R vs Python differences in floating point resolution : Is it possible to set Python act like R?

Question

I am working on a personal project involving the manipulation of linear algebra concepts. This project involves, in the first instance, replicating (translating) an R (64bit version) code into a Python code. The problem described here attempts to reproduce the problem I am facing.

Based on some internet search I have done, this problem seems to be related to the handling of decimal numbers with great precision.

Here are the two pieces of code:

R code:

library(base)

# load, subsetting and convert df to matrix
df <- read.csv('hansen99.csv') 
arr <- data.matrix(df[c(4:20)])

# crossprod leads to a square matrix and do inverse calculation
A <- t(arr) %*% arr # crossprod leads to a square matrix
A_inv <- solve(A)   

# affichage avec 100 décimales
sprintf("%.100f",A_inv[1,1]) 
# '0.0023559636604639148764472889041599046322517096996307373046875000000000000000000000000000000000000000'

Python code:

import numpy as np
import numpy.linalg as npla
import pandas as pd

# load, subsetting and convert df to array
df = pd.read_csv('hansen99.csv')
arr = df.iloc[:, 3:21].values

# crossprod leads to a square matrix and do inverse calculation
A = np.transpose(arr) @ arr
A_inv = npla.inv(A)

# affichage avec 100 décimales
"{:.100f}".format(A_inv[0,0])
# '0.0023559636604639157438090268925634518382139503955841064453125000000000000000000000000000000000000000'

A_inv is therefore the inverse of a given matrix A (csv file). If I take the first element of this matrix and print it with 100 decimal places, I obtain :

R      : '0.0023559636604639*1*48764472889041599046322517096996307373046875000000000000000000000000000000000000000'

Python : '0.0023559636604639*1*57438090268925634518382139503955841064453125000000000000000000000000000000000000000'

We can see that after the 17th decimal place (between stars) the sequence of numbers displayed is different between R and Python.

I've done some research and it seems to be related to the decimal resolution of types.

In python, I have the command for the float64 type (Don't know how to get it in R):

np.finfo(np.float64)
# finfo(resolution=1e-15, min=-1.7976931348623157e+308, max=1.7976931348623157e+308, dtype=float64)

Notes:

This difference does not seem to be related to a difference between the respective functions of the two languages. I called the R functions in Python and the Python functions in R to make sure.

Data:

If you are interested in data follow https://github.com/ceteuf/PSTR/tree/main/data/hansen99.csv

Questions:

1) What do you think is the reason for this difference? (need clarification, and possibly good resources to learn more)

2) Is it possible, by modifying a configuration or whatever, to reproduce exactly the same result and behaviour as R?

2b) Or was I completely wrong? In that case, do you have any leads?