Different gradient calculations in a logistic regression

Question

I am trying to find how one variable (EVI) may predict a binary outcome (an_larv_bin) using glmer from lme4 package. The code I input is:

univ_points_evi <- glmer(an_larv_bin ~ EVI + (1|grid_no), family="binomial", data=univ_points)

and I come up against the following warning message:

In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
  Model failed to converge with max|grad| = 0.0331857 (tol = 0.002, component 1)

I have rescaled and centered the variable as such:

scale(EVI, center = TRUE, scale = TRUE)

And still receive the same warning. The next steps I took from Rstudio troubleshooting for lme4 convergence warnings. I checked for singularity, but it was not a problem:

> tt <- getME(univ_points_evi,"theta")
> ll <- getME(univ_points_evi,"lower")
> min(tt[ll==0])
[1] 0.80632

I restarted the function from a different fit:

ss <- getME(univ_points_evi,c("theta","fixef"))
m2 <- update(univ_points_evi,start=ss,control=glmerControl(optCtrl=list(maxfun=2e4)))

And get the same warning again. I checked the gradient calculations and got the embedded results:

#pre-computed information 
derivs1 <- univ_points_evi@optinfo$derivs
sc_grad1 <- with(derivs1,solve(Hessian,gradient))
max(abs(sc_grad1))
[1] 4.991948e-05


max(pmin(abs(sc_grad1),abs(derivs1$gradient)))
[1] 4.991948e-05

#numDeriv
dd <- update(univ_points_evi,devFunOnly=TRUE)
pars <- unlist(getME(univ_points_evi,c("theta","fixef")))
grad2 <- grad(dd,pars)
hess2 <- hessian(dd,pars)
sc_grad2 <- solve(hess2,grad2)
max(pmin(abs(sc_grad2),abs(grad2)))
[1] 3.308277

The third figure is obviously quite different from the first two. I am unsure what this is actually telling me, and whether it is the root of the warning?

Lastly, I checked the fit of all optimisers:

univ_points_evi.all <- allFit(univ_points_evi)

And got the following:

bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]

And tried all available optimisers, which converge to values that are practically equivalent.

ss <- summary(univ_points_evi.all)
ss$ fixef
ss$ llik
ss$ sdcor
ss$ theta

The lme4 package guidelines state that they would in this case consider the convergence warnings to be false positives. I am wondering whether this is true in my case, or if there is an issue with the gradient calculations which is causing the issue.

My data:

> dput(EVI)
c(0.499929994, 0.589900017, 0.593994021, 0.589900017, 0.601158023, 
0.492922992, 0.546519995, 0.601045012, 0.536565006, 0.592272997, 
0.592227995, 0.645565987, 0.61619997, 0.516200006, 0.516200006, 
0.4639, 0.4639, 0.561200023, 0.5898, 0.564800024, 0.5898, 0.5898, 
0.605099976, 0.595300019, 0.545300007, 0.572000027, 0.599600017, 
0.585300028, 0.591700017, 0.533399999, 0.552100003, 0.569700003, 
0.592499971, 0.596199989, 0.53490001, 0.53490001, 0.53490001, 
0.553300023, 0.582899988, 0.545000017, 0.592100024, 0.582899988, 
0.59009999, 0.569299996, 0.612900019, 0.533500016, 0.583299994, 
0.772599995, 0.772599995, 0.682500005, 0.682500005, 0.682500005, 
0.772599995, 0.628099978, 0.626299977, 0.628099978, 0.747399986, 
0.640200019, 0.531899989, 0.680199981, 0.535099983, 0.680199981, 
0.535099983, 0.565299988, 0.680199981, 0.703199983, 0.703199983, 
0.541700006, 0.678200006, 0.678200006, 0.547100008, 0.634899974, 
0.696399987, 0.688199997, 0.574899971, 0.574899971, 0.669799984, 
0.611000001, 0.61559999, 0.639100015, 0.669799984, 0.669799984, 
0.611000001, 0.59890002, 0.639100015, 0.604799986, 0.604799986, 
0.604799986, 0.606599987, 0.606599987, 0.640600026, 0.624899983, 
0.640600026, 0.624899983, 0.624899983, 0.640600026, 0.640600026, 
0.516200006, 0.507499993, 0.507499993, 0.46540001, 0.530300021, 
0.530300021, 0.565100014, 0.546599984, 0.546599984, 0.530399978, 
0.530399978, 0.530399978, 0.523199975, 0.523199975, 0.546400011, 
0.546599984, 0.496600002, 0.530799985, 0.537800014, 0.545000017, 
0.496600002, 0.496600002, 0.514100015, 0.530799985, 0.530799985, 
0.537800014, 0.530200005, 0.530200005, 0.546599984, 0.546599984, 
0.576399982, 0.46540001, 0.516200006, 0.530399978, 0.655300021, 
0.680999994, 0.660000026, 0.661499977, 0.661499977, 0.680999994, 
0.655300021, 0.617799997, 0.647099972, 0.647099972, 0.617799997, 
0.673300028, 0.673300028, 0.507700026, 0.507700026, 0.507700026, 
0.651799977, 0.591799974, 0.591799974, 0.688300014, 0.661499977, 
0.661499977, 0.661499977, 0.661499977, 0.648500025, 0.648500025, 
0.495799989, 0.495799989, 0.495799989, 0.648899972, 0.648899972, 
0.673300028, 0.673300028, 0.648500025, 0.647099972, 0.691999972, 
0.647099972, 0.647099972, 0.617799997, 0.657199979, 0.706499994, 
0.591799974, 0.661499977, 0.661499977, 0.641600013, 0.648500025, 
0.648500025, 0.688300014, 0.495799989, 0.495799989, 0.688300014, 
0.582000017, 0.582000017, 0.57069999, 0.582000017, 0.62559998, 
0.565500021, 0.565500021, 0.62559998, 0.593599975, 0.604700029, 
0.599699974, 0.536800027, 0.600300014, 0.600300014, 0.604700029, 
0.566699982, 0.566699982, 0.626900017, 0.626900017, 0.594900012, 
0.594900012, 0.584500015, 0.586199999, 0.605700016, 0.584699988, 
0.553799987, 0.542900026, 0.584699988, 0.584699988, 0.575399995, 
0.579999983, 0.579299986, 0.596899986, 0.594900012, 0.565500021, 
0.579299986, 0.594900012, 0.565500021, 0.549499989, 0.549499989, 
0.549499989, 0.549499989, 0.606899977, 0.539600015, 0.584699988, 
0.571699977, 0.56129998, 0.595600009, 0.62559998, 0.565500021, 
0.565500021, 0.620299995, 0.620299995, 0.594900012, 0.579999983, 
0.654299974, 0.654299974, 0.627600014, 0.627600014, 0.64349997, 
0.687699974, 0.64349997, 0.59859997, 0.59859997, 0.649999976, 
0.518299997, 0.658299983, 0.658299983, 0.658299983, 0.627600014, 
0.658299983, 0.658299983, 0.627600014, 0.667500019, 0.653100014, 
0.564899981, 0.561999977, 0.629000008, 0.639999986, 0.639999986, 
0.675100029, 0.675100029, 0.658299983, 0.659300029, 0.658299983, 
0.659300029, 0.657400012, 0.645299971, 0.425599992, 0.425599992, 
0.474299997, 0.598800004, 0.595200002, 0.416399986, 0.564899981, 
0.564899981, 0.70599997, 0.70599997, 0.664699972, 0.484299988, 
0.496199995, 0.496199995, 0.484299988, 0.517499983, 0.517499983, 
0.517499983, 0.535899997, 0.51730001, 0.562399983, 0.540000021, 
0.540000021, 0.501299977, 0.501299977, 0.528599977, 0.532400012, 
0.51730001, 0.562399983, 0.501299977, 0.574299991, 0.528599977, 
0.528599977, 0.528599977, 0.503499985, 0.568700016, 0.521799982, 
0.503499985, 0.521799982, 0.557699978, 0.557699978, 0.545099974, 
0.532400012, 0.563399971, 0.530700028, 0.431100011, 0.431100011, 
0.510900021, 0.556400001, 0.501299977, 0.48120001, 0.48120001, 
0.528800011, 0.528800011, 0.62470001, 0.62470001, 0.707899988, 
0.707899988, 0.62529999, 0.62529999, 0.630500019, 0.646300018, 
0.604900002, 0.62529999, 0.669799984, 0.634199977, 0.634199977, 
0.634199977, 0.612999976, 0.662400007, 0.698700011, 0.632799983, 
0.682099998, 0.428499997, 0.513300002, 0.569700003, 0.519500017, 
0.519500017, 0.48120001, 0.48120001, 0.646399975, 0.559899986, 
0.564899981, 0.564899981, 0.564899981, 0.602699995, 0.602699995, 
0.60650003, 0.575699985, 0.5722, 0.584299982, 0.584900022, 0.584900022, 
0.5722, 0.584299982, 0.5722, 0.560699999, 0.560699999)
> dput(an_larv_bin)
c(1L, 0L, 1L, 1L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 1L, 0L, 0L, 1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 0L, 
0L, 0L, 1L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 1L, 1L, 1L, 1L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 
1L, 0L, 0L, 0L, 1L, 1L, 1L, 0L, 1L, 1L, 0L, 0L, 1L, 0L, 0L, 0L, 
0L, 1L, 1L, 1L, 1L, 1L, 0L, 1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 
1L, 1L, 0L, 1L, 1L, 1L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 1L, 1L, 1L, 
0L, 1L, 0L, 0L, 0L, 0L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 0L, 
0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 1L, 1L, 
1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 1L, 1L, 1L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 1L, 1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 1L, 0L, 1L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 0L, 
0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 1L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 
0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 1L, 0L, 
1L)



 > dput(grid_no)
c(4937L, 3270L, 2854L, 3270L, 2582L, 2584L, 2585L, 2584L, 4663L, 
3416L, 3416L, 3979L, 2986L, 4839L, 4839L, 4937L, 4937L, 3264L, 
2854L, 2854L, 2289L, 2289L, 2582L, 3978L, 3834L, 3416L, 3416L, 
3547L, 3687L, 2852L, 4388L, 4388L, 4538L, 4538L, 4937L, 4937L, 
4937L, 2854L, 2854L, 2996L, 2996L, 2289L, 2582L, 3416L, 3692L, 
2983L, 2301L, 4937L, 3264L, 3264L, 3547L, 3547L, 3547L, 3264L, 
3822L, 3683L, 3683L, 3678L, 2427L, 2427L, 2427L, 2289L, 2427L, 
2289L, 4117L, 2710L, 2582L, 2303L, 2854L, 2854L, 4520L, 3692L, 
3692L, 3416L, 4526L, 4527L, 3264L, 3685L, 3685L, 4937L, 3264L, 
3264L, 3685L, 4801L, 4937L, 2290L, 2289L, 2289L, 2854L, 2854L, 
2581L, 2719L, 2719L, 2578L, 2578L, 2582L, 2581L, 3416L, 3978L, 
3978L, 3416L, 3549L, 3549L, 2986L, 2700L, 2700L, 4680L, 4680L, 
4680L, 4670L, 4670L, 2428L, 4527L, 3264L, 2854L, 4937L, 2582L, 
3264L, 3264L, 3264L, 2854L, 2854L, 4937L, 2289L, 2289L, 4527L, 
4680L, 4680L, 3416L, 3416L, 4680L, 3409L, 3547L, 3685L, 3685L, 
3685L, 3409L, 3547L, 2861L, 2581L, 2578L, 2861L, 2430L, 2430L, 
2293L, 2293L, 2293L, 3977L, 3684L, 4523L, 4669L, 3264L, 3264L, 
3264L, 3264L, 2854L, 2854L, 2289L, 2289L, 2289L, 2577L, 2577L, 
4937L, 4937L, 2577L, 2582L, 2582L, 2578L, 2578L, 3416L, 3416L, 
4527L, 4801L, 3685L, 3822L, 2302L, 2855L, 2855L, 4669L, 2287L, 
2287L, 4669L, 3549L, 3549L, 4798L, 3549L, 4680L, 4680L, 4680L, 
4822L, 4258L, 4948L, 3273L, 4677L, 4677L, 4677L, 4948L, 2854L, 
2854L, 3264L, 3264L, 3264L, 4937L, 4937L, 2582L, 2582L, 2578L, 
2578L, 2289L, 2289L, 2289L, 3416L, 2573L, 3416L, 4527L, 3685L, 
3547L, 4801L, 3685L, 3547L, 2287L, 2287L, 2287L, 2287L, 2436L, 
2291L, 2718L, 2718L, 4099L, 3131L, 4680L, 4680L, 4680L, 3260L, 
3260L, 3977L, 2571L, 2578L, 2578L, 2854L, 2854L, 3264L, 3264L, 
3264L, 4937L, 4937L, 2582L, 2582L, 2289L, 2289L, 2289L, 2573L, 
2573L, 2573L, 2573L, 3132L, 3407L, 3416L, 3416L, 3685L, 3685L, 
3685L, 4527L, 4801L, 2991L, 2287L, 2287L, 2426L, 3399L, 2301L, 
4680L, 4680L, 4680L, 4541L, 4390L, 3277L, 3277L, 3277L, 3978L, 
3978L, 3978L, 4937L, 4801L, 4801L, 4937L, 2289L, 2289L, 2289L, 
2573L, 2854L, 3264L, 3264L, 3264L, 3684L, 3684L, 2582L, 2582L, 
2854L, 3264L, 3684L, 4527L, 2578L, 2578L, 2718L, 2718L, 2296L, 
4665L, 4665L, 4665L, 3416L, 3416L, 3277L, 3277L, 2443L, 2300L, 
2302L, 4680L, 4680L, 4680L, 3546L, 3546L, 4937L, 4937L, 4801L, 
4801L, 2854L, 2854L, 3264L, 3264L, 2289L, 2289L, 2582L, 2582L, 
2578L, 2289L, 3416L, 3416L, 3416L, 3556L, 3277L, 3685L, 3978L, 
4680L, 4110L, 4237L, 4527L, 4237L, 4937L, 4937L, 4801L, 4801L, 
3264L, 3685L, 3416L, 3416L, 3416L, 2289L, 2289L, 2289L, 2582L, 
2578L, 2582L, 2293L, 2857L, 2721L, 2443L, 4680L, 4680L)

Thanks in advance!

user_8684551 · Answer

data:

df <- structure(list(an_larv_bin = c(1L, 0L, 1L, 1L, 0L, 0L, 0L, 1L,
1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 1L, 1L, 1L, 1L, 0L,
0L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 1L, 0L,
0L, 0L, 1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 0L, 0L, 0L, 1L, 1L, 1L, 0L, 1L,
1L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 1L, 1L, 1L, 1L, 1L, 0L, 1L, 1L,
1L, 1L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 0L, 1L, 1L, 1L, 0L, 0L, 0L,
1L, 1L, 0L, 0L, 1L, 1L, 1L, 0L, 1L, 0L, 0L, 0L, 0L, 1L, 1L, 1L,
1L, 1L, 1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L,
1L, 1L, 0L, 0L, 0L, 1L, 1L, 1L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 1L, 0L, 0L,
1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 1L, 1L, 1L, 1L, 0L, 0L, 0L,
0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 1L, 0L,
0L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 1L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 0L, 1L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 1L, 0L, 0L, 0L,
0L, 0L, 1L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L, 0L,
0L, 1L, 1L, 0L, 0L, 1L, 0L, 1L), EVI = c(0.499929994, 0.589900017,
0.593994021, 0.589900017, 0.601158023, 0.492922992, 0.546519995,
0.601045012, 0.536565006, 0.592272997, 0.592227995, 0.645565987,
0.61619997, 0.516200006, 0.516200006, 0.4639, 0.4639, 0.561200023,
0.5898, 0.564800024, 0.5898, 0.5898, 0.605099976, 0.595300019,
0.545300007, 0.572000027, 0.599600017, 0.585300028, 0.591700017,
0.533399999, 0.552100003, 0.569700003, 0.592499971, 0.596199989,
0.53490001, 0.53490001, 0.53490001, 0.553300023, 0.582899988,
0.545000017, 0.592100024, 0.582899988, 0.59009999, 0.569299996,
0.612900019, 0.533500016, 0.583299994, 0.772599995, 0.772599995,
0.682500005, 0.682500005, 0.682500005, 0.772599995, 0.628099978,
0.626299977, 0.628099978, 0.747399986, 0.640200019, 0.531899989,
0.680199981, 0.535099983, 0.680199981, 0.535099983, 0.565299988,
0.680199981, 0.703199983, 0.703199983, 0.541700006, 0.678200006,
0.678200006, 0.547100008, 0.634899974, 0.696399987, 0.688199997,
0.574899971, 0.574899971, 0.669799984, 0.611000001, 0.61559999,
0.639100015, 0.669799984, 0.669799984, 0.611000001, 0.59890002,
0.639100015, 0.604799986, 0.604799986, 0.604799986, 0.606599987,
0.606599987, 0.640600026, 0.624899983, 0.640600026, 0.624899983,
0.624899983, 0.640600026, 0.640600026, 0.516200006, 0.507499993,
0.507499993, 0.46540001, 0.530300021, 0.530300021, 0.565100014,
0.546599984, 0.546599984, 0.530399978, 0.530399978, 0.530399978,
0.523199975, 0.523199975, 0.546400011, 0.546599984, 0.496600002,
0.530799985, 0.537800014, 0.545000017, 0.496600002, 0.496600002,
0.514100015, 0.530799985, 0.530799985, 0.537800014, 0.530200005,
0.530200005, 0.546599984, 0.546599984, 0.576399982, 0.46540001,
0.516200006, 0.530399978, 0.655300021, 0.680999994, 0.660000026,
0.661499977, 0.661499977, 0.680999994, 0.655300021, 0.617799997,
0.647099972, 0.647099972, 0.617799997, 0.673300028, 0.673300028,
0.507700026, 0.507700026, 0.507700026, 0.651799977, 0.591799974,
0.591799974, 0.688300014, 0.661499977, 0.661499977, 0.661499977,
0.661499977, 0.648500025, 0.648500025, 0.495799989, 0.495799989,
0.495799989, 0.648899972, 0.648899972, 0.673300028, 0.673300028,
0.648500025, 0.647099972, 0.691999972, 0.647099972, 0.647099972,
0.617799997, 0.657199979, 0.706499994, 0.591799974, 0.661499977,
0.661499977, 0.641600013, 0.648500025, 0.648500025, 0.688300014,
0.495799989, 0.495799989, 0.688300014, 0.582000017, 0.582000017,
0.57069999, 0.582000017, 0.62559998, 0.565500021, 0.565500021,
0.62559998, 0.593599975, 0.604700029, 0.599699974, 0.536800027,
0.600300014, 0.600300014, 0.604700029, 0.566699982, 0.566699982,
0.626900017, 0.626900017, 0.594900012, 0.594900012, 0.584500015,
0.586199999, 0.605700016, 0.584699988, 0.553799987, 0.542900026,
0.584699988, 0.584699988, 0.575399995, 0.579999983, 0.579299986,
0.596899986, 0.594900012, 0.565500021, 0.579299986, 0.594900012,
0.565500021, 0.549499989, 0.549499989, 0.549499989, 0.549499989,
0.606899977, 0.539600015, 0.584699988, 0.571699977, 0.56129998,
0.595600009, 0.62559998, 0.565500021, 0.565500021, 0.620299995,
0.620299995, 0.594900012, 0.579999983, 0.654299974, 0.654299974,
0.627600014, 0.627600014, 0.64349997, 0.687699974, 0.64349997,
0.59859997, 0.59859997, 0.649999976, 0.518299997, 0.658299983,
0.658299983, 0.658299983, 0.627600014, 0.658299983, 0.658299983,
0.627600014, 0.667500019, 0.653100014, 0.564899981, 0.561999977,
0.629000008, 0.639999986, 0.639999986, 0.675100029, 0.675100029,
0.658299983, 0.659300029, 0.658299983, 0.659300029, 0.657400012,
0.645299971, 0.425599992, 0.425599992, 0.474299997, 0.598800004,
0.595200002, 0.416399986, 0.564899981, 0.564899981, 0.70599997,
0.70599997, 0.664699972, 0.484299988, 0.496199995, 0.496199995,
0.484299988, 0.517499983, 0.517499983, 0.517499983, 0.535899997,
0.51730001, 0.562399983, 0.540000021, 0.540000021, 0.501299977,
0.501299977, 0.528599977, 0.532400012, 0.51730001, 0.562399983,
0.501299977, 0.574299991, 0.528599977, 0.528599977, 0.528599977,
0.503499985, 0.568700016, 0.521799982, 0.503499985, 0.521799982,
0.557699978, 0.557699978, 0.545099974, 0.532400012, 0.563399971,
0.530700028, 0.431100011, 0.431100011, 0.510900021, 0.556400001,
0.501299977, 0.48120001, 0.48120001, 0.528800011, 0.528800011,
0.62470001, 0.62470001, 0.707899988, 0.707899988, 0.62529999,
0.62529999, 0.630500019, 0.646300018, 0.604900002, 0.62529999,
0.669799984, 0.634199977, 0.634199977, 0.634199977, 0.612999976,
0.662400007, 0.698700011, 0.632799983, 0.682099998, 0.428499997,
0.513300002, 0.569700003, 0.519500017, 0.519500017, 0.48120001,
0.48120001, 0.646399975, 0.559899986, 0.564899981, 0.564899981,
0.564899981, 0.602699995, 0.602699995, 0.60650003, 0.575699985,
0.5722, 0.584299982, 0.584900022, 0.584900022, 0.5722, 0.584299982,
0.5722, 0.560699999, 0.560699999), grid_no = c(4937L, 3270L,
2854L, 3270L, 2582L, 2584L, 2585L, 2584L, 4663L, 3416L, 3416L,
3979L, 2986L, 4839L, 4839L, 4937L, 4937L, 3264L, 2854L, 2854L,
2289L, 2289L, 2582L, 3978L, 3834L, 3416L, 3416L, 3547L, 3687L,
2852L, 4388L, 4388L, 4538L, 4538L, 4937L, 4937L, 4937L, 2854L,
2854L, 2996L, 2996L, 2289L, 2582L, 3416L, 3692L, 2983L, 2301L,
4937L, 3264L, 3264L, 3547L, 3547L, 3547L, 3264L, 3822L, 3683L,
3683L, 3678L, 2427L, 2427L, 2427L, 2289L, 2427L, 2289L, 4117L,
2710L, 2582L, 2303L, 2854L, 2854L, 4520L, 3692L, 3692L, 3416L,
4526L, 4527L, 3264L, 3685L, 3685L, 4937L, 3264L, 3264L, 3685L,
4801L, 4937L, 2290L, 2289L, 2289L, 2854L, 2854L, 2581L, 2719L,
2719L, 2578L, 2578L, 2582L, 2581L, 3416L, 3978L, 3978L, 3416L,
3549L, 3549L, 2986L, 2700L, 2700L, 4680L, 4680L, 4680L, 4670L,
4670L, 2428L, 4527L, 3264L, 2854L, 4937L, 2582L, 3264L, 3264L,
3264L, 2854L, 2854L, 4937L, 2289L, 2289L, 4527L, 4680L, 4680L,
3416L, 3416L, 4680L, 3409L, 3547L, 3685L, 3685L, 3685L, 3409L,
3547L, 2861L, 2581L, 2578L, 2861L, 2430L, 2430L, 2293L, 2293L,
2293L, 3977L, 3684L, 4523L, 4669L, 3264L, 3264L, 3264L, 3264L,
2854L, 2854L, 2289L, 2289L, 2289L, 2577L, 2577L, 4937L, 4937L,
2577L, 2582L, 2582L, 2578L, 2578L, 3416L, 3416L, 4527L, 4801L,
3685L, 3822L, 2302L, 2855L, 2855L, 4669L, 2287L, 2287L, 4669L,
3549L, 3549L, 4798L, 3549L, 4680L, 4680L, 4680L, 4822L, 4258L,
4948L, 3273L, 4677L, 4677L, 4677L, 4948L, 2854L, 2854L, 3264L,
3264L, 3264L, 4937L, 4937L, 2582L, 2582L, 2578L, 2578L, 2289L,
2289L, 2289L, 3416L, 2573L, 3416L, 4527L, 3685L, 3547L, 4801L,
3685L, 3547L, 2287L, 2287L, 2287L, 2287L, 2436L, 2291L, 2718L,
2718L, 4099L, 3131L, 4680L, 4680L, 4680L, 3260L, 3260L, 3977L,
2571L, 2578L, 2578L, 2854L, 2854L, 3264L, 3264L, 3264L, 4937L,
4937L, 2582L, 2582L, 2289L, 2289L, 2289L, 2573L, 2573L, 2573L,
2573L, 3132L, 3407L, 3416L, 3416L, 3685L, 3685L, 3685L, 4527L,
4801L, 2991L, 2287L, 2287L, 2426L, 3399L, 2301L, 4680L, 4680L,
4680L, 4541L, 4390L, 3277L, 3277L, 3277L, 3978L, 3978L, 3978L,
4937L, 4801L, 4801L, 4937L, 2289L, 2289L, 2289L, 2573L, 2854L,
3264L, 3264L, 3264L, 3684L, 3684L, 2582L, 2582L, 2854L, 3264L,
3684L, 4527L, 2578L, 2578L, 2718L, 2718L, 2296L, 4665L, 4665L,
4665L, 3416L, 3416L, 3277L, 3277L, 2443L, 2300L, 2302L, 4680L,
4680L, 4680L, 3546L, 3546L, 4937L, 4937L, 4801L, 4801L, 2854L,
2854L, 3264L, 3264L, 2289L, 2289L, 2582L, 2582L, 2578L, 2289L,
3416L, 3416L, 3416L, 3556L, 3277L, 3685L, 3978L, 4680L, 4110L,
4237L, 4527L, 4237L, 4937L, 4937L, 4801L, 4801L, 3264L, 3685L,
3416L, 3416L, 3416L, 2289L, 2289L, 2289L, 2582L, 2578L, 2582L,
2293L, 2857L, 2721L, 2443L, 4680L, 4680L)), class = "data.frame", row.names = c(NA,
-368L))

For a random-effects model -- like the one you are estimating -- to make sense, you usually need more than one observation per level to estimate the group level variance.

In your case, table(df$grid_no) shows:

2287 2289 2290 2291 2293 2296 2300 2301 2302 2303 2426 2427 2428 2430 2436 2443
   8   27    1    1    4    1    1    2    2    1    1    4    1    2    1    2
2571 2573 2577 2578 2581 2582 2584 2585 2700 2710 2718 2719 2721 2852 2854 2855
   1    6    3   13    3   18    2    1    2    1    4    2    1    1   22    2
2857 2861 2983 2986 2991 2996 3131 3132 3260 3264 3270 3273 3277 3399 3407 3409
   1    2    1    2    1    2    1    1    2   28    2    1    6    1    1    2
3416 3546 3547 3549 3556 3678 3683 3684 3685 3687 3692 3822 3834 3977 3978 3979
  24    2    8    5    1    1    2    4   14    1    3    2    1    2    7    1
4099 4110 4117 4237 4258 4388 4390 4520 4523 4526 4527 4538 4541 4663 4665 4669
   1    1    1    2    1    2    1    1    1    1    8    2    1    1    3    3
4670 4677 4680 4798 4801 4822 4839 4937 4948
   2    3   21    1   10    1    2   23    2

That is, about 12% of your data has only one observation per group.

If we throw out the one-observation groups, convergence issues go away:

library(tidyverse)
df %>% group_by(grid_no) %>% mutate(count_obs = n()) -> df
summary( glmer(an_larv_bin ~ EVI + (1|grid_no), family="binomial", data=df[df$count_obs > 1,]))
Generalized linear mixed model fit by maximum likelihood (Laplace
  Approximation) [glmerMod]
 Family: binomial  ( logit )
Formula: an_larv_bin ~ EVI + (1 | grid_no)
   Data: df[df$count_obs > 1, ]

     AIC      BIC   logLik deviance df.resid
   374.7    386.1   -184.4    368.7      327

Scaled residuals:
    Min      1Q  Median      3Q     Max
-1.0048 -0.5813 -0.4693  0.8706  2.4995

Random effects:
 Groups  Name        Variance Std.Dev.
 grid_no (Intercept) 0.6866   0.8286
Number of obs: 330, groups:  grid_no, 51

Fixed effects:
            Estimate Std. Error z value Pr(>|z|)
(Intercept)    1.896      1.322   1.434    0.151
EVI           -5.039      2.247  -2.242    0.025 *
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Correlation of Fixed Effects:
    (Intr)
EVI -0.989

Different gradient calculations in a logistic regression

1 Answers