Lie Groups for 2D and 3D Transformations
|
|
- Cecilia Chandler
- 6 years ago
- Views:
Transcription
1 Lie Groups for 2D and 3D Transformations Ethan Eade Updated May 20, 2017 * 1 Introduction This document derives useful formulae for working with the Lie groups that represent transformations in 2D and 3D space. A Lie group is a topological group that is also a smooth manifold, with some other nice properties. Associated with every Lie group is a Lie algebra, which is a vector space discussed below. Importantly, a Lie group and its Lie algebra are intimately related, allowing calculations in one to be mapped usefully into the other. This document does not give a rigorous introduction to Lie groups, nor does it discuss all of the mathematical details of Lie groups in general. It does attempt to provide enough information that the Lie groups representing spatial transformations can be employed usefully in robotics and computer vision. Here are the Lie groups that this document addresses: Group Description Dim. Matrix Representation SO3 3D Rotations 3 3D rotation matrix SE3 3D Rigid transformations 6 Linear transformation on homogeneous 4-vectors SO2 2D Rotations 1 2D rotation matrix SE2 2D Rigid transformations 3 Linear transformation on homogeneous 3-vectors Sim3 3D Similarity transformations Linear transformation on 7 rigid motion + scale homogeneous 4-vectors For each of these groups, the representation is described, and the exponential map and adjoint are derived. 1.1 Why use Lie groups for robotics or computer vision? Many problems in robotics and computer vision involve manipulation and estimation in the 3D geometry. Without a coherent and robust framework for representing and working with 3D transformations, * Added Ÿ2.4.2 to clarify the notation for dierentiation of/by group element perturbations. 1
2 these tasks are onerous and treacherous. Transformations must be composed, inverted, dierentiated and interpolated. Lie groups and their associated machinery address all of these operations, and do so in a principled way, so that once intuition is developed, it can be followed with condence. 1.2 Lie algebras and other general properties Every Lie group has an associated Lie algebra, which is the tangent space around the identity element of the group. That is, the Lie algebra is a vector space generated by dierentiating the group transformations along chosen directions in the space, at the identity transformation. The tangent space has the same structure at all group elements, though tangent vectors undergo a coordinate transformation when moved from one tangent space to another. The basis elements of the Lie algebra and thus of the tangent space are called generators in this document. All tangent vectors represent linear combinations of the generators. Importantly, the tangent space associated with a Lie group provides an optimal space in which to represent dierential quantities related to the group. For instance, velocities, Jacobians, and covariances of transformations are well-represented in the tangent space around a transformation. This is the optimal space in which to represent dierential quantities because ˆ The tangent space is a vector space with the same dimension as the number of degrees of freedom of the group transformations ˆ The exponential map converts any element of the tangent space exactly into a transformation in the group ˆ The adjoint linearly and exactly transforms tangent vectors from one tangent space to another The adjoint property is what ensures that the tangent space has the same structure at all points on the manifold, because a tangent vector can always be transormed back to the tangent space around the identity. Each Lie group described below also has a group action on 3D space. For instance, 3D rigid transformations have the action of rotating and translating points. The matrix representations given below make these actions explicit. 2 SO3: Rotations in 3D space 2.1 Representation Elements of the 3D rotation group, SO3, are represented by 3D rotation matrices. Composition and inversion in the group correspond to matrix multiplication and inversion. Because rotation matrices are orthogonal, inversion is equivalent to transposition. R SO3 1 R 1 R T 2 2
3 The Lie algebra, so3, is the set of 3 3 skew-symmetric matrices. The generators of so3 correspond to the derivatives of rotation around the each of the standard axes, evaluated at the identity: G , G , G An element of so3 is then represented as a linear combination of the generators: 3 ω R 3 4 ω 1 G 1 + ω 2 G 2 + ω 3 G 3 so3 5 We will simply write ω so3 as a 3-vector of the coecients, and use ω to represent the corresponding skew symmetric matrix. 2.2 Exponential Map The exponential map that takes skew symmetric matrices to rotation matrices is simply the matrix exponential over a linear combination of the generators: exp ω exp 0 ω 3 ω 2 ω 3 0 ω 1 ω 2 ω I + ω + 1 2! ω ! ω3 + 7 Writing the terms in pairs, we have: exp ω I + [ ] ω 2i+1 2i + 1! + ω2i+2 2i + 2! i0 8 Now we can take advantage of a property of skew-symmetric matrices: First extend this identity to the general case: ω 3 ω T ω ω 9 θ 2 ω T ω 10 ω 2i+1 1 i θ 2i ω 11 ω 2i+2 1 i θ 2i ω 2 12 Now we can factor the exponential map series and recognize the Taylor expansions in the coecients: 3
4 exp ω I + i0 1 i θ 2i ω + 2i + 1! i0 I + 1 θ2 3! + θ4 5! + ω + sin θ 1 cos θ I + ω + θ θ 2 1 i θ 2i ω i + 2! 1 2! θ2 4! + θ4 6! + ω 2 14 ω 2 15 Equation 15 is the familiar Rodrigues formula. The exponential map yields a rotation by θ radians around the axis given by ω. Practical implementation of the Rodrigues formula should use the Taylor expansions of the coecients of the second and third terms when θ is small. The exponential map can be inverted to give the logarithm, going from SO3 to so3: R SO3 16 trr 1 θ arccos 17 2 θ ln R 2 sin θ R R T 18 The vector ω is then taken from the o-diagonal elements of ln R. Again, the Taylor expansion of θ the coecient should be used when θ is small. 2 sin θ 2.3 Adjoint In Lie groups, it is often necessary to transform a tangent vector from the tangent space around one element to the tangent space of another. The adjoint performs this transformation. One very nice property of Lie groups in general is that this transformation is linear. For an element X of a Lie group, the adjoint is written Adj X : ω so3, R SO3 19 R exp ω exp Adj R ω R 20 The adjoint can be computed from the generators of the Lie algebra: exp Adj R ω R exp ω R Adj R ω R ω i G i R 1 22 i1 R ω R 1 23 Rω 24 Adj R R 25 4
5 In the case of SO3, the adjoint transformation for an element is particularly simple: it is the same rotation matrix used to represent the element. Rotating a tangent vector by an element moves it from the tangent space on the right side of the element to the tangent space on the left. 2.4 Jacobians Dierentiating the action of SO3 on R 3 Consider R SO3 and x R 3. The rotation of vector x by matrix R is given by multiplication: y fr, x R x 26 Then dierentiation by the vector is straightforward, as f is linear in x: y x R 27 Dierentiation by the rotation parameters is performed by implicitly left multiplying the rotation by the exponential of a tangent vector and dierentiating the resulting expression around the zero perturbation. This is equivalent to left multiplying the product by the generators. y R ω ω0 exp ω R x 28 ω ω0 exp ω R x 29 ω ω0 exp ω y 30 G 1 y G 2 y G 3 y 31 y Dierentiating a group-valued function by an argument in the group Consider a Lie group G and a function f : G G. Neither the domain nor the range is a vector space, but by introducing tangent space perturbations on the argument and result, we can use the dierentation notation as a shorthand for the mapping from input to output perturbations: exp ɛ f g f exp δ g 33 f ɛ g δ δ0 34 Solving Eq. 33 for ɛ and dierentiating yields an explicit formula for the dierential of the output perturbation ɛ by the input perturbation δ: 5
6 ɛ log f exp δ g f g 1 35 f exp δ g f g 1 f g log δ δ0 36 Eq. 36 produces a linear mapping from left-tangent-space perturbations of the argument to lefttangent-space perturbations of the result. As expected, applying this dierentiation shorthand to the identity function f g g yields the identity matrix. For a nontrivial example application of this procedure, consider a product of elements in G SO3 by the second factor R 0 : R 2 f R 0 R 1 R 0 37 First, the input and output perturbations in the tangent space so3 are made explicit. exp ɛ R 2 R 1 exp ω R 0 38 Dierentation of ɛ by the input perturbation ω is performed around ω 0. The adjoint is employed to shift the tangent vector to the left side of the expression. The remainder of the expression cancels and the result is simple. R 2 R 0 log R 1 exp ω R 0 R 1 R 0 1 ω0 39 ω [ exp ω ω0 log ω AdjR1 R 1 R 0 R1 R 0 1] 40 [ log exp AdjR1 ω ] 41 ω ω0 ω [ ω0 AdjR1 ω ] 42 Adj R1 43 R Gaussians in SO Sampling We can encode Gaussian distributions over 3D rotations by representing the mean with an element of SO3 and the covariance as a quadratic form over tangent vectors in so3. More precisely, consider a Gaussian distribution given by mean R SO3 and covariance Σ R 3 3. We can draw a sample rotation S from the distribution by sampling the zero-mean distribution in the tangent space and left multiplying the mean: 6
7 ɛ N 0, Σ 45 S exp ɛ R Composition of uncertain rotations Given two Gaussian distributions on rotation, we can compose the two uncertain transformations using the adjoint. Let one mean-covariance pair be R 0, Σ 0 and the other be R 1, Σ 1. Then the distribution of rotations by rst transforming by R 0 and then by R 1 is given by: R 1, Σ 1 R 0, Σ 0 R 1 R 0, Σ 1 + R 1 Σ 0 R T Bayesian combination of rotation estimates The information from the two Gaussians can be combined in a Bayesian manner to yield R c, Σ c by rst nding the deviation between the two means in the tangent space, and then weighting by the information of the two estimates. The information inverse covariance adds, as usual: 2.6 Extended Kalman Filtering in SO3 Σ c Σ Σ Σ 0 Σ 0 Σ 0 + Σ 1 1 Σ 0 49 v R 1 R 0 50 ln R 1 R R c exp Σ c Σ 1 1 v R 0 52 Equation 47 could be used as the dynamics update in an extended Kalman lter EKF, where R 0, Σ 0 is the prior state and R 1, Σ 1 is the dynamic model. Note that Equation 49 is actually the EKF measurement update for the covariance and Equation 52 is the measurement update for the mean, assuming a trivial measurement Jacobian identity matrix. The tangent vector v is the innovation. In this case of trivial measurement Jacobian, the Kalman gain K is dened so that the Kalman update can be written in its standard form: K Σ 0 Σ 0 + Σ R c R 0 K v 54 exp K v R 0 55 Σ c I K Σ
8 Labelling the above in the standard EKF framework, the state covariance is given by Σ 0 and the measurement noise is given by Σ 1. Note that Eq. 55 is mathematically identical to Eq. 52, and Eq. 56 is identical to Eq. 49. The case of non-trivial measurement or dynamics Jacobians is a simple modication of the equations given here. 3 SE3: Rigid transformations in 3D space 3.1 Representation The group of rigid transformations in 3D space, SE3, is well represented by linear transformations on homogeneous four-vectors: R SO3, t R 3 57 R t C SE3 58 Note that, in an implementation, only R and t need to be stored. The remaining matrix structure can be implicitly imposed. This representation, as in SO3, means that transformation composition and inversion are coincident with matrix multiplication and inversion: C 1, C 2 SE3 59 R1 t C 1 C 2 1 R2 t 2 60 R1 R 2 R 1 t 2 + t 1 61 R C1 1 T 1 R T 1 t 62 The matrix representation also makes the group action on 3D points and vectors clear: x x y z w T RP 3 λx x λ R R t C x x 63 T R x y z + wt 64 w 8
9 Typically, w 1, so that x is a Cartesian point. The action by matrix-vector multiplication corresponds to rst rotating x and then translating it. For direction vectors, encoded with w 0, translation is ignored. The Lie algebra se3 is the set of 4 4 matrices corresponding to dierential translations and rotations as in so3. There are thus six generators of the algebra: G 1 G , G 2, G , G 3, G , 65 An element of se3 is then represented by multiples of the generators: u ω T R 6 66 u 1 G 1 + u 2 G 2 + u 3 G 3 + ω 1 G 4 + ω 2 G 5 + ω 3 G 6 se3 67 For convenience, we write u ω T se3, with multiplication against the generators implied. 3.2 Exponential Map The exponential map from se3 to SE3 is the matrix exponential on a linear combination of the generators: δ u ω se3 68 ω u exp δ exp ω u I ω 2 ω u + 1 ω 3 ω 2 u ! 0 0 3! 0 0 The rotation block is the same as for SO3, but the translation component is a dierent power series: ω u exp 0 0 exp ω Vu 71 V I + 1 2! ω + 1 3! ω
10 Again using the identity from Eq. 9, we split the terms by odd and even powers, and factor out : [ ] ω 2i+1 V I + 2i + 2! + ω2i i + 3! i0 1 i θ 2i 1 i θ 2i I + ω + ω i + 2! 2i + 3! i0 The coecients can be identied with Taylor expansions, yielding a formula for V: i0 1 V I + 2! θ2 4! + θ4 1 cos θ I + θ 2 6! + ω + 1 ω + θ sin θ Thus the exponential map has a closed-form representation: θ 3 7! + 3! θ2 5! + θ4 ω 2 75 ω u, ω R 3 77 θ ω T ω 78 A sin θ θ 79 B 1 cos θ θ 2 80 C 1 A θ 2 81 R I + Aω + Bω 2 82 V I + Bω + Cω 2 83 u R Vu exp 84 ω For implementation purposes, Taylor expansions of A, B, and C should be used when θ 2 is small. The matrix V has a closed-form inverse: V 1 I 1 2 ω + 1 θ 2 1 A ω B The ln function on SE3 can be implemented by rst nding lnr as shown in Eq. computing u V 1 t. 18, then 10
11 3.3 Adjoint The adjoint in SE3 is computed from the generators, just as in SO3: δ u ω T se3, C R t C exp δ exp Adj C δ C SE3 86 exp Adj C δ C exp δ C Adj C δ C δ i G i C 1 88 Adj C i1 Ru + t Rω 89 Rω R t R R R Note that moving a tangent vector via the adjoint mixes the rotation component into the translation component. 3.4 Jacobians Consider C multiplication: R t SE3 and x R 3. The transformation of vector x by C is given by y fc, x R t x 1 91 R x + t 92 Then dierentiation by the vector is straightforward, as f is linear in x: y x R 93 Just as with SO3, dierentiation by the transformation parameters is performed by left multiplying the product by the generators here with their last rows removed: y C G 1 y G 6 y I y 94 Again, dierentiation of a product of transformations is trivial given the adjoint: 11
12 C C 1 C 0 95 C C 0 δ [C 1 exp δ C 0 ] 96 Adj C SO2: Rotations in 2D space Having treated SO3, the 2D equivalent SO3 is straightforward. 4.1 Representation Elements of the rotation group in two dimensions, SO2, are represented by 2D rotation matrices. Composition and inversion in the group correspond to matrix multiplication and inversion. Because rotation matrices are orthogonal, inversion is equivalent to transposition. R SO2 98 R 1 R T 99 The Lie algebra, so2, is the set of 2 2 skew-symmetric matrices. The single generator of so2 corresponds to the derivative of 2D rotation, evaluated at the identity: G An element of so2 is then any scalar multiple of the generator: θ R 101 θg so2 102 We will simply write θ so2, and use θ to represent the skew symmetric matrix θg. 4.2 Exponential Map The exponential map that takes skew symmetric matrices to rotation matrices is simply the matrix exponential over a linear combination of the generators: 12
13 0 θ exp θ exp θ I + θ + 1 2! θ ! θ θ I θ 2 0 θ 0 2! 0 θ θ 3 3! θ The resulting elements form the Taylor series expansion of sin θ and cos θ: cos θ sin θ exp θ sin θ cos θ SO2 106 Thus the exponential map yields a rotation by θ radians. The exponential map can be inverted, going from SO2 to so2: R SO2 107 ln R θ arctan R 21, R Adjoint Because rotations in the plane commute, the adjoint of SO2 is the identity function. 5 SE2: Rigid transformations in 2D space The group SE2 is the lower-dimensional analogue of SE3. The group has three dimensions, corresponding to translation and rotation in the plane. 5.1 Representation The group of rigid transformations in 2D space, SE2, is represented by linear transformations on homogeneous three-vectors: R SO2, t R R t C SE2 110 Note that, in an implementation, only R and t need to be stored. The remaining matrix structure can remain implicit. 13
14 Transformation composition and inversion are coincident with matrix multiplication and inversion: C 1, C 2 SE2 111 R1 t C 1 C 2 1 R2 t R1 R 2 R 1 t 2 + t R C1 1 T 1 R T 1 t 114 The matrix representation also makes the group action on 2D points and vectors explicit: x x y w T RP 2 λx x λ R R t C x x 115 T R x y + wt 116 w Typically, w 1, so that x is a Cartesian point. The action by matrix-vector multiplication corresponds to rst rotating x and then translating it. For direction vectors, encoded with w 0, translation is ignored. The Lie algebra se2 is the set of 3 3 matrices corresponding to dierential translations and rotation around the identity. There are thus three generators of the algebra: G , G , G An element of se2 is then represented by linear combinations of the generators: 117 u1 u 2 θ T R u 1 G 1 + u 2 G 2 + θg 3 se2 119 For convenience, we write u θ T se2, with multiplication against the generators implied. 5.2 Exponential Map As for all Lie groups in this document, the exponential map from se2 to SE2 is the matrix exponential on a linear combination of the generators: 14
15 δ u θ se2 120 θ u exp δ exp θ u I θ 2 θ u + 1 θ 3 θ u ! 0 0 3! 0 0 The rotation block is the same as for SO2, but the translation component is a dierent power series: θ u exp 0 0 exp θ Vu 123 V I + 1 2! θ + 1 3! θ We split the terms by odd and even powers: V [ ] θ 2i 2i + 1! + θ2i+1 2i + 2! i0 125 Two identities easily conrmed by induction are useful for collapsing the series: θ 2i 1 i θ 2i 1 0 θ 2i+1 1 i θ 2i Direct application of the identies yields a reduced expression for V in terms of diagonal and skewsymmetric components: V [ 1 i θ 2i i + 1! + i0 1 i θ 2i i + 1! i0 i0 θ 0 1 2i + 2! i θ 2i+1 2i + 2! ] The coecients can be identied with Taylor expansions: 15
16 V 1 θ2 3! + θ4 5! + sin θ θ sin θ 1 cos θ 1 cos θ sin θ 1 θ cos θ θ θ + 2! θ3 4! + θ ! For implementation purposes, Taylor expansions should be used for V when θ is small. The ln function on SE2 can be implemented by rst recovering θ lnr as shown in Eq. 108, then solving Vu t for u in closed form: 5.3 Adjoint R t ln A sin θ θ B 1 cos θ θ V 1 1 The adjoint in SE2 is computed from the generators: A B 135 A 2 + B 2 B A V 1 t se2 136 θ δ u θ T R t se2, C 3 Adj C δ C δ i G i Adj C i1 Ru + θ θ R t 2 t 1 SE2 137 C t2 t R Note that moving a tangent vector via the adjoint mixes the rotation component into the translation component. 16
17 6 Sim3: Similarity Transformations in 3D space 6.1 Representation Similarity transformations are combinations of rigid transformation and scaling. The group of similarity transforms in 3D space, Sim3, has a nearly identical representation to SE3, with an additional scale factor: R SO3, t R 3, s R 141 R t T 0 s 1 Sim3 142 Again, group operations map are isomorphic with matrix operations: T 1, T 2 Sim3 143 R1 t T 1 T 2 1 R2 t 0 s s R1 R 2 R 1 t 2 + s 1 2 t 1 0 s 1 s R T1 1 T 1 s 1 R T 1 t s 1 The group action on 3D points also encodes scaling by s: x x y z w T RP 3 λx x λ R R t T x 0 s 1 x 147 T R x y z + wt s w s R x y z T + wt 149 w In the typical case with w 1, this corresponds to rigid transformation followed by scaling. The generators of the Lie algebra sim3 are identical to those of se3 Eq. 65, with the addition of a generator corresponding to scale change: G
18 An element of sim3 is represented by multiples of the generators: u ω λ T R u 1 G 1 + u 2 G 2 + u 3 G 3 + ω 1 G 4 + ω 2 G 5 + ω 3 G 6 + λg 7 sim3 152 For convenience, we write u ω λ T sim3, with multiplication against the generators implied. 6.2 Exponential Map As above, the exponential map from sim3 to Sim3 is the matrix exponential on a linear combination of the generators: δ u ω λ sim3 ω u exp δ exp 0 λ ω u I + 0 λ + 1 2! ω 2 ω u λu 0 λ ω 3 ω 2 u λω u + λ 2 u 3! 0 λ The series is similar to that for se3, but now rotation, translation and scale are being interleaved innitesimally. The rotation and scale components of the exponential are immediately clear, but the translation component involves the mixing of the three. We can write out the series for the translation multiplier: ω u exp 0 0 V exp ω Vu 0 exp λ n ω n k λ k n + 1! n0 k0 k0 nk k0 j0 ω n k λ k n + 1! ω j λ k j + k + 1! Again letting θ 2 ω T ω, and using the identity from Eq. 9, we partition the terms into odd and even powers of ω, and factor: V λ k I + λ k k + 1! k0 k0 λ k I + k + 1! k0 k0 i0 i0 [ ω 2i+1 2i + k + 2! + ω 2i+2 1 i θ 2i λ k 2i + k + 2! 2i + k + 3! ω + k0 i0 ] 1 i θ 2i λ k 2i + k + 3! 160 ω
19 The rst coecient is easily identied as a Taylor series, leaving the other two coecients to be analyzed: Consider coecient B: V AI + Bω + Cω A B C 1 exp λ λ 1 i θ 2i λ k 2i + k + 2! k0 i0 1 i θ 2i λ k 2i + k + 3! k0 i B k0 i0 [ 1 i θ 2i i0 i0 i0 i0 i0 1 i θ 2i λ k 2i + k + 2! [ 1 i θ 2i λ 2i [ 1 i θ 2i λ 2i 1 i θ 2i λ 2i 1 i θ 2i λ 2i k0 k0 λ k 2i + k + 2! λ 2i+k 2i + k + 2! λ m ] m + 2! [ m2i m0 [ m0 ] ] λ m 2i 1 m + 2! λ m m + 2! ] m0 ] λ m m + 2! 1 i θ 2i i0 λ 2i 2i 1 m0 λ m m + 2! Manipulating the second term yields a more useful form: 19
20 L Relabel the factors: i0 B L L 1 i θ 2i i0 i0 λ 2i [ m0 1 i θ 2i λ 2i 1 i θ 2i λ 2i ] λ m m + 2! 2i 1 λ m m + 2! m0 [ i 1 p0 λ 2p 2p + 2! + λ2p+1 2p + 3! ] i θ 2i i 1 [ ] L λ 2i λ 2p 1 2p + 2! λ 175 2p + 3! i0 p0 1 i θ 2i [ ] L λ 2i λ 2p 1 2p + 2! λ 176 2p + 3! p<i 1 i θ 2i [ ] 1 L λ 2i p 2p + 2! λ 177 2p + 3! p<i 1 q θ 2q [ 1 p θ 2p 1 p L λ 2q 2p + 2! λ θ 2p ] 178 2p + 3! p0 q1 1 q θ 2q [ 1 p θ 2p 1 p L λ 2q 2p + 2! λ θ 2p ] 179 2p + 3! q1 p0 1 i θ 2i λ m i0 λ 2i m0 180 m + 2! 1 q θ 2q [ 1 p θ 2p 1 p q0 λ 2q 1 p0 2p + 2! λ θ 2p ] 2p + 3! B α β α 1 γ 181 α β γ α β γ + γ i θ 2i λ 2i i0 183 λ m 184 m + 2! m0 [ 1 p θ 2p 1 p 2p + 2! λ θ 2p ] 185 2p + 3! p0 20
21 The factors are then recognised as Taylor series: λ 2 α 186 λ 2 + θ 2 exp λ 1 + λ β 187 λ 2 γ 1 cos θ θ sin θ θ 2 λ 188 θ 3 By similar algebra, a formula for coecient C can be derived from Eq. 165: C α µ ν + ν 189 µ 1 λ λ2 exp λ 190 λ 2 ν θ sin θ cos θ 1 + θ 2 2 θ 3 λ 191 θ 4 Combining these results gives a closed-form exponential map for sim3: 21
22 u ω λ T sim3 192 exp u ω λ θ 2 ω T ω 193 X sin θ θ 194 Y 1 cos θ θ Z 1 X θ W 1 2 Y θ λ 2 α λ 2 + θ β exp λ 1 + λ λ γ Y λz 200 µ 1 λ λ2 exp λ λ ν Z λw 202 A 1 exp λ λ 203 B α β γ + γ 204 C α µ ν + ν 205 R I + aω + bω V AI + Bω + Cω R Vu exp λ Again, Taylor expansions should be used when λ 2 or θ 2 is small. The ln function can be implemented by rst recovering ω and λ, constructing V, and then solving for u as in the SE3 case. 6.3 Adjoint The adjoint is computed from a linear combination of the generators: 22
23 δ u ω λ T sim3, T T exp δ exp Adj T δ T R t 0 s 1 Sim3 209 exp Adj T δ T exp δ T Adj T δ T δ i G i T Adj T i1 s Ru + t Rω st Rω λ sr st R st 0 R R Interpolation Consider a Lie group G, with two elements a, b G. We would like to interpolate between these elements, according to a parameter t [0, 1]. Dene a function that will perform the interpolation: f : G G R G 214 f a, b, 0 a 215 f a, b, 1 b 216 The function is dened by transforming the interpolation operation into the tangent space, performing the linear combination there, and then transforming the resulting tangent vector back onto the manifold. First consider the group element that takes a to b: d b a 1 G 217 d a b 218 Now compute the corresponding Lie algebra vector and scale it in the tangent space: dt t ln d 219 Then transform back into the manifold using the exponential map, yielding a partial transformation: Combining these three steps gives a denition for f: d t exp dt
24 fa, b, t d t a 221 exp t ln b a 1 a 222 Note that the result is always on the manifold, due to the properties of the exponential map. No projection or coercion is required. Furthermore, the linear transformation in the tangent space corresponds to moving along a geodesic of the manifold. So the interpolation always moves along the shortest transformation in the Lie group. 8 Uncertain Transformations 8.1 Sampling Consider a Lie group G and its associated Lie algebra vector space g, with k degrees of freedom. We wish to represent Gaussian distributions over transformations in this group. Each such distribution has a mean transformation, µ G, and a covariance matrix Σ R k k. The algebra corresponds to tangent vectors around the identity element of the group. Thus, it is natural to express a sample x from the desired distribution in terms of a sample δ drawn from a zero-mean Gaussian and the mean transformation µ: δ N 0; Σ 223 x exp δ µ Transforming This formulation allows convenient transformations of distributions through other group elements using the adjoint. Consider x, y G, and the distribution x, Σ with mean x. Consider y x, where x is a sample drawn from x, Σ: x exp δ x 225 y x y exp δ x 226 exp Adj y δ y x 227 By the denition of covariance, [ ] Σ E δ δ T 228 Given a linear transformation L, by linearity of expectation we have: [ ] [ ] E Lδ Lδ T E L δ δ T L T 229 [ ] L E δ δ T L T 230 L Σ L T
25 So we can express the parameters of the transformed distribution: y x N y x; Adj y Σ Adj T y 232 Thus the mean of the transformed distribution is the transformed mean, and the covariance is mapped linearly by the adjoint. 8.3 Distribution of Inverse Similarly, the distribution of the inverse transformation is easily computed: z x x 1 exp δ 234 exp Adj x 1δ x z N x 1 ; Adj x 1 Σ Adj T x Projection through Group Actions Because the covariance is expressed in terms of tangent vectors exponentiated and multiplied on the left-hand side, all the Jacobians given in the group descriptions above can be used to transform covariance or information matrices through arbitrary mappings. 8.5 Computing Moments from Samples Given a Lie group G with algebra g and a set of N samples x i G, we can estimate the mean and covariance iteratively. First, any of the samples provides an initial guess for the mean: µ 0 x Then, new estimates of the mean and covariance can be computed in terms of deviations in the tangent space: v i,k ln x i µ 1 k 238 Σ k 1 [ vi,k v N i,k] T 239 i 1 µ k+1 exp v i,k µ k 240 N Iterating these updates should rapidly converge typically in two or three iterations. Replacing the factor 1 N with 1 in Eq. 239 yields an unbiased estimate of the covariance. N 1 i 25
Differential of the Exponential Map
Differential of the Exponential Map Ethan Eade May 20, 207 Introduction This document computes ɛ0 log x + ɛ x ɛ where and log are the onential mapping and its inverse in a Lie group, and x and ɛ are elements
More informationVisual SLAM Tutorial: Bundle Adjustment
Visual SLAM Tutorial: Bundle Adjustment Frank Dellaert June 27, 2014 1 Minimizing Re-projection Error in Two Views In a two-view setting, we are interested in finding the most likely camera poses T1 w
More informationLinear Regression and Its Applications
Linear Regression and Its Applications Predrag Radivojac October 13, 2014 Given a data set D = {(x i, y i )} n the objective is to learn the relationship between features and the target. We usually start
More informationMethods in Computer Vision: Introduction to Matrix Lie Groups
Methods in Computer Vision: Introduction to Matrix Lie Groups Oren Freifeld Computer Science, Ben-Gurion University June 14, 2017 June 14, 2017 1 / 46 Definition and Basic Properties Definition (Matrix
More informationRigid Geometric Transformations
Rigid Geometric Transformations Carlo Tomasi This note is a quick refresher of the geometry of rigid transformations in three-dimensional space, expressed in Cartesian coordinates. 1 Cartesian Coordinates
More informationLinear Algebra. Min Yan
Linear Algebra Min Yan January 2, 2018 2 Contents 1 Vector Space 7 1.1 Definition................................. 7 1.1.1 Axioms of Vector Space..................... 7 1.1.2 Consequence of Axiom......................
More informationLinear Algebra (part 1) : Matrices and Systems of Linear Equations (by Evan Dummit, 2016, v. 2.02)
Linear Algebra (part ) : Matrices and Systems of Linear Equations (by Evan Dummit, 206, v 202) Contents 2 Matrices and Systems of Linear Equations 2 Systems of Linear Equations 2 Elimination, Matrix Formulation
More informationMATRIX ALGEBRA AND SYSTEMS OF EQUATIONS. + + x 1 x 2. x n 8 (4) 3 4 2
MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS SYSTEMS OF EQUATIONS AND MATRICES Representation of a linear system The general system of m equations in n unknowns can be written a x + a 2 x 2 + + a n x n b a
More informationKalman Filter. Predict: Update: x k k 1 = F k x k 1 k 1 + B k u k P k k 1 = F k P k 1 k 1 F T k + Q
Kalman Filter Kalman Filter Predict: x k k 1 = F k x k 1 k 1 + B k u k P k k 1 = F k P k 1 k 1 F T k + Q Update: K = P k k 1 Hk T (H k P k k 1 Hk T + R) 1 x k k = x k k 1 + K(z k H k x k k 1 ) P k k =(I
More informationMath 302 Outcome Statements Winter 2013
Math 302 Outcome Statements Winter 2013 1 Rectangular Space Coordinates; Vectors in the Three-Dimensional Space (a) Cartesian coordinates of a point (b) sphere (c) symmetry about a point, a line, and a
More informationMATH 23a, FALL 2002 THEORETICAL LINEAR ALGEBRA AND MULTIVARIABLE CALCULUS Solutions to Final Exam (in-class portion) January 22, 2003
MATH 23a, FALL 2002 THEORETICAL LINEAR ALGEBRA AND MULTIVARIABLE CALCULUS Solutions to Final Exam (in-class portion) January 22, 2003 1. True or False (28 points, 2 each) T or F If V is a vector space
More informationLinear Algebra and Robot Modeling
Linear Algebra and Robot Modeling Nathan Ratliff Abstract Linear algebra is fundamental to robot modeling, control, and optimization. This document reviews some of the basic kinematic equations and uses
More informationCOMP 558 lecture 18 Nov. 15, 2010
Least squares We have seen several least squares problems thus far, and we will see more in the upcoming lectures. For this reason it is good to have a more general picture of these problems and how to
More informationECE 275A Homework 6 Solutions
ECE 275A Homework 6 Solutions. The notation used in the solutions for the concentration (hyper) ellipsoid problems is defined in the lecture supplement on concentration ellipsoids. Note that θ T Σ θ =
More informationVectors. January 13, 2013
Vectors January 13, 2013 The simplest tensors are scalars, which are the measurable quantities of a theory, left invariant by symmetry transformations. By far the most common non-scalars are the vectors,
More information1 Matrices and Systems of Linear Equations
Linear Algebra (part ) : Matrices and Systems of Linear Equations (by Evan Dummit, 207, v 260) Contents Matrices and Systems of Linear Equations Systems of Linear Equations Elimination, Matrix Formulation
More informationKinematics. Chapter Multi-Body Systems
Chapter 2 Kinematics This chapter first introduces multi-body systems in conceptual terms. It then describes the concept of a Euclidean frame in the material world, following the concept of a Euclidean
More informationPURE MATHEMATICS AM 27
AM SYLLABUS (2020) PURE MATHEMATICS AM 27 SYLLABUS 1 Pure Mathematics AM 27 (Available in September ) Syllabus Paper I(3hrs)+Paper II(3hrs) 1. AIMS To prepare students for further studies in Mathematics
More informationIntroduction to Group Theory
Chapter 10 Introduction to Group Theory Since symmetries described by groups play such an important role in modern physics, we will take a little time to introduce the basic structure (as seen by a physicist)
More informationSolutions Preliminary Examination in Numerical Analysis January, 2017
Solutions Preliminary Examination in Numerical Analysis January, 07 Root Finding The roots are -,0, a) First consider x 0 > Let x n+ = + ε and x n = + δ with δ > 0 The iteration gives 0 < ε δ < 3, which
More information. D CR Nomenclature D 1
. D CR Nomenclature D 1 Appendix D: CR NOMENCLATURE D 2 The notation used by different investigators working in CR formulations has not coalesced, since the topic is in flux. This Appendix identifies the
More informationRigid Geometric Transformations
Rigid Geometric Transformations Carlo Tomasi This note is a quick refresher of the geometry of rigid transformations in three-dimensional space, expressed in Cartesian coordinates. 1 Cartesian Coordinates
More informationPURE MATHEMATICS AM 27
AM Syllabus (014): Pure Mathematics AM SYLLABUS (014) PURE MATHEMATICS AM 7 SYLLABUS 1 AM Syllabus (014): Pure Mathematics Pure Mathematics AM 7 Syllabus (Available in September) Paper I(3hrs)+Paper II(3hrs)
More informationChoice of Riemannian Metrics for Rigid Body Kinematics
Choice of Riemannian Metrics for Rigid Body Kinematics Miloš Žefran1, Vijay Kumar 1 and Christopher Croke 2 1 General Robotics and Active Sensory Perception (GRASP) Laboratory 2 Department of Mathematics
More informationManaging Uncertainty
Managing Uncertainty Bayesian Linear Regression and Kalman Filter December 4, 2017 Objectives The goal of this lab is multiple: 1. First it is a reminder of some central elementary notions of Bayesian
More informationLinear Algebra (part 1) : Vector Spaces (by Evan Dummit, 2017, v. 1.07) 1.1 The Formal Denition of a Vector Space
Linear Algebra (part 1) : Vector Spaces (by Evan Dummit, 2017, v. 1.07) Contents 1 Vector Spaces 1 1.1 The Formal Denition of a Vector Space.................................. 1 1.2 Subspaces...................................................
More informationUNIVERSITY OF DUBLIN
UNIVERSITY OF DUBLIN TRINITY COLLEGE JS & SS Mathematics SS Theoretical Physics SS TSM Mathematics Faculty of Engineering, Mathematics and Science school of mathematics Trinity Term 2015 Module MA3429
More informationUnit 2: Solving Scalar Equations. Notes prepared by: Amos Ron, Yunpeng Li, Mark Cowlishaw, Steve Wright Instructor: Steve Wright
cs416: introduction to scientific computing 01/9/07 Unit : Solving Scalar Equations Notes prepared by: Amos Ron, Yunpeng Li, Mark Cowlishaw, Steve Wright Instructor: Steve Wright 1 Introduction We now
More informationVector Space Basics. 1 Abstract Vector Spaces. 1. (commutativity of vector addition) u + v = v + u. 2. (associativity of vector addition)
Vector Space Basics (Remark: these notes are highly formal and may be a useful reference to some students however I am also posting Ray Heitmann's notes to Canvas for students interested in a direct computational
More information4.1 Eigenvalues, Eigenvectors, and The Characteristic Polynomial
Linear Algebra (part 4): Eigenvalues, Diagonalization, and the Jordan Form (by Evan Dummit, 27, v ) Contents 4 Eigenvalues, Diagonalization, and the Jordan Canonical Form 4 Eigenvalues, Eigenvectors, and
More informationAPPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2.
APPENDIX A Background Mathematics A. Linear Algebra A.. Vector algebra Let x denote the n-dimensional column vector with components 0 x x 2 B C @. A x n Definition 6 (scalar product). The scalar product
More informationSpring, 2012 CIS 515. Fundamentals of Linear Algebra and Optimization Jean Gallier
Spring 0 CIS 55 Fundamentals of Linear Algebra and Optimization Jean Gallier Homework 5 & 6 + Project 3 & 4 Note: Problems B and B6 are for extra credit April 7 0; Due May 7 0 Problem B (0 pts) Let A be
More informationGravitation: Special Relativity
An Introduction to General Relativity Center for Relativistic Astrophysics School of Physics Georgia Institute of Technology Notes based on textbook: Spacetime and Geometry by S.M. Carroll Spring 2013
More informationSPRING 2006 PRELIMINARY EXAMINATION SOLUTIONS
SPRING 006 PRELIMINARY EXAMINATION SOLUTIONS 1A. Let G be the subgroup of the free abelian group Z 4 consisting of all integer vectors (x, y, z, w) such that x + 3y + 5z + 7w = 0. (a) Determine a linearly
More informationENGI 9420 Lecture Notes 2 - Matrix Algebra Page Matrix operations can render the solution of a linear system much more efficient.
ENGI 940 Lecture Notes - Matrix Algebra Page.0. Matrix Algebra A linear system of m equations in n unknowns, a x + a x + + a x b (where the a ij and i n n a x + a x + + a x b n n a x + a x + + a x b m
More informationMODELLING OF FLEXIBLE MECHANICAL SYSTEMS THROUGH APPROXIMATED EIGENFUNCTIONS L. Menini A. Tornambe L. Zaccarian Dip. Informatica, Sistemi e Produzione
MODELLING OF FLEXIBLE MECHANICAL SYSTEMS THROUGH APPROXIMATED EIGENFUNCTIONS L. Menini A. Tornambe L. Zaccarian Dip. Informatica, Sistemi e Produzione, Univ. di Roma Tor Vergata, via di Tor Vergata 11,
More informationLecture Note 4: General Rigid Body Motion
ECE5463: Introduction to Robotics Lecture Note 4: General Rigid Body Motion Prof. Wei Zhang Department of Electrical and Computer Engineering Ohio State University Columbus, Ohio, USA Spring 2018 Lecture
More informationElementary maths for GMT
Elementary maths for GMT Linear Algebra Part 2: Matrices, Elimination and Determinant m n matrices The system of m linear equations in n variables x 1, x 2,, x n a 11 x 1 + a 12 x 2 + + a 1n x n = b 1
More informationECE 275A Homework #3 Solutions
ECE 75A Homework #3 Solutions. Proof of (a). Obviously Ax = 0 y, Ax = 0 for all y. To show sufficiency, note that if y, Ax = 0 for all y, then it must certainly be true for the particular value of y =
More informationMILNOR SEMINAR: DIFFERENTIAL FORMS AND CHERN CLASSES
MILNOR SEMINAR: DIFFERENTIAL FORMS AND CHERN CLASSES NILAY KUMAR In these lectures I want to introduce the Chern-Weil approach to characteristic classes on manifolds, and in particular, the Chern classes.
More information! 4 4! o! +! h 4 o=0! ±= ± p i And back-substituting into the linear equations gave us the ratios of the amplitudes of oscillation:.»» = A p e i! +t»»
Topic 6: Coupled Oscillators and Normal Modes Reading assignment: Hand and Finch Chapter 9 We are going to be considering the general case of a system with N degrees of freedome close to one of its stable
More information1 Kalman Filter Introduction
1 Kalman Filter Introduction You should first read Chapter 1 of Stochastic models, estimation, and control: Volume 1 by Peter S. Maybec (available here). 1.1 Explanation of Equations (1-3) and (1-4) Equation
More information1 Determinants. 1.1 Determinant
1 Determinants [SB], Chapter 9, p.188-196. [SB], Chapter 26, p.719-739. Bellow w ll study the central question: which additional conditions must satisfy a quadratic matrix A to be invertible, that is to
More informationDerivation of the Kalman Filter
Derivation of the Kalman Filter Kai Borre Danish GPS Center, Denmark Block Matrix Identities The key formulas give the inverse of a 2 by 2 block matrix, assuming T is invertible: T U 1 L M. (1) V W N P
More informationOptimisation on Manifolds
Optimisation on Manifolds K. Hüper MPI Tübingen & Univ. Würzburg K. Hüper (MPI Tübingen & Univ. Würzburg) Applications in Computer Vision Grenoble 18/9/08 1 / 29 Contents 2 Examples Essential matrix estimation
More informationII. DIFFERENTIABLE MANIFOLDS. Washington Mio CENTER FOR APPLIED VISION AND IMAGING SCIENCES
II. DIFFERENTIABLE MANIFOLDS Washington Mio Anuj Srivastava and Xiuwen Liu (Illustrations by D. Badlyans) CENTER FOR APPLIED VISION AND IMAGING SCIENCES Florida State University WHY MANIFOLDS? Non-linearity
More informationDifferential Kinematics
Differential Kinematics Relations between motion (velocity) in joint space and motion (linear/angular velocity) in task space (e.g., Cartesian space) Instantaneous velocity mappings can be obtained through
More informationVectors To begin, let us describe an element of the state space as a point with numerical coordinates, that is x 1. x 2. x =
Linear Algebra Review Vectors To begin, let us describe an element of the state space as a point with numerical coordinates, that is x 1 x x = 2. x n Vectors of up to three dimensions are easy to diagram.
More information, respectively to the inverse and the inverse differential problem. Check the correctness of the obtained results. Exercise 2 y P 2 P 1.
Robotics I July 8 Exercise Define the orientation of a rigid body in the 3D space through three rotations by the angles α β and γ around three fixed axes in the sequence Y X and Z and determine the associated
More informationA matrix over a field F is a rectangular array of elements from F. The symbol
Chapter MATRICES Matrix arithmetic A matrix over a field F is a rectangular array of elements from F The symbol M m n (F ) denotes the collection of all m n matrices over F Matrices will usually be denoted
More informationMath 52H: Multilinear algebra, differential forms and Stokes theorem. Yakov Eliashberg
Math 52H: Multilinear algebra, differential forms and Stokes theorem Yakov Eliashberg March 202 2 Contents I Multilinear Algebra 7 Linear and multilinear functions 9. Dual space.........................................
More informationProperties of Linear Transformations from R n to R m
Properties of Linear Transformations from R n to R m MATH 322, Linear Algebra I J. Robert Buchanan Department of Mathematics Spring 2015 Topic Overview Relationship between the properties of a matrix transformation
More informationDS-GA 1002 Lecture notes 0 Fall Linear Algebra. These notes provide a review of basic concepts in linear algebra.
DS-GA 1002 Lecture notes 0 Fall 2016 Linear Algebra These notes provide a review of basic concepts in linear algebra. 1 Vector spaces You are no doubt familiar with vectors in R 2 or R 3, i.e. [ ] 1.1
More informationThe groups SO(3) and SU(2) and their representations
CHAPTER VI The groups SO(3) and SU() and their representations Two continuous groups of transformations that play an important role in physics are the special orthogonal group of order 3, SO(3), and the
More informationPage 52. Lecture 3: Inner Product Spaces Dual Spaces, Dirac Notation, and Adjoints Date Revised: 2008/10/03 Date Given: 2008/10/03
Page 5 Lecture : Inner Product Spaces Dual Spaces, Dirac Notation, and Adjoints Date Revised: 008/10/0 Date Given: 008/10/0 Inner Product Spaces: Definitions Section. Mathematical Preliminaries: Inner
More informationNotes on multivariable calculus
Notes on multivariable calculus Jonathan Wise February 2, 2010 1 Review of trigonometry Trigonometry is essentially the study of the relationship between polar coordinates and Cartesian coordinates in
More informationMatrices and Vectors
Matrices and Vectors James K. Peterson Department of Biological Sciences and Department of Mathematical Sciences Clemson University November 11, 2013 Outline 1 Matrices and Vectors 2 Vector Details 3 Matrix
More informationx 1 x 2. x 1, x 2,..., x n R. x n
WEEK In general terms, our aim in this first part of the course is to use vector space theory to study the geometry of Euclidean space A good knowledge of the subject matter of the Matrix Applications
More informationFraction-free Row Reduction of Matrices of Skew Polynomials
Fraction-free Row Reduction of Matrices of Skew Polynomials Bernhard Beckermann Laboratoire d Analyse Numérique et d Optimisation Université des Sciences et Technologies de Lille France bbecker@ano.univ-lille1.fr
More informationLINEAR ALGEBRA - CHAPTER 1: VECTORS
LINEAR ALGEBRA - CHAPTER 1: VECTORS A game to introduce Linear Algebra In measurement, there are many quantities whose description entirely rely on magnitude, i.e., length, area, volume, mass and temperature.
More informationMATH 583A REVIEW SESSION #1
MATH 583A REVIEW SESSION #1 BOJAN DURICKOVIC 1. Vector Spaces Very quick review of the basic linear algebra concepts (see any linear algebra textbook): (finite dimensional) vector space (or linear space),
More informationReview for Exam Find all a for which the following linear system has no solutions, one solution, and infinitely many solutions.
Review for Exam. Find all a for which the following linear system has no solutions, one solution, and infinitely many solutions. x + y z = 2 x + 2y + z = 3 x + y + (a 2 5)z = a 2 The augmented matrix for
More informationMATH 320, WEEK 7: Matrices, Matrix Operations
MATH 320, WEEK 7: Matrices, Matrix Operations 1 Matrices We have introduced ourselves to the notion of the grid-like coefficient matrix as a short-hand coefficient place-keeper for performing Gaussian
More informationELEMENTARY LINEAR ALGEBRA WITH APPLICATIONS. 1. Linear Equations and Matrices
ELEMENTARY LINEAR ALGEBRA WITH APPLICATIONS KOLMAN & HILL NOTES BY OTTO MUTZBAUER 11 Systems of Linear Equations 1 Linear Equations and Matrices Numbers in our context are either real numbers or complex
More informationELEMENTARY LINEAR ALGEBRA
ELEMENTARY LINEAR ALGEBRA K R MATTHEWS DEPARTMENT OF MATHEMATICS UNIVERSITY OF QUEENSLAND First Printing, 99 Chapter LINEAR EQUATIONS Introduction to linear equations A linear equation in n unknowns x,
More informationLinear Algebra. Preliminary Lecture Notes
Linear Algebra Preliminary Lecture Notes Adolfo J. Rumbos c Draft date May 9, 29 2 Contents 1 Motivation for the course 5 2 Euclidean n dimensional Space 7 2.1 Definition of n Dimensional Euclidean Space...........
More informationPhysics 70007, Fall 2009 Answers to Final Exam
Physics 70007, Fall 009 Answers to Final Exam December 17, 009 1. Quantum mechanical pictures a Demonstrate that if the commutation relation [A, B] ic is valid in any of the three Schrodinger, Heisenberg,
More informationElliptically Contoured Distributions
Elliptically Contoured Distributions Recall: if X N p µ, Σ), then { 1 f X x) = exp 1 } det πσ x µ) Σ 1 x µ) So f X x) depends on x only through x µ) Σ 1 x µ), and is therefore constant on the ellipsoidal
More informationNumerical Methods for Inverse Kinematics
Numerical Methods for Inverse Kinematics Niels Joubert, UC Berkeley, CS184 2008-11-25 Inverse Kinematics is used to pose models by specifying endpoints of segments rather than individual joint angles.
More informationParameter Estimation
Parameter Estimation Tuesday 9 th May, 07 4:30 Consider a system whose response can be modeled by R = M (Θ) where Θ is a vector of m parameters. We take a series of measurements, D (t) where t represents
More informationLinear Algebra Primer
Introduction Linear Algebra Primer Daniel S. Stutts, Ph.D. Original Edition: 2/99 Current Edition: 4//4 This primer was written to provide a brief overview of the main concepts and methods in elementary
More informationLecture 4: Extended Kalman filter and Statistically Linearized Filter
Lecture 4: Extended Kalman filter and Statistically Linearized Filter Department of Biomedical Engineering and Computational Science Aalto University February 17, 2011 Contents 1 Overview of EKF 2 Linear
More informationMath 4A Notes. Written by Victoria Kala Last updated June 11, 2017
Math 4A Notes Written by Victoria Kala vtkala@math.ucsb.edu Last updated June 11, 2017 Systems of Linear Equations A linear equation is an equation that can be written in the form a 1 x 1 + a 2 x 2 +...
More informationAutonomous Mobile Robot Design
Autonomous Mobile Robot Design Topic: Extended Kalman Filter Dr. Kostas Alexis (CSE) These slides relied on the lectures from C. Stachniss, J. Sturm and the book Probabilistic Robotics from Thurn et al.
More informationDS-GA 1002 Lecture notes 10 November 23, Linear models
DS-GA 2 Lecture notes November 23, 2 Linear functions Linear models A linear model encodes the assumption that two quantities are linearly related. Mathematically, this is characterized using linear functions.
More information2 Eigenvectors and Eigenvalues in abstract spaces.
MA322 Sathaye Notes on Eigenvalues Spring 27 Introduction In these notes, we start with the definition of eigenvectors in abstract vector spaces and follow with the more common definition of eigenvectors
More informationa s 1.3 Matrix Multiplication. Know how to multiply two matrices and be able to write down the formula
Syllabus for Math 308, Paul Smith Book: Kolman-Hill Chapter 1. Linear Equations and Matrices 1.1 Systems of Linear Equations Definition of a linear equation and a solution to a linear equations. Meaning
More informationLinear Combinations of Optic Flow Vectors for Estimating Self-Motion a Real-World Test of a Neural Model
Linear Combinations of Optic Flow Vectors for Estimating Self-Motion a Real-World Test of a Neural Model Matthias O. Franz MPI für biologische Kybernetik Spemannstr. 38 D-72076 Tübingen, Germany mof@tuebingen.mpg.de
More informationLinear Algebra. The analysis of many models in the social sciences reduces to the study of systems of equations.
POLI 7 - Mathematical and Statistical Foundations Prof S Saiegh Fall Lecture Notes - Class 4 October 4, Linear Algebra The analysis of many models in the social sciences reduces to the study of systems
More informationMath 3108: Linear Algebra
Math 3108: Linear Algebra Instructor: Jason Murphy Department of Mathematics and Statistics Missouri University of Science and Technology 1 / 323 Contents. Chapter 1. Slides 3 70 Chapter 2. Slides 71 118
More informationA Riemannian Framework for Denoising Diffusion Tensor Images
A Riemannian Framework for Denoising Diffusion Tensor Images Manasi Datar No Institute Given Abstract. Diffusion Tensor Imaging (DTI) is a relatively new imaging modality that has been extensively used
More informationMTH Linear Algebra. Study Guide. Dr. Tony Yee Department of Mathematics and Information Technology The Hong Kong Institute of Education
MTH 3 Linear Algebra Study Guide Dr. Tony Yee Department of Mathematics and Information Technology The Hong Kong Institute of Education June 3, ii Contents Table of Contents iii Matrix Algebra. Real Life
More informationa11 a A = : a 21 a 22
Matrices The study of linear systems is facilitated by introducing matrices. Matrix theory provides a convenient language and notation to express many of the ideas concisely, and complicated formulas are
More informationDynamical Systems. August 13, 2013
Dynamical Systems Joshua Wilde, revised by Isabel Tecu, Takeshi Suzuki and María José Boccardi August 13, 2013 Dynamical Systems are systems, described by one or more equations, that evolve over time.
More information4.2. ORTHOGONALITY 161
4.2. ORTHOGONALITY 161 Definition 4.2.9 An affine space (E, E ) is a Euclidean affine space iff its underlying vector space E is a Euclidean vector space. Given any two points a, b E, we define the distance
More informationOctober 25, 2013 INNER PRODUCT SPACES
October 25, 2013 INNER PRODUCT SPACES RODICA D. COSTIN Contents 1. Inner product 2 1.1. Inner product 2 1.2. Inner product spaces 4 2. Orthogonal bases 5 2.1. Existence of an orthogonal basis 7 2.2. Orthogonal
More information10. Smooth Varieties. 82 Andreas Gathmann
82 Andreas Gathmann 10. Smooth Varieties Let a be a point on a variety X. In the last chapter we have introduced the tangent cone C a X as a way to study X locally around a (see Construction 9.20). It
More informationLifting to non-integral idempotents
Journal of Pure and Applied Algebra 162 (2001) 359 366 www.elsevier.com/locate/jpaa Lifting to non-integral idempotents Georey R. Robinson School of Mathematics and Statistics, University of Birmingham,
More informationVectors Coordinate frames 2D implicit curves 2D parametric curves. Graphics 2008/2009, period 1. Lecture 2: vectors, curves, and surfaces
Graphics 2008/2009, period 1 Lecture 2 Vectors, curves, and surfaces Computer graphics example: Pixar (source: http://www.pixar.com) Computer graphics example: Pixar (source: http://www.pixar.com) Computer
More informationMath 108b: Notes on the Spectral Theorem
Math 108b: Notes on the Spectral Theorem From section 6.3, we know that every linear operator T on a finite dimensional inner product space V has an adjoint. (T is defined as the unique linear operator
More informationLinear Algebra. Preliminary Lecture Notes
Linear Algebra Preliminary Lecture Notes Adolfo J. Rumbos c Draft date April 29, 23 2 Contents Motivation for the course 5 2 Euclidean n dimensional Space 7 2. Definition of n Dimensional Euclidean Space...........
More informationMath113: Linear Algebra. Beifang Chen
Math3: Linear Algebra Beifang Chen Spring 26 Contents Systems of Linear Equations 3 Systems of Linear Equations 3 Linear Systems 3 2 Geometric Interpretation 3 3 Matrices of Linear Systems 4 4 Elementary
More informationPHYSICS 219 Homework 2 Due in class, Wednesday May 3. Makeup lectures on Friday May 12 and 19, usual time. Location will be ISB 231 or 235.
PHYSICS 219 Homework 2 Due in class, Wednesday May 3 Note: Makeup lectures on Friday May 12 and 19, usual time. Location will be ISB 231 or 235. No lecture: May 8 (I m away at a meeting) and May 29 (holiday).
More informationZero Energy Modes in One Dimension: An Introduction to Hourglass Modes
Zero Energy Modes in One Dimension: An Introduction to Hourglass Modes David J. Benson March 9, 2003 Reduced integration does a lot of good things for an element: it reduces the computational cost, it
More informationCore A-level mathematics reproduced from the QCA s Subject criteria for Mathematics document
Core A-level mathematics reproduced from the QCA s Subject criteria for Mathematics document Background knowledge: (a) The arithmetic of integers (including HCFs and LCMs), of fractions, and of real numbers.
More informationCS 246 Review of Linear Algebra 01/17/19
1 Linear algebra In this section we will discuss vectors and matrices. We denote the (i, j)th entry of a matrix A as A ij, and the ith entry of a vector as v i. 1.1 Vectors and vector operations A vector
More informationRegression. Oscar García
Regression Oscar García Regression methods are fundamental in Forest Mensuration For a more concise and general presentation, we shall first review some matrix concepts 1 Matrices An order n m matrix is
More informationNumerical Linear Algebra Homework Assignment - Week 2
Numerical Linear Algebra Homework Assignment - Week 2 Đoàn Trần Nguyên Tùng Student ID: 1411352 8th October 2016 Exercise 2.1: Show that if a matrix A is both triangular and unitary, then it is diagonal.
More informationEquality: Two matrices A and B are equal, i.e., A = B if A and B have the same order and the entries of A and B are the same.
Introduction Matrix Operations Matrix: An m n matrix A is an m-by-n array of scalars from a field (for example real numbers) of the form a a a n a a a n A a m a m a mn The order (or size) of A is m n (read
More informationPower Series Solutions We use power series to solve second order differential equations
Objectives Power Series Solutions We use power series to solve second order differential equations We use power series expansions to find solutions to second order, linear, variable coefficient equations
More information