For s.e. of flux multiply cv by mean flux over time period Damage: penetration depends on size
sagtu17.pdf Ascona12.pdf
Use of A(, ). bandpass filtering Suppose X(x,y) j,k jk exp{i( j x + k y)} Y(x,y) = A[X](x,y) j,k A( j, k ) jk exp{i( j x + k y)} e.g. If A(, ) = 1, | ± 0 |, | ± 0 | = 0 otherwise Y(x,y) contains only these terms Repeated xeroxing Filtering/smoothing.
Approximating an ideal low-pass filter. Transfer function A( ) = 1 | | Ideal Y(t) = a(u) X(t-u) t,u in Z A( ) = a(u) exp{-i u) - < a(u) = exp{iu }A( )d / 2 = |lamda|<Omega exp{i u }d /2 = / u=0 = sin u/ u u 0
Bank of bandpass filters
Fourier series. How close is A (n) ( ) to A( ) ?
By substitution
Error
Convergence factors. Fejer (1900) Replace (*) by Fejer kernel integrates to 1 non-negative approximate Dirac delta
General class. h(u) = 0, |u|>1 h(u/n) exp{-i u} a(u) = H (n) ( ) A( - ) d (**) with H (n) ( ) = (2 ) -1 h(u/n) exp{-i u} h(.): convergence factor, taper, data window, fader (**) = A( ) + n -1 H( )d A'( ) + ½n -2 2 H( )d A"( ) +...
Lowpass filter.
Smoothing/smoothers. goal: retain smooth/low frequency components of signal while reducing the more irregular/high frequency ones difficulty: no universal definition of smooth curve Example. running mean ave t-k s t+k Y(s)
Kernel smoother. S(t) = w b (t-s)Y(s) / w b (t-s) w b (t) = w(t/b) b: bandwidth ksmooth()
Local polynomial. Linear case Obtain a t, b t OLS intercept and slope of points {(s,Y(s)): t-k s t+k} S(t) = a t + b t t span: (2k+1)/n lowess(), loess(): WLS can be made resistant
Running median med t-k s t+k Y(s) Repeat til no change Other things: parametric model, splines,... choice of bandwidth/binwidth
Finite Fourier transforms. Considered
Empirical Fourier analysis. Uses. Estimation - parameters and periods Unification of data types Approximation of distributions System identification Speeding up computations Model assessment...
Examples. 1. Constant. X(t)=1
Inversion. fft()
Convolution. Lemma If |X(t) M, a(0) and |ua(u)| A, Y(t) = a(t-u)X(u) then, |d Y T ( ) – A( ) d Y T ( ) | 4MA Application. Filtering Add S-T zeroes
Periodogram. |d T ( )| 2
Chandler wobble.
Interpretation of frequency.
Some other empirical FTs. 1. Point process on the line. {0 j <T}, j=1,...,N N(t), 0 t<T dN(t)/dt = j (t- j ) Might approximate by a 0-1 time series Y t = 1 point in [0,t) = 0 otherwise j Y t exp{-i t}
2. M.p.p. (sampled time series). { j, M j } {Y( j )} j M j exp{-i j } j Y( j ) exp{-i j }
3. Measure, processes of increments 4. Discrete state-valued process Y(t) values in N, g:N R t g(Y(t)) exp{-i t} 5. Process on circle Y( ), 0 < Y( ) = k k exp{i k}
Other processes. process on sphere, line process, generalized process, vector-valued time, LCA group