (15%): enfocados en desarrollar destrezas básicas.
    • Casos de estudio (30%): aplicaciones interesantes, enfatizando la parte computacional.
    • Evaluaciones (15%): exámenes cortos* y course exit survey.
  • Laboratorio - 40%
    • Prácticas semanales: iniciando con la culminación de Control 1.

¿Control moderno?

¿De dónde surge?

Control clásico

 

dominio de frecuencia

teoría ingenieril

SISO

LTI

técnicas analíticas simples

Control moderno

 

dominio de tiempo

teoría matemática

MIMO

no lineal

herramientas numéricas

Control clásico

 

dominio de frecuencia

teoría ingenieril

SISO

LTI

técnicas analíticas simples

Sistemas dinámicos como modelo fundamental

Sistemas dinámicos como modelo fundamental

dispositivo, máquina, proceso o fenómeno que evoluciona a través del tiempo.

el concepto fundamental surge de las observaciones de Newton en su estudio de la mecánica de cuerpos celestes

t_1:
t_2:
x_1,y_1,\dot{x}_1,\dot{y}_1, \newline ...,\phi_1,\dot{\phi}_1
x_2,y_2,\dot{x}_2,\dot{y}_2, \newline ...,\phi_2,\dot{\phi}_2

"El movimiento futuro de los planetas puede predecirse únicamente con los valores actuales de cantidades fundamentales como la posición y la velocidad"

t_1:
t_2:
x_1,y_1,\dot{x}_1,\dot{y}_1, \newline ...,\phi_1,\dot{\phi}_1
x_2,y_2,\dot{x}_2,\dot{y}_2, \newline ...,\phi_2,\dot{\phi}_2

"El movimiento futuro de los planetas puede predecirse únicamente con los valores actuales de cantidades fundamentales como la posición y la velocidad"

t_1:
t_2:
x_1,y_1,\dot{x}_1,\dot{y}_1, \newline ...,\phi_1,\dot{\phi}_1
x_2,y_2,\dot{x}_2,\dot{y}_2, \newline ...,\phi_2,\dot{\phi}_2

en conjunto se denominan el estado del sistema

El estado de un sistema dinámico es una colección de variables o cantidades físicas que caracteriza completamente (en la ausencia de excitaciones externas) el comportamiento actual del sistema y puede usarse para predecir su comportamiento futuro

\begin{cases} \dot{\mathbf{x}}(t)=\mathbf{f}(\mathbf{x}(t)) \\ \mathbf{x}(t_0)=\mathbf{x}_0 \end{cases}

la intuición de Newton colapsa entonces al siguiente modelo "simple"

Sistema dinámico autónomo continuo en el tiempo

\begin{cases} \dot{\mathbf{x}}(t)=\mathbf{f}(\mathbf{x}(t)) \\ \mathbf{x}(t_0)=\mathbf{x}_0 \end{cases}

Sistema dinámico autónomo continuo en el tiempo

vector de estado

condiciones iniciales

dinámica

(campo vectorial)

\begin{cases} \dot{\mathbf{x}}(t)=\mathbf{f}(\mathbf{x}(t)) \\ \mathbf{x}(t_0)=\mathbf{x}_0 \end{cases}

Sistema dinámico autónomo continuo en el tiempo

\dot{\mathbf{x}}=\mathbf{f}(\mathbf{x})

Sistema de control no lineal en el espacio de estados

\dot{\mathbf{x}}=\mathbf{f}(\mathbf{x})
\begin{cases} \dot{\mathbf{x}}=\mathbf{f}(\mathbf{x}, \mathbf{u}) \\ \mathbf{x}(t_0)=\mathbf{x}_0 \\ \mathbf{y}=\mathbf{h}(\mathbf{x},\mathbf{u}) \end{cases}

Sistema de control no lineal en el espacio de estados

\dot{\mathbf{x}}=\mathbf{f}(\mathbf{x})
\begin{cases} \dot{\mathbf{x}}=\mathbf{f}(\mathbf{x}, \mathbf{u}) \\ \mathbf{x}(t_0)=\mathbf{x}_0 \\ \mathbf{y}=\mathbf{h}(\mathbf{x},\mathbf{u}) \end{cases}

esta notación simple esconde una gran complejidad

\mathbf{x}=\begin{bmatrix} x_1 \\ x_2 \\ x_3 \\ \vdots \\ x_n \end{bmatrix}
\mathbf{u}=\begin{bmatrix} u_1 \\ u_2 \\ \vdots \\ u_m \end{bmatrix}
\mathbf{y}=\begin{bmatrix} y_1 \\ y_2 \\ \vdots \\ y_p \end{bmatrix}
\mathbf{x}=\begin{bmatrix} x_1 \\ x_2 \\ x_3 \\ \vdots \\ x_n \end{bmatrix}
\mathbf{u}=\begin{bmatrix} u_1 \\ u_2 \\ \vdots \\ u_m \end{bmatrix}
\mathbf{y}=\begin{bmatrix} y_1 \\ y_2 \\ \vdots \\ y_p \end{bmatrix}

variables de estado

entradas o controles

salidas o mediciones

\mathbf{x}=\begin{bmatrix} x_1 \\ x_2 \\ x_3 \\ \vdots \\ x_n \end{bmatrix}
\mathbf{u}=\begin{bmatrix} u_1 \\ u_2 \\ \vdots \\ u_m \end{bmatrix}
\mathbf{y}=\begin{bmatrix} y_1 \\ y_2 \\ \vdots \\ y_p \end{bmatrix}

variables de estado

entradas o controles

salidas o mediciones

el espacio \(n\)-dimensional cuyos ejes están conformados por las variables de estado se denomina espacio de estados

\dot{\mathbf{x}} = \begin{bmatrix} \dot{x}_1 \\ \dot{x}_2 \\ \vdots \quad \\ \dot{x}_n \end{bmatrix} = \mathbf{f}(\mathbf{x},\mathbf{u}) = \begin{bmatrix} f_1(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \\ f_2(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \\ \vdots \\ f_n(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \end{bmatrix}
\dot{\mathbf{x}} = \begin{bmatrix} \dot{x}_1 \\ \dot{x}_2 \\ \vdots \quad \\ \dot{x}_n \end{bmatrix} = \mathbf{f}(\mathbf{x},\mathbf{u}) = \begin{bmatrix} f_1(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \\ f_2(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \\ \vdots \\ f_n(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \end{bmatrix}

la notación compacta esconde un sistema (vectorial) de ecuaciones diferenciales y algebráicas no lineales

\dot{\mathbf{x}} = \begin{bmatrix} \dot{x}_1 \\ \dot{x}_2 \\ \vdots \quad \\ \dot{x}_n \end{bmatrix} = \mathbf{f}(\mathbf{x},\mathbf{u}) = \begin{bmatrix} f_1(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \\ f_2(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \\ \vdots \\ f_n(x_1,x_2,...,x_n,u_1,u_2,...,u_m) \end{bmatrix}

la notación compacta esconde un sistema (vectorial) de ecuaciones diferenciales y algebráicas no lineales

¿Qué herramientas necesitamos para poder trabajar con estos sistemas?

Álgebra lineal y cálculo vectorial

Vectores

un vector \(\mathbf{v}\) es una lista ordenada de números reales (escalares) \(v_1,v_2,...,v_n\) apilados verticalmente de la forma:

\mathbf{v}=\begin{bmatrix} v_1 \\ v_2 \\ \vdots \\ v_n \end{bmatrix}=\begin{bmatrix} v_1 & v_2 & \cdots & v_n \end{bmatrix}^\top \in\mathbb{R}^n

puede emplearse la transpuesta para obtener un vector fila a partir de un vector columna.

gráficamente, puede representarse a un vector como una flecha dirigida desde el origen hasta el punto situado en las coordenadas dadas por los elementos del vector

\mathbf{v}=\begin{bmatrix} 1 \\ 2 \end{bmatrix}
\mathbf{s}=\begin{bmatrix} 3 & 4 & 5 \end{bmatrix}^\top

gráficamente, puede representarse a un vector como una flecha dirigida desde el origen hasta el punto situado en las coordenadas dadas por los elementos del vector

\mathbf{v}=\begin{bmatrix} 1 \\ 2 \end{bmatrix}
\mathbf{s}=\begin{bmatrix} 3 & 4 & 5 \end{bmatrix}^\top

sin embargo, puede considerarse también al vector como un objeto (el punto mismo) dentro de algún espacio \(n\)-dimensional

Escalamiento

se reparte el escalar, \(\alpha\in\mathbb{R}\), elemento por elemento

Escalamiento

se reparte el escalar, \(\alpha\in\mathbb{R}\), elemento por elemento

si \(\alpha>0\) entonces la operación sólo cambia el tamaño del vector, mientras que si \(\alpha<0\) también le cambia el sentido

Suma y resta

elemento por elemento, sólo está definida entre vectores de la misma dimensión

\mathbf{v}=\begin{bmatrix} 1 \\ 2 \end{bmatrix}, \qquad \mathbf{w}=\begin{bmatrix} -1 \\ 0 \\ 2 \end{bmatrix}, \qquad \mathbf{z}=\begin{bmatrix} 3 \\ 2 \end{bmatrix}
\mathbf{v}+\mathbf{z}=\begin{bmatrix} 1 \\ 2 \end{bmatrix}+\begin{bmatrix} 3 \\ 2 \end{bmatrix}=\begin{bmatrix} 4 \\ 4 \end{bmatrix}

claramente \(\mathbf{v}+\mathbf{w}\) no está definido, sin embargo

Producto interno

se define como la suma de los productos, elemento por elemento, de dos vectores \(\mathbf{v}, \mathbf{w} \in \mathbb{R}^n\) de la misma dimensión

\mathbf{v} \cdot \mathbf{w}\equiv \langle \mathbf{v}, \mathbf{w} \rangle\equiv \mathbf{v}^\top\mathbf{w}=\displaystyle\sum_{i=1}^{n} v_iw_i
\begin{bmatrix} 1 \\ 2 \\ 5\end{bmatrix} \cdot \begin{bmatrix} 3 \\ 4 \\ -1 \end{bmatrix}=(1)(3)+(2)(4)+(5)(-1)=6

Norma

el producto interno de un vector consigo mismo brinda una noción de magnitud denominada norma, la más común es la norma Euclideana (longitud del vector) y se denota como

\|\mathbf{v}\|_2 \equiv \|\mathbf{v}\|=\sqrt{\mathbf{v} \cdot \mathbf{v}}=\displaystyle \sqrt{\sum_{i=1}^{n}v_i^2}
\mathbf{x}=\begin{bmatrix} 1 & 2 & 3 \end{bmatrix}^\top, \quad \|\mathbf{x}\|=\sqrt{1^2+2^2+3^2}=\sqrt{14}

existen otras normas además de la Euclideana y se emplea el sub-índice para diferenciarlas entre sí

norma del valor absoluto:

\|\mathbf{v}\|_1 =\displaystyle \sum_{i=1}^{n} |v_i|
\|\mathbf{v}\|_\infty =\displaystyle \max_{1\le i \le n} |v_i|

norma uniforme (o del supremo):

Propiedades del producto interno

\(\mathbf{v} \cdot \mathbf{w}=\mathbf{w} \cdot \mathbf{v}\)

\(\mathbf{v} \cdot (\alpha\mathbf{w})=\alpha(\mathbf{v} \cdot \mathbf{w})\)

\(\mathbf{u} \cdot (\mathbf{v}+\mathbf{w})=\mathbf{u} \cdot \mathbf{v}+\mathbf{u} \cdot \mathbf{w}\)

\( \|\mathbf{v} + \mathbf{w}\|^2 = \|\mathbf{v}\|^2+2(\mathbf{v}\cdot\mathbf{w})+\|\mathbf{w}\|^2\)

\( \|\mathbf{v} + \mathbf{w}\|\le \|\mathbf{v}\|+\|\mathbf{w}\|\)

\(\mathbf{v} \cdot \mathbf{w}=\|\mathbf{v}\|\|\mathbf{w}\|\cos\theta\)

Ortogonalidad

si \(\mathbf{v}\in\mathbb{R}^n\) presenta \(\|\mathbf{v}\|=1\) entonces se conoce como vector unitario y puede denotarse como \(\hat{\mathbf{v}}\)

 

puede obtenerse un vector unitario a partir de cualquier vector mediante la normalización \(\hat{\mathbf{w}}=\mathbf{w}/\|\mathbf{w}\|\)

Ortogonalidad

si el producto interno entre dos vectores es cero, entonces se dice que estos son ortogonales

 

si, adicionalmente, dos vectores ortogonales tienen norma unitaria se dice que estos son ortonormales

ortogonales más no ortonormales

ortonormales

Matrices

se define una matriz como un arreglo rectangular de elementos con \(m\) filas y \(n\) columnas, estas pueden considerarse como una generalización para los vectores

\mathbf{A}=\begin{bmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \vdots & \vdots & \ddots & \vdots \\ a_{m1} & a_{m2} & \cdots & a_{mn} \end{bmatrix} \in\mathbb{R}^{m \times n}

para cada elemento: \(a_{ij}\in\mathbb{R}\)

 

si \(m=n\) se dice que la matriz es cuadrada

 

si ya sea \(n=1\) o \(m=1\) entonces se obtiene un vector columna o fila respectivamente

 

las operaciones de suma, resta y escalamiento funcionan igual para matrices que con vectores

la operación de transposición intercambia las filas con las columnas de la matriz

 

si \(\mathbf{A}\in\mathbb{R}^{m \times n}\) entonces  \(\mathbf{A}^\top\in\mathbb{R}^{n \times m}\) y recibe el nombre de la transpuesta de \(\mathbf{A}\)

\mathbf{A}=\begin{bmatrix} 1 & 2 & 3 \\ 4 & 5 & 6 \end{bmatrix} \in\mathbb{R}^{2 \times 3}
\mathbf{A}^\top=\begin{bmatrix} 1 & 4 \\ 2 & 5 \\ 3 & 6 \end{bmatrix} \in\mathbb{R}^{3\times 2}

Multiplicación de matrices

\mathbf{A}=\begin{bmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ a_{21} & a_{22} & \cdots & a_{2n} \\ \vdots & \vdots & \ddots & \vdots \\ a_{m1} & a_{m2} & \cdots & a_{mn} \end{bmatrix}\in\mathbb{R}^{m \times n}\equiv \begin{bmatrix} \mathbf{a}_1 \\ \mathbf{a}_2 \\ \vdots \\ \mathbf{a}_m \end{bmatrix}, \quad \mathbf{a}_i=\begin{bmatrix} a_{i1} & a_{i2} & \cdots & a_{in} \end{bmatrix}
\mathbf{B}=\begin{bmatrix} b_{11} & b_{12} & \cdots & b_{1p} \\ b_{21} & b_{22} & \cdots & b_{2p} \\ \vdots & \vdots & \ddots & \vdots \\ b_{n1} & b_{n2} & \cdots & b_{np} \end{bmatrix}\in\mathbb{R}^{n \times p}\equiv \begin{bmatrix} \mathbf{b}_1 & \mathbf{b}_2 & \cdots & \mathbf{b}_p \end{bmatrix}, \quad \mathbf{b}_j=\begin{bmatrix} b_{j1} \\ b_{j2} \\ \vdots \\ b_{jn} \end{bmatrix}

se define la multiplicación \(\mathbf{A}\mathbf{B}\) entre matrices como

\mathbf{AB}=\begin{bmatrix} \mathbf{a}_1 \cdot \mathbf{b}_1 & \mathbf{a}_1 \cdot \mathbf{b}_2 & \cdots & \mathbf{a}_1 \cdot \mathbf{b}_p \\ \mathbf{a}_2 \cdot \mathbf{b}_1 & \mathbf{a}_2 \cdot \mathbf{b}_2 & \cdots & \mathbf{a}_2 \cdot \mathbf{b}_p \\ \vdots & \vdots & \ddots & \vdots \\ \mathbf{a}_m \cdot \mathbf{b}_1 & \mathbf{a}_m \cdot \mathbf{b}_2 & \cdots & \mathbf{a}_m \cdot \mathbf{b}_p \end{bmatrix}\in\mathbb{R}^{m \times p}

es decir, se efectúa el producto interno de cada fila de \(\mathbf{A}\) con cada columna de \(\mathbf{B}\)

por lo tanto, para que la multiplicación entre matrices esté definida, el número de columnas de \(\mathbf{A}\) debe ser igual al número de filas de \(\mathbf{B}\)

 

de esto se concluye que la multiplicación entre matrices no es conmutativa

por ejemplo, si se tienen las matrices

\mathbf{C}=\begin{bmatrix} 3 & 1 & -2 \\ 4 & 0 & 6 \end{bmatrix}\in\mathbb{R}^{2 \times 3}, \qquad \mathbf{D}=\begin{bmatrix} 1 & 0 \\ -1 & 1 \end{bmatrix}\in\mathbb{R}^{2 \times 2}
\mathbf{D}\mathbf{C}

claramente \(\mathbf{C}\mathbf{D}\) no está definida, sin embargo

Inversa

"división entre matrices" pero bajo la definición de un recíproco o inverso multiplicativo, que cumple con

\mathbf{A}\mathbf{A}^{-1}=\mathbf{A}^{-1}\mathbf{A}=\mathbf{I}

donde \(\mathbf{A}^{-1}\) se denomina la inversa de \(\mathbf{A}\)

dado que la multiplicación por la inversa es conmutativa, se concluye que esta existe sólo para matrices cuadradas

Inversa

"división entre matrices" pero bajo la definición de un recíproco o inverso multiplicativo, que cumple con

\mathbf{A}\mathbf{A}^{-1}=\mathbf{A}^{-1}\mathbf{A}=\mathbf{I}

donde \(\mathbf{A}^{-1}\) se denomina la inversa de \(\mathbf{A}\)

sin embargo, no toda matriz cuadrada posee inversa y se refiere a ellas como singulares

en general, los procedimientos analíticos para encontrar la inversa de una matriz son tediosos y complicados, especialmente para matrices de dimensiones superiores

 

por esta razón, a menos que se trate de una matriz de \(2 \times 2\), se acudirá frecuentemente a métodos numéricos para determinarla

\mathbf{A}=\begin{bmatrix} a & b \\ c & d \end{bmatrix}, \qquad \mathbf{A}^{-1}=\dfrac{1}{\mathrm{det}(\mathbf{A})}\begin{bmatrix} d & -b \\ -c & a \end{bmatrix}

Matrices cuadradas de importancia

\mathbf{0}=\begin{bmatrix} 0 & 0 & \cdots & 0 \\ 0 & 0 & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \cdots & 0 \end{bmatrix}
\mathbf{I}=\begin{bmatrix} 1 & 0 & \cdots & 0 \\ 0 & 1 & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \cdots & 1 \end{bmatrix}

matriz cero

matriz identidad

matrices diagonales

matrices simétricas

\mathbf{A}=\begin{bmatrix} a_{11} & 0 & \cdots & 0 \\ 0 & a_{22} & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \cdots & a_{nn} \end{bmatrix}
\mathbf{A}^\top=\mathbf{A}

matrices antisimétricas

\mathbf{A}^\top=-\mathbf{A}

matrices ortogonales

\mathbf{A}^\top=\mathbf{A} \newline \mathbf{A}\mathbf{A}^\top=\mathbf{A}^\top\mathbf{A}=\mathbf{I} \newline \mathbf{A}^{-1}=\mathbf{A}^\top

matrices triangulares superiores

matrices triangulares inferiores

\mathbf{A}=\begin{bmatrix} a_{11} & a_{12} & \cdots & a_{1n} \\ 0 & a_{22} & \cdots & a_{2n} \\ \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & \cdots & a_{nn} \end{bmatrix}
\mathbf{A}=\begin{bmatrix} a_{11} & 0 & \cdots & 0 \\ a_{21} & a_{22} & \cdots & 0 \\ \vdots & \vdots & \ddots & \vdots \\ a_{n1} & a_{n2} & \cdots & a_{nn} \end{bmatrix}

Determinante

permite obtener un escalar a partir de una matriz cuadrada

 

se emplea en la solucion de sistemas de ecuaciones lineales, para definir una fórmula general para invertir matrices y para determinar el factor de escalamiento de una transformación lineal

no existe una fórmula general para el determinante ya que la complejidad de determinar el mismo depende de la dimensión de la matriz, sin embargo, existen las siguientes fórmulas

\mathbf{A}=\begin{bmatrix} a & b \\ c & d \end{bmatrix}
\mathrm{det}(\mathbf{A})\equiv|A|=ad-bc
\mathbf{B}=\begin{bmatrix} a & b & c \\ d & e & f \\ g & h & i \end{bmatrix}
\mathrm{det}(\mathbf{B})=aei+bfg+cdh-(gec+hfa+idb)

Algunas identidades

\(\mathbf{A}+\mathbf{B}=\mathbf{B}+\mathbf{A}\).

\(\mathbf{A}\mathbf{B}\ne \mathbf{B}\mathbf{A}\).

\(\mathbf{A}+(\mathbf{B}+\mathbf{C})=(\mathbf{A}+\mathbf{B})+\mathbf{C}\).

\(\mathbf{A}(\mathbf{B}\mathbf{C})=(\mathbf{A}\mathbf{B})\mathbf{C}\).

\((\mathbf{A}+\mathbf{B})^\top=\mathbf{A}^\top+\mathbf{B}^\top\).

\((\mathbf{A}\mathbf{B})^\top=\mathbf{B}^\top\mathbf{A}^\top\).

\((\mathbf{A}+\mathbf{B})^{-1}\ne \mathbf{A}^{-1}+\mathbf{B}^{-1}\).

\((\mathbf{A}\mathbf{B})^{-1}=\mathbf{B}^{-1}\mathbf{A}^{-1}\).

\(\left(\mathbf{A}^{-1}\right)^{-1}=\mathbf{A}\).

\(\mathrm{det}(\mathbf{A}\mathbf{B})=\mathrm{det}(\mathbf{A})\mathrm{det}(\mathbf{B})\).

\(\mathrm{det}\left(\mathbf{A}^\top\right)=\mathrm{det}(\mathbf{A})\).

\(\mathrm{det}\left(\mathbf{A}^{-1}\right)=1/\mathrm{det}(\mathbf{A})\).

Eigenvalores y eigenvectores

para una matriz \(\mathbf{A}\in\mathbb{R}^{n\times n}\) su polinomio característico está dado por

los eigenvalores de \(\mathbf{A}\) son las soluciones de la ecuación característica

\chi_\mathbf{A}(\lambda)=\det(\lambda\mathbf{I}-\mathbf{A})=\lambda^n+\alpha_{n-1}\lambda^{n-1}+\cdots+\alpha_0
\chi_\mathbf{A}(\lambda)=\det(\lambda\mathbf{I}-\mathbf{A})=0

son por lo general complejos y \(\mathbf{A}\) (en este caso) siempre tendrá \(n\) eigenvalores (puede que algunos se repitan)

se llama espectro de \(\mathbf{A}\) al conjunto formado por todos los eigenvalores de \(\mathbf{A}\) y se denota como

\sigma\left(\mathbf{A}\right)=\left\{ \lambda_1, \lambda_2, ..., \lambda_n\right\}

los eigenvalores resuelven el problema de eigenvalores en álgebra lineal

donde \(\mathbf{v}_i\in\mathbb{R}^n\) se conocen como eigenvectores y hay uno asociado a cada eigenvalor de \(\mathbf{A}\)

\mathbf{A}\mathbf{v}_i=\lambda_i\mathbf{v}_i

en otras palabras, los eigenvectores son aquellos a los cuales la aplicación de la matriz \(\mathbf{A}\) es equivalente a un escalamiento de magnitud igual a los eigenvalores

Cálculo vectorial y matricial

sea \(\mathbf{x}\in \mathbb{R}^n\), una función escalar multivariable está definida como \(f:\mathbb{R}^n \to \mathbb{R}\), tal que

f(\mathbf{x})=f(x_1,x_2,...,x_n)

donde \(x_i\) corresponde a los elementos de \(\mathbf{x}\)

se define el gradiente de la función escalar como el vector columna dado por

\nabla f(\mathbf{x})=\begin{bmatrix} \frac{\partial f(\mathbf{x})}{\partial x_1} \\ \frac{\partial f(\mathbf{x})}{\partial x_2} \\ \vdots \\ \frac{\partial f(\mathbf{x})}{\partial x_n} \end{bmatrix} \in\mathbb{R}^n

un campo vectorial (o función vectorial) está definido como \(\mathbf{f}:\mathbb{R}^n \to \mathbb{R}^m\), tal que

\mathbf{f}(\mathbf{x})=\mathbf{f}(x_1,x_2,...,x_n)= \begin{bmatrix} f_1(\mathbf{x}) \\ f_2(\mathbf{x}) \\ \vdots \\ f_m(\mathbf{x}) \end{bmatrix}= \begin{bmatrix} f_1(x_1,x_2,...,x_n) \\ f_2(x_1,x_2,...,x_n) \\ \vdots \\ f_m(x_1,x_2,...,x_n) \end{bmatrix}\in\mathbb{R}^m

en donde \(f_1,f_2,...,f_m\) son funciones escalares multivariable

se define el jacobiano del campo vectorial como la matriz

D\mathbf{f}(\mathbf{x})=\mathbf{f}'(\mathbf{x})=\dfrac{d \mathbf{f}(\mathbf{x})}{d \mathbf{x}}= \begin{bmatrix} \frac{\partial f_1(\mathbf{x})}{\partial x_1} & \cdots & \frac{\partial f_1(\mathbf{x})}{\partial x_n} \\ \vdots & \ddots & \vdots \\ \frac{\partial f_m(\mathbf{x})}{\partial x_1} & \cdots & \frac{\partial f_m(\mathbf{x})}{\partial x_n} \end{bmatrix}\in\mathbb{R}^{m \times n}

el jacobiano es, efectivamente, la derivada del campo vectorial

si \(\mathbf{A}(t)\) es una matriz que depende del tiempo

\mathbf{A}(t)=\begin{bmatrix} f_{11}(t) & \cdots & f_{1n}(t) \\ \vdots & \ddots & \vdots \\ f_{m1}(t) & \cdots & f_{mn}(t) \end{bmatrix}\in\mathbb{R}^{m \times n}

entonces se define

\dfrac{d \mathbf{A}(t)}{dt}=\begin{bmatrix} \frac{d f_{11}(t)}{dt} & \cdots & \frac{d f_{1n}(t)}{dt} \\ \vdots & \ddots & \vdots \\ \frac{d f_{m1}(t)}{dt} & \cdots & \frac{d f_{mn}(t)}{dt} \end{bmatrix}
\displaystyle \int \mathbf{A}(t) dt=\begin{bmatrix} \int f_{11}(t) dt & \cdots & \int f_{1n}(t) dt \\ \vdots & \ddots & \vdots \\ \int f_{m1}(t) dt & \cdots & \int f_{mn}(t) dt \end{bmatrix}

¿Muy complicado a mano?

MATLAB

\mathbf{x}=\begin{bmatrix} x_1 \\ x_2 \\ x_3 \end{bmatrix}=\begin{bmatrix} 1 \\ 2 \\ 3 \end{bmatrix}
\mathbf{v}=\begin{bmatrix} -1 & 0 & 1 & 2 \end{bmatrix}
\mathbf{w}=\begin{bmatrix} 0.5 & -8 & 1 \end{bmatrix}^\top
x = [1; 2; 3]
v = [-1, 0, 1, 2]
w = [0.5, -8, 1]'
x_2
x(2)
x(1:2)
\begin{bmatrix} x_1 \\ x_2 \end{bmatrix}
x_3
x(end)
\mathbf{x}+\mathbf{w}
x + w
\langle\mathbf{x},\mathbf{w}\rangle
x' * w
\|\mathbf{x}\|_2
norm(x, 2)
\mathbf{A}=\begin{bmatrix} a_{11} & a_{12} & a_{13} \\ a_{21} & a_{22} & a_{23} \\ a_{31} & a_{32} & a_{33} \end{bmatrix}= \begin{bmatrix} 1 & 0 & -1 \\ -2 & 3 & 0 \\ 0 & 0 & -0.5 \end{bmatrix}
A = [1, 0, -1; -2, 3, 0; 0, 0, -0.5]
a_{23}
A(2, 3)
\begin{bmatrix} a_{12} & a_{13} \\ a_{22} & a_{23} \\ a_{32} & a_{33} \end{bmatrix}
A(:, 2:end)
\mathbf{A}^\top
A'
\det(\mathbf{A})
det(A)
\mathbf{I}
I = eye(3)
\mathbf{0}
zeros(2, 5)
\mathbf{A}^{-1}
A^(-1)
I/A
inv(A)
\mathbf{A}\mathbf{I}
A * I
eig(A)
\sigma(\mathbf{A})=\left\{\lambda_1, \lambda_2, \cdots, \lambda_n \right\}
f(\mathbf{x})=f(x_1,x_2,x_3)=x_1^2+\cos(x_2)-e^{-x_3}
syms x1 x2 x3
f(x1, x2, x3) = x1^2 + cos(x2) - exp(-x3)
\nabla f(\mathbf{x})
g = gradient(f)
\mathbf{f}(\mathbf{x})=\begin{bmatrix} f_1(x_1,x_2,x_3) \\ f_2(x_1,x_2,x_3) \end{bmatrix}= \begin{bmatrix} 2x_1-\sin(x_2)-e^{-x_3} \\ x_1^2+\cos(x_2)-x_3 \end{bmatrix}
syms x1 x2 x3
f(x1, x2, x3) = [ 2*x1 - sin(x2) - exp(-x3); 
                  x1^2 + cos(x2) - x3       ]
D\mathbf{f}(\mathbf{x})=\dfrac{d\mathbf{f}(\mathbf{x})}{d\mathbf{x}}
Df = jacobian(f, [x1, x2, x3])
\mathbf{A}(t)=\begin{bmatrix} f_{11}(t) & f_{12}(t) \\ f_{21}(t) & f_{22}(t) \end{bmatrix}= \begin{bmatrix} t^2 & \cos(t) \\ e^{-t} & 1/t \end{bmatrix}
syms t
A(t) = [t^2, cos(t); exp(-t), 1 / t]
\dfrac{d\mathbf{A}(t)}{dt}
dAdt = diff(A)
\displaystyle\int \mathbf{A}(t)dt
intA = int(A)

IE3041 - Lecture 0 (2025)

By Miguel Enrique Zea Arenales

IE3041 - Lecture 0 (2025)

  • 137