I cannot predict more than 1000 points using the mgcv package at a time, otherwise it starts repeating the values

mgcv_predict1000

library(mgcv)
library(tidyverse)

# Data

BPR <- 
  tibble::tribble(
  ~time,              ~bpr,
  50, 0.892042350231095,
  170, 0.888132434012036,
  230,  0.87170777383751,
  290, 0.838972349740963,
  350, 0.930837665079573,
  410, 0.831346095761359,
  470, 0.947754901271871,
  530, 0.763998744037918,
  620, 0.834635757258697,
  1190,  0.70956560704307,
  1790, 0.551889549531392,
  2390, 0.697936911346988,
  3041, 0.658171861836563,
  3590, 0.656165891852214,
  4190,    0.687996980066,
  4790, 0.654859498649481,
  5390, 0.627913384922731
)


# Fit

bprspline <- mgcv::gam(bpr ~ splines::bs(time), 
                       data=BPR)


# Predict - for given values

BPR$splinepred <- predict(bprspline)

ggplot(BPR, aes(x=time, y=bpr)) + 
  geom_point() +
  geom_line(aes(y=splinepred))



# Predict - for 500 interpolated values

interptime <- seq(50, 5390, length.out=500)


splinepred_df <- tibble::tibble(time = interptime,
                                splinepred = predict(bprspline, 
                                                     newdata=data.frame(
                                                       time=interptime)))

ggplot(BPR, aes(x=time, y=bpr)) +
  geom_point() +
  geom_line(data=splinepred_df, aes(x=time, y=splinepred))



# Predict - for 1000 interpolated values

interptime <- seq(50, 5390, length.out=1000)


splinepred_df <- tibble::tibble(time = interptime,
                                splinepred = predict(bprspline, 
                                                     newdata=data.frame(
                                                       time=interptime)))

ggplot(BPR, aes(x=time, y=bpr)) +
  geom_point() +
  geom_line(data=splinepred_df, aes(x=time, y=splinepred))


# Predict - for 4000 interpolated values

interptime <- seq(50, 5390, length.out=4000)


splinepred_df <- tibble::tibble(time = interptime,
                                splinepred = predict(bprspline, 
                                                     newdata=data.frame(
                                                       time=interptime)))

ggplot(BPR, aes(x=time, y=bpr)) +
  geom_point() +
  geom_line(data=splinepred_df, aes(x=time, y=splinepred))