Hello, I'm experiencing an issue where the Line Search optimization for my set up takes an extremely long amount of time and seems to iterate well past 1,000,000 values for k. I set up my experiment exactly the same as in the relax_fit script.
Here's the exact spot where the error occurs. Fitting to spin '#Ap4Aase_new_3_mol1:3@N' ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Newton minimisation ~~~~~~~~~~~~~~~~~~~ Line search: Backtracking line search. Hessian modification: The Gill, Murray, and Wright modified Cholesky algorithm. k: 0 xk: [ 20, 2000] fk: 10.5392026372 k: 100 xk: [ 494.4015814, 20474015.74] fk: 6.96842623012 k: 200 xk: [ 517.0492257, 30502358.26] fk: 6.9683206805 k: 300 xk: [ 527.0537514, 36375387.07] fk: 6.96828589187 k: 400 xk: [ 533.5623637, 40790463.33] fk: 6.96826633601 k: 500 xk: [ 538.3959132, 44412527.99] fk: 6.96825319617 k: 600 xk: [ 542.2426565, 47523593.85] fk: 6.96824350921 k: 700 xk: [ 545.4380817, 50272959.33] fk: 6.96823594648 k: 800 xk: [ 548.1711776, 52750386.01] fk: 6.96822980706 k: 900 xk: [ 550.5589954, 55014551.85] fk: 6.96822467975 k: 1000 xk: [ 552.6790841, 57106170.59] fk: 6.96822030448 I've also noticed that my run will optimize at an extra spin at residue 3 whereas the demo run does not load that spin during execution. Is this a setup issue on my part? I'm also running relax on macOS. Best, Kapil
_______________________________________________ nmr-relax-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nmr-relax-users
