After the quake, Portuguese priests were asked to document their observations. Their records, still preserved, represent the first systematic attempt to investigate an earthquake and its effects. Since then, detailed records have been kept of almost
Another great earth- quake occurred on January 23,1812, and a third, the strongest yet, on February 7,. 1812. Aftershocks were nearly continu- ous between these great .... Earth's crust. Even if a fault zone has recently experienced an earthquake, ho
(crusted over lava blobs), ashflows, and tephra (airborne ash and dust). A volcano is most commonly a con- ical hill or mountain built around a vent that connects with reservoirs of molten rock below the surface of the. Earth. The term volcano also r
Download Jom FTEKNIK Volume 3 No.2Oktober 2016. 1. MANAJEMEN SISTEM PERGUDANGAN/WAREHOUSE PERALATAN LABORATORIUM TEKNIK ..... X. Jurnal Titra 5(2). [3] Sorooshian, S., Salimi, M., Bavani, S. dan. Aminattaheri, H. 2012, “Experience of 5S. Implem
Download Jom FTEKNIK Volume 3 No.2Oktober 2016. 1. MANAJEMEN SISTEM PERGUDANGAN/WAREHOUSE PERALATAN LABORATORIUM TEKNIK ..... X. Jurnal Titra 5(2). [3] Sorooshian, S., Salimi, M., Bavani, S. dan. Aminattaheri, H. 2012, “Experience of 5S. Implem
Mindanao Island from Apex Mining Co. Apex previously operated the underground mine from 1981 to 1989, producing almost 12.5 mt of gold before closing it ..... 1994, pp. 8. 8-9. Mining Journal (London). V. 323, No. 8289, Aug. 14 , 1994, p. 131. 9. Sou
May 18, 1980 ... that the eruption reached its peak between 3:00 and 5:00 p.m. Over the course of the day, ... episodes. Each of the dome-building episodes added between 1 and 29 million cubic yards of new lava to the dome. Most of the growth occurre
Download MANAJEMEN SISTEM PERGUDANGAN/WAREHOUSE PERALATAN LABORATORIUM TEKNIK .... terbaru. 2.6 Perancangan Perangkat Lunak. Tahap perancangan sistem merupakan tahapan dalam membuat rincian sistem agar dimengerti oleh pengguna (user). 1. .....
Download MANAJEMEN SISTEM PERGUDANGAN/WAREHOUSE PERALATAN LABORATORIUM TEKNIK .... terbaru. 2.6 Perancangan Perangkat Lunak. Tahap perancangan sistem merupakan tahapan dalam membuat rincian sistem agar dimengerti oleh pengguna (user). 1. .....
disasters, such as hurricanes, tsunamis, and forest fires, and by ... you do, many different types of crises have the poten- tial to significantly disrupt .... FedEx quickly responded to the surprise of the crisis (see http://www.youtube .com/ watch?
2 July 2017 Publications WASTONG GAMIT NG MGA SALITA SA WIKANG FILIPINO, MAHALAGA PA BA? ni: Camela C. Mila “Ang hindi magmahal sa kanyang salita, mahigit sa hayop
Download serta melengkapi secara komprehensif atas implementasi Sistem Informasi Manajemen. Dokumen ini adalah kerangka dasar yang .... MANAGER. •Cek Stok Kebutuhan. •Approval Manager. PEMBELIAN. •Validasi Data. Pembelian. •Cetak Order. Pembeli
australian mathematics trust 1 publications from the australian mathematics trust www.amt.edu.au competition materials extension materials international mathematics
Download Jurnal Sistem Informasi MTI-UI, Volume 4, Nomor 1, ISBN 1412-8896. PEMANFAATAN DATA WAREHOUSE DI PERGURUAN TINGGI INDONESIA. Iik Wilarso. Fakultas Ilmu Komputer, Universitas Indonesia, Depok, Indonesia iik@cs .ui.ac.id. Abstrak. Paradig
Download Biro Logistik dan Warehouse Proyek Indarung VI PT. Semen Padang bertanggung jawab dalam pemesanan dan pengolahan barang yang dibutuhkan untuk membangun pabrik VI PT. Semen Padang. Penelitian ini bertujuan untuk mengetahui bagaimana pen
Download Perancangan Data Warehouse Untuk Penerimaan Mahasiswa Baru. Jurnal Ilmu Komputer Volume 9 Nomor 2, September 2013. 148 ... Saat ini banyak perguruan tinggi yang sudah memiliki aplikasi yang memanfaatkan database.
professionals to implement innovative solutions to effectively manage complex global supply chains while simultaneously reducing operational expenses
Download Biro Logistik dan Warehouse Proyek Indarung VI PT. Semen Padang bertanggung jawab dalam pemesanan dan pengolahan barang yang dibutuhkan untuk membangun pabrik VI PT. Semen Padang. Penelitian ini bertujuan untuk mengetahui bagaimana pen
Technical Publications Technical Manual Development • Military page-based Technical Manuals (TM), Electronic Technical Manuals (ETM), Interactive Electronic
Download Kata kunci: pergudangan, supply chain management, warehouse management sistem. 1. ... Prosiding SNST ke-8 Tahun 2017. Fakultas Teknik .... Jurnal. Ilmiah Komputer Dan Informatika (Komputa) Issn : 2089-9033, 1-8. Agustina, M., & Oktasar
Download PEMANFAATAN DATA WAREHOUSE DI PERGURUAN TINGGI INDONESIA. Iik Wilarso. Fakultas Ilmu Komputer, Universitas Indonesia, Depok, Indonesia iik@cs .ui.ac.id. Abstrak. Paradigma Baru Pendidikan Tinggi [1,2] yang dicanangkan oleh Direktorat Je
Data Warehouse Design: Modern Principles and Methodologies 2.3.3 Conceptual Design 52 2.3.4 Workload Refinement and Validation of Conceptual Schemata 53
Download Biro Logistik dan Warehouse Proyek Indarung VI PT. Semen Padang bertanggung jawab dalam pemesanan dan pengolahan barang yang dibutuhkan untuk membangun pabrik VI PT. Semen Padang. Penelitian ini bertujuan untuk mengetahui bagaimana pen
Download Perancangan Data Warehouse Untuk Penerimaan Mahasiswa Baru. Jurnal Ilmu Komputer Volume 9 Nomor 2, September 2013. 148 .... Berikut ini gambaran lebih rinci tentang lima kekuatan ini dalam kaitannya dengan pendidikan. Gambar 2.
Improved digital filters for evaluating Fourier and Hankel transform integrals
By
Walter L. Anderson U.S. Geological Survey Denver, Colorado 80225
.I)SGS - GO -
75 -0 1 2
f$TANDA;W TITLE PAGE I~ECHNICAL REPORTS
· Re~prS~A.·
.u ~li~ -
GO
-
75 -
0
1
2
PB 242 800 '
4. Tlt fe and Subtitle
5. Report Date
Improved digital filters for evaluating Fourier and Hankel transform integrals
1975
h.--:----:-~:---~--:----;:::--;--- -1
Performing OrR ani zat ion Code
j
I
17.
8. Performing
Orga n i z atio~ -R-ept.N·o-:- ·1
i
Walter L. Anderson 9. Performing Organization Name and Address
10. Projec t/ Task/ Work Un it No.
U.S. Geological Survey Denver Federal Center, Bldg. 85 Denver, Colorado 80225
TI.
contrac ITra.nt No.--
·- --- -
-1
I I
12.-Sporisori;g· Agency. Name and Address------ -· - ··- -- -- - - -
New algorithms are described for evaluating Fourier (cosine, sine) and Hankel (J ,J ) transform integrals by means of digital filters. The filters have been designed 0 1 with extended lengths so that a variable convolution operation can be applied to a large class of integral transforms having the same system transfer function. A f ' lagged-convolution method is also presented to significantly decrease the computation time when computing a series of like-transforms over a parameter set spaced the same as the filters. Accuracy of the new filters is comparable to Gaussian integration, provided moderate parameter ranges and well-behaved kernel functions are used. A collection of Fortran IV subprograms is included for both real and complex functions for each filter type. The algorithms have been successfully used in geophysical applications containing a wide variety of integral transforms. 17. Key Words and Document Analysis. (a). Descr iptcrs
1201 0902
Numerical integration Digital filters Fortran program
17b. Identifiers/ Open-Ended Terms
Fourier transforms Hankel transforms
REPRODUCED BY
NATIONAl TECHNICAL INFORMATION SERVICE U.S. DEPARTMENT OF COMMERCE SPRINGFIELD, VA. 22161
Improved Digital Filters for Evaluating Fourier and Hankel Transform Integrals By Walter L. Anderson Abstract New algorithms are described for evaluating Fourier (cosine, sine) and Hankel (J ,J ) transform integrals by means of digital filters. 0 1
The
filters have been designed with extended lengths so that a variable convolution operation can be applied to a large class of integral transforms \ having the same system transfer function.
A lagged-convolution method is
also presented to significantly decrease the computation time when computing a series of like-transforms over a parameter set spaced the same as the filters.
Accuracy of the new filters is comparable to Gaussian integration,
provided moderate parameter ranges and well-behaved kernel functions are used.
A collection of Fortran IV subprograms is included for both real
and complex functions for each filter type.
The algorithms have been
successfully used in geophysical applications containing a wide variety of integral transforms.
1
Acknowledgment The author thanks Bruce D. Smith, Raymond D. Watts, and Frank C. Frischknecht, U. S. Geological Survey, who proposed this study and provided pertinent suggestions.
I
\
2
Introduction Digital filtering techniques to evaluate oscillatory-type convolution integrals have become a widespread method reported in the geophysical literature.
For example, Koefoed et al (1972)
developed Hankel transform
filters for functions encountered in electromagnetic depth soundings. Anderson (1973) extended this method to Fourier cosine transforms.
Das
et al (1974a, 1974b) proposed several specific filters used in transformations of resistivity sounding curves.
Anderson (1974) used
a standard set of Hankel and Fourier transform filters to evaluate a wide class of integral transforms required in computing electromagnetic field components about a finite electric wire source. In all these papers, the main advantage cited for using digital filters over direct numerical integration is the increased speed of the calculation. An order of magnitude or more in speed improvement is not uncommon,
but
this is usually achieved with a reduction in accuracy. This report briefly reviews the filter design process for convolution integrals.
A new method is presented to improve the designed filter
coefficients so that increased accuracy is possible at very little sacrifice in computational speed.
A variable cutoff convolution method is
described which utilizes the nature of the input kernel function and filter response to approximate the integral given a truncation tolerance. An algorithm is described for evaluating Fourier (cosine, sine) and Hankel (J ,J ) transform integrals for well-behaved kernels.
0
1 Additionally, a special lagged-convolution algorithm is developed to
significantly decrease the overall computational time over variable convolution.
This method is especially advantageous when computing a set
of like-integrals where the transform parameter is equivalent to an incremented shift in the sampled filter.
3
Sixteen similar, but self-contained, Fortran IV computer subprograms utilizing the new filtering algorithms are listed in appendix 1. subprogram naming convention used is given in tables 2 and 3.
The
Each of
these routines contains comments to explicitly define the calling parameters.
With minor changes, the Fortran IV subprograms should be
acceptable to most current-day digital computers.
Some examples, using
the improved filters, are tabulated in appendix 2. Finally, the new filtering methods are discussed in light of other recently published techniques.
4
Filter design The method of designing digital filters for Hankel transform integrals presented by Koefoed et al (1972)
may be generalized to other type
integrals expressed as a linear system convolution in the form 00
K(x) =
where
f k(y)s(x-y) dy,
(l)
k(y)
=
input function (also called kernel function below)
s(y)
=
system transfer (or filter) function
K(x)
=
convolved output function.
We may consider (1) as either time-domain or frequency-domain convolution.
By Fourier transform theory (e.g., see Papoulis, 1962) linear
convolution is equivalent to multiplication in the transformed-domain; that is, K(x)
k(x)s(x),
(2)
where K(x)++ K(x),k(x) ++ k(x),and s(x) ++ s(x) are Fourier transform pairs.
For a known system input-output relation, the transformed
system (filter) response is s(x) provided
k(x)
(3)
K(x)/k(x),
is not identically zero for some x , and
bo~h
and output functions have band-limited Fourier transforms. required so that
s(x)
~
0 for x
~
±00
•
5
system input
The latter is
The filter impulse response is the inverse Fourier transform of s(x)
, but would not be useful for convolution unless truncated to a
reasonable finite length. of N points spaced
~x
To this
end, we first choose a discrete sampling
apart and obtain
s(xj),j = 1,2, •.. ,N>>O.
l/(2~x).
results in a Nyquist frequency of
The sampling
A suitable truncation may be
obtained by cutting-off the transformed response at the Nyquist frequency. Note this is equivalent to multiplying
by the Fourier transform of
s(x)
sinc(x.) = sin(nx./~x)/(nx./~x) • . J J J The truncated response is inverse Fourier transformed to obtain the
the function
sampled filter impulse response (hereafter simply called filter weights or filter response).
In some cases, it is advantageous to shift the
resulting filter response to further minimize the magnitude of the filter tails.
For example, see Koefoed (1972)
for the reason of the oscillating
behavior of the filter response and the suggested shift at zero crossings. If proper input-output functionals are used, then the final filter response should asymptotically approach zero in both abscissa directions. Application of the filter weights to specific input functions in (1) is given by the non-circular convolution sum
K(x)
where
w
NW
E W.k(x-a.),
i=l
~
(4)
~
filter weights; i = 1,2, •.. ,NW=number of weights,
i
x = transform parameter ai
c
filter abscissa corresponding to
w1
.
Thus, the convolution integral (1) has been reduced to a straightforward summation given in (4), and the predetermined filter weights remain
6
constant for all classes of convolution integrals having the same system transfer function. In practice, many transform integrals encountered range over (O,oo) therefore a transformation to the range (4) to a given kernel function. integer order
(-oo,oo)
and
is required before applying
For example, the Hankel transform of
is defined as
n~O
j gh(g)Jn(bg) dg,
H(b)
(5)
0
= modified kernel* for Hankel transforms,
where
gh(g)
and
J (bg) = Bessel function of the first kind of order n. n
Similarly, the Fourier sine or cosine integral is given by
oo
b
F(b)
sin f(g)cos(bg)
dg.
(6)
If we let x = ln(b),y = ln(l/g)and after multiplying by ex, equations (5) and (6) become respectively,
(7)
and e
X
X
F(e )
*Many texts define gJn(bg)
(8)
f
as the Hankel
the definition given in (1) and consider function with ·h*(g)
=
kernel~
however, we will adopt the system transfer
gh(g) the system input kernel function. 7
Equations (7) and (8) are expressed in the convolution form (1), which is assumed when applying equation (4).
Multiplying by ex ensures
that the system response approaches zero for both large and small filter abscissas. An important consideration in designing digital filters is the choice of known input-output function pairs.
In previous work by Koefoed et al
(1972) and Anderson ( 1973), certain known Lipschitz integrals were used to develop Hankel (J , J ) and Fourier cosine filters, respectively. 0 1
Both
papers illustrate filter responses that are characterized as oscillatory decreasing functions in both abscissa directions.
The integral forms
previously used for filter design are indicated by (¢) in table 1 (The sine-transform form was previously used by the author, even though the filter weights were never published).