Some strange behavior on HPC

[jakobbossek]

I get the following strange behavior on a HPC for method = "ta" producing discrepancy values between 1 and 2. However, this does not happen locally.
Note: it seems that this happens in the first phase of the TA algorithm only.

Trial 9/10
1.56809 at 2 : 109 114 96 109 107 116 117 111 106 111
1.60991 at 4 : 114 114 97 109 114 116 111 111 106 111
1.64886 at 5 : 114 114 97 109 114 116 111 111 116 111
1.66496 at 7 : 114 108 105 109 114 116 111 111 116 111
1.69675 at 10 : 114 115 105 109 114 116 111 111 114 111
1.72251 at 75 : 115 103 110 114 116 114 111 110 114 117
1.72667 at 285 : 115 117 112 115 110 113 106 113 114 111
1.73547 at 312 : 101 115 115 116 114 113 116 113 112 115
1.74633 at 582 : 115 117 110 115 113 117 112 99 112 117
1.74775 at 592 : 115 109 115 113 113 107 109 116 114 117
1.76164 at 813 : 112 110 115 113 110 115 112 114 112 115
1.78077 at 999 : 112 117 115 114 116 109 112 114 112 113
1.80121 at 1148 : 114 114 114 116 114 116 111 113 110 117