Dear colleagues,

I need to determine a linear boundary of a data cloud. More specifically:

I have a set of experimental points y(x). On the plane x-y, the data points form a cloud of a triangular shape, with the legs coinciding with the x and y axes.  I need to best-fit the outer (upper-right), boundary of the data cloud, i.e. the hypotenuse of the "triangle",    with a linear function. It is not necessary for the line to go through the boundary points exactly, I need a best-fitting approximation (actually, the gradient of the line). Could someone give me any idea how to do it? (Or, at least, advice how I can google it, as my searches like "approximate boundary data cloud" return loads of irrelevant stuff.

Thanks in advance!

Perhaps the question is trivial, but I could not find the solution.

I am solving numerically an ODE (e.g., a simple harmonic oscillator) with the righthand side that contains a random part. For example it is

eq:=diff(a(t),t,t) + a(t) = (1+0.01*R(t))*cos(5*t);

where R(t) is a random function of t. How can I make such a function?

The naive attempt: 

eq:=diff(a(t),t,t) + a(t) = 3.*(1+0.1*rand()/1000000000000.)*cos(5.*t);

gave me a fixed (while random) value, e.g.

...

But, I need this coefficeint to be random each time step.

Any suggestions are very welcome!

I am trying to use Maple to study a 1D wave equation with non-uniform coefficients. In general the problem does not have analytical solutions, hence I am using pdsolve(..., numeric). As the first test I am looking at the development of a Gaussian pulse in a uniform medium, doing:

PDE := diff(u(x, t), t, t) = diff(u(x, t), x, x);

IBC := {u(-10, t) = exp(-(-10)^2), u(10, t) = exp(-10^2), u(x, 0) = exp(-x^2), (D[2](u))(x, 0) = 0};

pds := pdsolve(PDE, IBC,...