Same data but different results when using `drm()` on different OSs

[annaquaglieri16]

Hi there,

I'd like to highlight some differences that I found when running the drm() function on my Mac (Processor 2.6 GHz 6-Core Intel Core i7) vs an Amazon Linux machine.

I attached the example dataset input_data.txt.

Example:

input_data = read.table(file.path(here(), "tests/test-data/input_data.txt"))
drm(response ~ dose, data = input_data, fct = L.4())

Results from my Mac

Model fitted: Logistic (ED50 as parameter) (4 parms)

Parameter estimates:

                 Estimate  Std. Error t-value   p-value    
b:(Intercept) -2.3306e-01  7.0496e-01 -0.3306 0.7434100    
c:(Intercept)  4.4357e+05  1.3346e+05  3.3236 0.0024854 ** 
d:(Intercept)  1.3001e+06  3.2732e+05  3.9720 0.0004529 ***
e:(Intercept)  5.2777e+01  1.2060e+01  4.3763 0.0001521 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error:

 654792.8 (28 degrees of freedom)

When running on an Amazon Linux machine:

Model fitted: Logistic (ED50 as parameter) (4 parms)
--
  |  
  | Parameter estimates:
  |  
  | Estimate  Std. Error t-value   p-value
  | b:(Intercept) -2.3552e-01  7.2686e-01 -0.3240 0.7483338
  | c:(Intercept)  4.4373e+05  1.3397e+05  3.3122 0.0025592 **
  | d:(Intercept)  1.2999e+06  3.2735e+05  3.9711 0.0004540 ***
  | e:(Intercept)  5.2747e+01  1.2068e+01  4.3709 0.0001543 ***
  | ---
  | Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
  |  
  | Residual standard error:
  |  
  | 654788.4 (28 degrees of freedom)

All the estimates are slightly but I believe not negligibly different.
Do you have any insights into how I can improve the reproducibility of the results? Thanks for all your work and your help,

Anna