Page 1 of 1

Bugs in Mathematica on Raspberry Pi 3?

Posted: Wed Mar 08, 2017 7:01 pm
by xxandra
Replaying a sample of examples from the Wolfram Language & System Documentation Center on my newly acquired Raspberry Pi 3, I discovered the following:
The results of calculations on my Raspberry Pi are mostly the same as those shown in the Documentation Center. However, there are notable exceptions. For example, the plots of the functions of the NDEigensystem-code

Code: Select all

{vals, funs} =   NDEigensystem[{-Laplacian[u[x, y], {x, y}],     DirichletCondition[u[x, y] == 0, True]},    u[x, y], {x, y} \[Element] Disk[], 6];

are distinctively different from the plots of these functions in the Documentation Center. Which is in striking contrast to the plots of the comparable DEigensystem-code

Code: Select all

{vals, funs} =   DEigensystem[{-Laplacian[u[x, y], {x, y}],     DirichletCondition[u[x, y] == 0, True]},    u[x, y], {x, y} \[Element] Disk[], 6];

The DEigensystem-plots on my Raspberry are the same as in the Documentation Center. That is to say, under version 10.3.1. Under version, however, four of the plots show white blots on their 3D surfaces, as if there were discontinuities in the functions.

Further, plots of the Eigenfunctions in an L-shaped region produced from

Code: Select all

 L = Polygon[{{1, 0}, {2, 0}, {2, 2}, {0, 2}, {0, 1}, {1, 1}}]; {vals, funs} =   NDEigensystem[{Laplacian[u[x, y], {x, y}],     DirichletCondition[u[x, y] == 0., True]},    u[x, y], {x, y} \[Element] L, 6];
look like plots of minus the Eigenfunctions. Possibly as a consequence of the warning (shown on my Raspberry, but not in the Documentation Center) that "the second matrix (..) is not positive definite".

It should also be noted that plotting the NDEigensystem-functions takes less than 20 seconds under version 10.3.1, but about half an hour under version!

One wonders, how reliable is Mathematica's implementation on Raspberry Pi?

N.B. Results under Raspbian version d.d. 2016-11-25