File Format

The file format consists of a number of fields, each with a text label and field length in bytes^5.1, followed by linefeed, then the data. Text labels are not case-dependent and contain no whitespace, though they may be proceeded by whitespace (spaces, tabs, CR, LF), and there must be whitespace between the label and bytecount. Text fields may include multiple lines.

Currently supported labels include:

• Author The author(s) of the file. If there are multiple Author fields, the last one will be official.
• Title The title of the file. If there are multiple Title fields, the last one will be official.
• Description The text description of the file. If there are multiple Description fields, the last one will be official.
• Comment A comment, can be inserted anywhere in the file.
• Dimensions This must come before Nodes, and just gives the integer number of dimensions in the problem.
• Nodes Text information on nodes. Each node consists of its integer group number (can be any integer) and D floating point numbers giving its coordinates. Node numbers are assigned sequentially starting with zero, and multiple Nodes fields just add to the end of the node list.
• N-Elements List of integer node numbers that make up each surface element. Interpolation is done using N-order polynomials, with $\frac{(N+D)!}{(N-1)!D!}$ evenly-spaced nodes per element (assuming 0!=-1!=1). Elements are numbered the same way as nodes are, and it is possible to mix several different orders of elements in a given model.
• Volume List of surface elements that define a closed volume. Elements listed with a dash in front of them (including -0) are switched to point outward.
• Equation List of integer volume index numbers and equation names to apply to them. Each equation comes with a defined number of variables per node.
• BoundaryCond List of boundary conditions, each of which consists of a group number, a variable number, and the boundary condition type and parameters. These are written as (case-insensitive):
  • D(a) for constant Dirichlet value u=a.
  • F(function) for Dirichlet BC given by a function of node position.
  • N(a) for constant outward flux a.
  • G(function) for outward flux as a function of node position.
  • R(a,b) for linear Robin BC giving outward flux as a+bu.
  • L(function1, function2) for linear Robin BC giving outward flux as function1+function2u where function1 and function2 take node position as arguments.
  • P(function) for generalized Robin BC giving outward flux as a function of node position and u.
  • S(string) for special boundary conditions unique to equations. The string is passed to the elemental matrix builder and the boundary condition maker, which are unique to each Equation.
  Functions are not yet implemented, so only D, N, R and S are presently available. Function syntax will be determined at a later time. Any number of these fields can be used; if more than one boundary condition is supplied for a given node group and variable then the last one will be used.
• DMatrix LU-decomposed matrix for this problem, stored in long double-precision format in little endian byte order.
• IMatrix Inverted matrix for this problem, same format.