Unscaled variant of the DIRECT algorithm was given incorrect constant…

… name (#537)

The constant for the unscaled variant of the DIRECT algorithm was incorrectly given as NLOPT_GNL_DIRECT_NOSCAL, when it is in fact NLOPT_GN_DIRECT_NOSCAL. Fixed.

doc/docs/NLopt_Algorithms.md

@@ -51,7 +51,7 @@ First, it contains a from-scratch re-implementation of both algorithms, specifie
 
 Second, there is a slightly randomized variant of DIRECT-L, specified by `NLOPT_GN_DIRECT_L_RAND`, which uses some randomization to help decide which dimension to halve next in the case of near-ties.
 
-The DIRECT and DIRECT-L algorithms start by rescaling the bound constraints to a hypercube, which gives all dimensions equal weight in the search procedure. If your dimensions do *not* have equal weight, e.g. if you have a "long and skinny" search space and your function varies at about the same speed in all directions, it may be better to use unscaled variants of these algorthms, which are specified as `NLOPT_GNL_DIRECT_NOSCAL`, `NLOPT_GN_DIRECT_L_NOSCAL`, and `NLOPT_GN_DIRECT_L_RAND_NOSCAL`, respectively. However, the unscaled variations make the most sense (if any) with the original DIRECT algorithm, since the design of DIRECT-L to some extent relies on the search region being a hypercube (which causes the subdivided hyperrectangles to have only a small set of side lengths).
+The DIRECT and DIRECT-L algorithms start by rescaling the bound constraints to a hypercube, which gives all dimensions equal weight in the search procedure. If your dimensions do *not* have equal weight, e.g. if you have a "long and skinny" search space and your function varies at about the same speed in all directions, it may be better to use unscaled variants of these algorthms, which are specified as `NLOPT_GN_DIRECT_NOSCAL`, `NLOPT_GN_DIRECT_L_NOSCAL`, and `NLOPT_GN_DIRECT_L_RAND_NOSCAL`, respectively. However, the unscaled variations make the most sense (if any) with the original DIRECT algorithm, since the design of DIRECT-L to some extent relies on the search region being a hypercube (which causes the subdivided hyperrectangles to have only a small set of side lengths).
 
 Finally, NLopt also includes separate implementations based on the [original Fortran code](http://www4.ncsu.edu/~ctk/SOFTWARE/DIRECTv204.tar.gz) by Gablonsky et al. (1998-2001), which are specified as `NLOPT_GN_ORIG_DIRECT` and `NLOPT_GN_ORIG_DIRECT_L`. These implementations have a number of hard-coded limitations on things like the number of function evaluations; I removed several of these limitations, but some remain. On the other hand, there seem to be slight differences between these implementations and mine; most of the time, the performance is roughly similar, but occasionally Gablonsky's implementation will do significantly better than mine or vice versa.