20060303-partial

1 files changed, 24 insertions, 0 deletions

diff --git a/module/linalg.m b/module/linalg.m
new file mode 100644
index 00000000..ccc3288d
--- /dev/null
+++ b/module/linalg.m
@@ -0,0 +1,24 @@
+# The convention used here for storing matrices is the same commonly
+# used for scientific programming in C, namely linearizing in Fortran order.
+# Let A be an m by n matrix.  We represent this by
+#	a: array of real;
+#	m, n, lda: int;
+# where the variable lda ("leading dimension of a") is used so that a
+# succession of matrix problems of varying sizes can be created without
+# wholesale copying of data.  The element of A in the i-th row and j-th column
+# is stored in a[i+lda*j], where 0<=i<m and 0<=j<n.  This 0-origin indexing
+# is used everywhere, and in particular in permutation vectors.
+
+LinAlg: module{
+	PATH:	con "/dis/math/linalg.dis";
+
+	Vector: type array of real;
+	Matrix: adt{
+		m, L, n: int;     # rows, column stride, columns
+		a: Vector; # data, stored A[i,j] = a[i+L*j]
+	};
+
+	dgefa:	fn(a:array of real, lda, n:int, ipvt:array of int): int;
+	dgesl:	fn(a:array of real, lda, n:int, ipvt:array of int, b:array of real, job:int);
+	printmat: fn(label:string, a:array of real, lda, m, n:int);
+};