If I understand correctly what you want to do (basically join by minimum distance) there are quite a few ways of doing this (kdtable, vptree, loc, distance, addins, ...). Below is one example
Names Default To Here(1);
dt_lookup = New Table("Untitled",
Add Rows(3),
Compress File When Saved(1),
New Column("Column 1", Numeric, "Continuous", Format("Best", 12), Set Values([3.8, 4.8, 4])),
New Column("Column 2", Numeric, "Continuous", Format("Best", 12), Set Values([50, 52, 51]))
);
dt = New Table("Untitled",
Add Rows(3),
Compress File When Saved(1),
New Column("Column 1", Numeric, "Continuous", Format("Best", 12), Set Values([3.801, 4.801, 4.000, 3.7, 4.802, 4.799]))
);
lookup_values = Column(dt_lookup, "Column 2") << get values;
search_values = Column(dt_lookup, "Column 1") << get values;
dt << New Column("Column 2", Numeric, Continuous, << Set Each Value(
As Constant(tab = KDTable(search_values););
{rows, dist} = tab << K nearest rows(1, Matrix(:Column 1));
lookup_values[rows[1]];
));
Below are some demonstrations for other options using matrices
Names Default To Here(1);
lookup_values = [50, 52, 51];
search_values = [3.8, 4.8, 4];
search_val = 4.3;
// These should return row from dt_lookup
kdt = KDTable(search_values);
{kd_rows, kd_dist} = kdt << K nearest rows(1, Matrix(search_val));
vpt = VPTree(search_values);
{vp_rows, vp_dist} = vpt << K nearest rows(1, Matrix(search_val));
dist = Distance(search_values, Matrix(search_val));
lmin_dist = LocMin(dist);
abs_calc = Abs(search_values - search_val);
lmin_abs = LocMin(abs_calc);
show(dist, lmin_dist, abs_calc, lmin_abs, kd_rows, kd_dist, vp_rows, vp_dist);
-Jarmo