Tf type

tf_type

Initial value: 2

Type of target function used for distance restraints. The same functional form is used for upper limits, lower limits, and van der Waals lower limits. type term for a violated upper limit () limiting cases

                small violation ()     large violation (d ' b)

1

2

3

4

d, b and b denote the actual distance, the upper distance bound, and the value of the system variable tf_beta. The larger the value of b, the longer the functional form will be close to the limiting case for small violations. Type 1 is the normal Diana target function (Guentert et al., 1991), and type 3 is the error-tolerant target function used by Noah (Mumenthaler et al., 1997). Target functions of type 1, 2 and 4 have unit A2, whereas the target function of type 3 is dimensionless, i.e. target function values obtained with type 3 cannot be compared with those obtained with other types. In all cases the contribution to the target function from a small violation is proportional to the square of the violation. For large violations, the target function types differ significantly: type 1 is proportional to d4, types 2 and 3 are proportional to d2, and type 4 is linear in d. Note that distance restraints with very small upper bound b can lead to problems when the target function of type 1 is used because they get an excessive weight over other restraints. To illustrate this, assume two upper limit distance restraints that are violated by the same amount:

       b = 0.1 A, d = 2 A: target function contribution = 398.0 A2
       b = 3.1 A, d = 5 A: target function contribution = 6.2 A2

The first restraint gives a more than 60 times larger contribution than the second! In such cases it is advisable to use the target function of type 2, to which both restraints would contribute the same amount.