Engineering: Reports
http://hdl.handle.net/10092/611
Sun, 02 Oct 2016 17:21:24 GMT2016-10-02T17:21:24ZA modula-2 cross-compiler for the Macintosh.
http://hdl.handle.net/10092/12760
A modula-2 cross-compiler for the Macintosh.
Gifford, Jonathan
There is a Modula-2 cross-compiler available on the Prime, which currently generates code for the Prime. This cross-compiler was developed at ETH (Eidgenössische Technische Hochschule) in Zürich, Switzerland, by a team lend by H. Seiler, and was based on the original compiler written by Niklaus Wirth. This cross-compiler is called Smiler-2, and has code generators for the MC-68000, MC6809, and the PDP-11. The code generation pass and linker for the Prime ere written by Greg Ewing at the University of Canterbury in 1985.
The aims of this project were to translate the ETH MC-68000 code generator from CDC-6000 Pascal to Sheffield Pascal, and modify either the Prime or ETH linker so that it produced a Macintosh resource fork, which could then be incorporated into an application file on the Macintosh.
Wed, 01 Jan 1986 00:00:00 GMThttp://hdl.handle.net/10092/127601986-01-01T00:00:00ZFast evaluation of radial basis functions : methods for generalised multiquadrics in ℝⁿ
http://hdl.handle.net/10092/12759
Fast evaluation of radial basis functions : methods for generalised multiquadrics in ℝⁿ
Cherrie, J. B.; Beatson, Richard Keith; Newsam, G.N.
A generalised multiquadric radial basis function is a function of the form s(x) = ∑ᴺ𝑖₌₁ 𝑑𝑖 𝜙 (𝗅x-t𝑖𝗅), where 𝜙(r) = (r² + 𝝉²)ᵏ/², x ∈ ℝⁿ, and k ∈ Z is odd. The direct evaluation of an N centre generalised multiquadric radial basis function at m
points requires 𝒪(mN) flops, which is prohibitive when m and N are large. Similar
considerations apparently rule out fitting an interpolating N centre generalised
multiquadric to N data points by either direct or iterative solution of the associated
system of linear equations in realistic problems.
In this paper we will develop far field expansions, recurrence relations for efficient
formation of the expansions, error estimates, and translation formulas, for generalised
multiquadric radial basis functions in n-variables. These pieces are combined
in a hierarchical fast evaluator requiring only 𝒪((m + N) log N llog 𝜖lⁿ⁺¹) flops for
evaluation of an N centre generalised multiquadric at m points. This flop count
compares very favourably with the cost of the direct method. Moreover, used to
compute matrix-vector products, the fast evaluator provides a basis for fast iterative
fitting strategies.
Sat, 01 Jan 2000 00:00:00 GMThttp://hdl.handle.net/10092/127592000-01-01T00:00:00ZSpanning and sampling in Lebesgue and Sobolev spaces
http://hdl.handle.net/10092/12719
Spanning and sampling in Lebesgue and Sobolev spaces
Bui, Huy-Qui; Laugesen, R. S.
We establish conditions under which the small-scale affine system {𝜓(𝑎j 𝑥 - 𝑘) : j ≥ J, 𝑘 ∈ℤᵈ}, with J∈ℤ fixed, spans the Lebesgue space 𝐿ᵖ(ℝᵈ) and the Sobolev space 𝑊ᵐ,ᵖ(ℝᵈ), 1 ≤ 𝑝 < ∞. The dilation matrices 𝑎j are expanding (meaning limj →∞ ll𝑎j ⁻¹ ll = 0) but they need not be diagonal.
For spanning 𝐿ᵖ we require ∫ℝᵈ 𝜓𝑑𝑥 ≠ 0 and (when 𝑝 > 1) that the periodization of l𝜓l or of 𝟙{𝜓≠₀} be bounded. To span 𝑊ᵐ,ᵖ we also require a Strang-Fix condition on 𝜓. But we impose this condition only to order 𝑚 - 1, whereas earlier authors required order 𝑚.
Our spanning results are derived from explicit "Shannon" type sampling formulas that express an arbitrary function 𝑓 as a limit of linear combinations of the 𝜓(𝑎j 𝑥 - 𝑘). The coefficients in these sampling formulas are local averages of 𝑓, or pointwise values off when
𝑓 has some regularity.
Thu, 01 Jan 2004 00:00:00 GMThttp://hdl.handle.net/10092/127192004-01-01T00:00:00ZExperiments in high precision clock synchronisation.
http://hdl.handle.net/10092/12717
Experiments in high precision clock synchronisation.
Jones, Richard
Hardware clocks used in computers tend to drift away from the correct time. In
a distributed computer system, each hardware clock drifts away from the correct
time at a different rate. In or4er to synchronise the clocks of each computer to
the same time, some form of adjustment must be made to each clock. Internal
clock synchronisation, where clocks can be adjusted so that all computers are
synchronised to the same time relative to each other, is sufficient for some
applications. Other applications require, external clock synchronisation, clocks
are synchronised to some external time standard, such as Coordinated Universal
Time (UTC), as well as to each other. Synchronisation between computers in
the network is important for certain applications such as global ordering of
events occurring throughout the system.
Clock synchronisation can be implemented using extra hardware devices or by
software alone. Many algorithms proposed to synchronise the clocks in a distributed
computer system. Most of the proposed algorithms are deterministic.
Deterministic algorithms are based on message passing between nodes. They
guarantee a maximum deviation between clocks of the nodes but this maximum
deviation is limited by the maximum end-to-end message transmission delay
which can be very large.
Recently some probabilistic algorithms proposed. These algorithms are still
based on message passing but cannot guarantee a maximum deviation between
computers in the network. Instead they guarantee a maximum deviation
with a certain probability of failure. Such algorithms are not bound by the
same constraint as the deterministic algorithms and can achieve much closer
synchronisation.
In chapter 2 backround information on clock synchronisation and the methods
that have been published so far, at the end of the chapter the aims and objectives of the project will be given. Chapter 3 describes discoveries that were made
about the clock systems in the deparment. Chapter 4 gives information on the
first of the two algorithms studied in this paper. Chapter 5 describes a new
algorithm that has been implimented on the departmental suns. Chapter 6
describes further work that could carry on from this project. Chapter 7 gives
the conclusions to the project.
Fri, 01 Jan 1993 00:00:00 GMThttp://hdl.handle.net/10092/127171993-01-01T00:00:00Z