Confusing behaviour with Compute FEP command

Hi everyone,

Lammps version: 3 June 2022
OS: Cent OS

I am trying to do some free energy calculations using FEP method using compute FEP command paired with the lj/cut/soft pair style in lammps. For this, I am running a simulation during which the \lambda parameter is decreased gradually after a fixed number of steps. And for each of the \lambda run, compute FEP command calculates the perturbation energy. This is accomplished using the following lines in input file:

variable simlength equal 20000
variable nsteps equal 5
variable nruns equal ${nsteps}+1
variable totalsimlength equal ${simlength}*${nruns}
variable samplefreq equal 10
variable l file Lambdas.txt
variable lambda equal next(l)
variable delta_lam_fwd equal 0.05
variable delta_lam_bwd equal -0.05

fix TVSTAT Rest nvt temp ${TK} ${TK} 100

fix FEP_vdW all adapt/fep ${simlength} pair lj/cut/soft lambda 1 1*6 v_lambda pair lj/cut/soft lambda 2 2*6 v_lambda scale no after yes
compute Fwd_FEP all fep ${TK} pair lj/cut/soft lambda 1 1*6 v_delta_lam_fwd pair lj/cut/soft lambda 2 2*6 v_delta_lam_fwd tail yes volume no
compute Bwd_FEP all fep ${TK} pair lj/cut/soft lambda 1 1*6 v_delta_lam_bwd pair lj/cut/soft lambda 2 2*6 v_delta_lam_bwd tail yes volume no

fix deltaU all ave/time 1 1 ${samplefreq} c_Fwd_FEP[1] c_Bwd_FEP[1] mode scalar file deltaUs_all.dat

thermo 1000
thermo_style custom step etotal ke pe evdwl temp vol press v_delta_lam_fwd v_delta_lam_bwd

So, what happens is, the simulation runs as expected till the second last \lambda value, but for the last one, the compute Fwd_FEP command starts to give values same as those for compute Bwd_FEP, which is caused by the value of the variable v_delta_lam_fwd changing to -0.05 instead of 0.05 as set in the input script, as printed using the thermo command:

   Step         TotEng         KinEng         PotEng         E_vdwl          Temp          Volume         Press      v_delta_lam_fwd v_delta_lam_bwd
...
 350093000  -86136.311      4273.3368     -90409.648      8735.9442      299.85692      75834.308      730.42089      0.05          -0.05
 350094000  -86106.398      4284.4323     -90390.83       8918.2387      300.63548      75834.308      1234.6366      0.05          -0.05
 350095000  -86135.691      4246.2832     -90381.974      8694.3414      297.95859      75834.308      510.14636      0.05          -0.05
 350096000  -85976.684      4261.9146     -90238.598      8660.8715      299.05544      75834.308      496.38937      0.05          -0.05
 350097000  -85919.287      4312.6694     -90231.956      8917.6138      302.61686      75834.308      1409.9869      0.05          -0.05
 350098000  -86058.619      4247.5722     -90306.191      8840.1815      298.04904      75834.308      1079.8458      0.05          -0.05
 350099000  -85896.704      4267.1954     -90163.899      8711.1602      299.42598      75834.308      820.38633      0.05          -0.05
 350100000  -85985.651      4268.8908     -90254.542      8670.4515      299.54495      75834.308      545.48806      0.05          -0.05      # second last lambda run ends here.
 350101000  -85966.823      4303.1725     -90269.995      8707.6271      301.95047      75834.308      706.52829     -0.05          -0.05
 350102000  -85982.031      4254.3892     -90236.42       8755.4484      298.52738      75834.308      811.1872      -0.05          -0.05
 350103000  -86021.924      4221.8595     -90243.783      8909.3352      296.24479      75834.308      1328.6266     -0.05          -0.05
 350104000  -86018.547      4242.1872     -90260.734      8780.0258      297.67118      75834.308      821.86939     -0.05          -0.05
 350105000  -86006.549      4258.5056     -90265.055      8685.4644      298.81623      75834.308      629.8689      -0.05          -0.05
 350106000  -85897.769      4280.0067     -90177.775      8701.0648      300.32494      75834.308      635.17761     -0.05          -0.05
 350107000  -86031.247      4208.3746     -90239.621      8684.3429      295.29857      75834.308      606.78614     -0.05          -0.05
 350108000  -85894.196      4251.2076     -90145.404      8779.1583      298.30413      75834.308      906.35704     -0.05          -0.05
 350109000  -85948.198      4193.3        -90141.498      8665.2021      294.2408       75834.308      456.46995     -0.05          -0.05
 350110000  -85846.401      4235.3346     -90081.735      8811.3486      297.19033      75834.308      1267.8322     -0.05          -0.05
 350111000  -85995.452      4256.681      -90252.133      8794.8207      298.68819      75834.308      946.96043     -0.05          -0.05
 350112000  -85961.304      4272.6076     -90233.911      8781.2769      299.80575      75834.308      928.04201     -0.05          -0.05
 350113000  -86201.837      4126.5484     -90328.385      8779.8829      289.55688      75834.308      835.88337     -0.05          -0.05
 350114000  -85968.249      4196.3182     -90164.567      8655.3571      294.45258      75834.308      532.46323     -0.05          -0.05
 350115000  -85897.515      4272.6101     -90170.125      8612.3793      299.80593      75834.308      487.07478     -0.05          -0.05
 350116000  -85855.7        4341.0402     -90196.74       8691.2443      304.60762      75834.308      571.6355      -0.05          -0.05
 350117000  -85963.257      4210.5771     -90173.834      8798.0102      295.45312      75834.308      999.98062     -0.05          -0.05
 350118000  -86018.397      4240.9741     -90259.371      8609.4535      297.58606      75834.308      320.38252     -0.05          -0.05
 350119000  -85891.813      4312.3972     -90204.21       8649.7025      302.59776      75834.308      633.75554     -0.05          -0.05
 350120000  -85944.851      4302.84       -90247.691      8664.7604      301.92714      75834.308      608.32034     -0.05          -0.05

The same behaviour is seen even if I run the simulation for just 1 \lambda value. Can someone help me identify what might be causing this? And what mistake I may be making.

I am attaching the input files as well.
data.n60_nocharge_eqm.lmp (1.2 MB) # input restart file
deltaUs_all.dat (315.8 KB) # fep data file (see last few lines)
in.lmp (2.1 KB) # input script
log.lammps (28.3 KB) # simulation log file
pair.lmp (3.1 KB) # pair coefficient definitions
Lambdas.txt (65 Bytes) # File from which \lambda values are read using next command.

Thanks, and regards,
Chaitanya