25 Rewriting R code in C++

library(Rcpp, warn.conflicts = FALSE)

25.1 Getting started with C++ (Exercises 25.2.6)

Q1. With the basics of C++ in hand, it’s now a great time to practice by reading and writing some simple C++ functions. For each of the following functions, read the code and figure out what the corresponding base R function is. You might not understand every part of the code yet, but you should be able to figure out the basics of what the function does.

#include <Rcpp.h>
using namespace Rcpp;

// [[Rcpp::export]]
double f1(NumericVector x) {
  int n = x.size();
  double y = 0;

  for(int i = 0; i < n; ++i) {
    y += x[i] / n;
  }
  return y;
}

// [[Rcpp::export]]
NumericVector f2(NumericVector x) {
  int n = x.size();
  NumericVector out(n);

  out[0] = x[0];
  for(int i = 1; i < n; ++i) {
    out[i] = out[i - 1] + x[i];
  }
  return out;
}

// [[Rcpp::export]]
bool f3(LogicalVector x) {
  int n = x.size();

  for(int i = 0; i < n; ++i) {
    if (x[i]) return true;
  }
  return false;
}

// [[Rcpp::export]]
int f4(Function pred, List x) {
  int n = x.size();

  for(int i = 0; i < n; ++i) {
    LogicalVector res = pred(x[i]);
    if (res[0]) return i + 1;
  }
  return 0;
}

// [[Rcpp::export]]
NumericVector f5(NumericVector x, NumericVector y) {
  int n = std::max(x.size(), y.size());
  NumericVector x1 = rep_len(x, n);
  NumericVector y1 = rep_len(y, n);

  NumericVector out(n);

  for (int i = 0; i < n; ++i) {
    out[i] = std::min(x1[i], y1[i]);
  }

  return out;
}

A1.

f1() is the same as mean():

x <- c(1, 2, 3, 4, 5, 6)

f1(x)
#> [1] 3.5
mean(x)
#> [1] 3.5

f2() is the same as cumsum():

x <- c(1, 3, 5, 6)

f2(x)
#> [1]  1  4  9 15
cumsum(x)
#> [1]  1  4  9 15

f3() is the same as any():

x1 <- c(TRUE, FALSE, FALSE, TRUE)
x2 <- c(FALSE, FALSE)

f3(x1)
#> [1] TRUE
any(x1)
#> [1] TRUE

f3(x2)
#> [1] FALSE
any(x2)
#> [1] FALSE

f4() is the same as Position():

x <- list("a", TRUE, "m", 2)

f4(is.numeric, x)
#> [1] 4
Position(is.numeric, x)
#> [1] 4

f5() is the same as pmin():

v1 <- c(1, 3, 4, 5, 6, 7)
v2 <- c(1, 2, 7, 2, 8, 1)

f5(v1, v2)
#> [1] 1 2 4 2 6 1
pmin(v1, v2)
#> [1] 1 2 4 2 6 1

Q2. To practice your function writing skills, convert the following functions into C++. For now, assume the inputs have no missing values.

all().
cumprod(), cummin(), cummax().
diff(). Start by assuming lag 1, and then generalise for lag n.
range().
var(). Read about the approaches you can take on Wikipedia. Whenever implementing a numerical algorithm, it’s always good to check what is already known about the problem.

A2. The performance benefits are not going to be observed if the function is primitive since those are already tuned to the max in R for performance. So, expect performance gain only for diff() and var().

is.primitive(all)
#> [1] TRUE
is.primitive(cumprod)
#> [1] TRUE
is.primitive(diff)
#> [1] FALSE
is.primitive(range)
#> [1] TRUE
is.primitive(var)
#> [1] FALSE

all()

#include <vector>
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
bool allC(std::vector<bool> x)
{
    for (const auto& xElement : x)
    {
        if (!xElement) return false;
    }

    return true;
}

v1 <- rep(TRUE, 10)
v2 <- c(rep(TRUE, 5), rep(FALSE, 5))

all(v1)
#> [1] TRUE
allC(v1)
#> [1] TRUE

all(v2)
#> [1] FALSE
allC(v2)
#> [1] FALSE

# performance benefits?
bench::mark(
  all(c(rep(TRUE, 1000), rep(FALSE, 1000))),
  allC(c(rep(TRUE, 1000), rep(FALSE, 1000))),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression                                      min
#>   <bch:expr>                                 <bch:tm>
#> 1 all(c(rep(TRUE, 1000), rep(FALSE, 1000)))    6.09µs
#> 2 allC(c(rep(TRUE, 1000), rep(FALSE, 1000)))   7.74µs
#>     median `itr/sec` mem_alloc `gc/sec`
#>   <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1   6.42µs   151475.    15.8KB        0
#> 2   8.07µs   122722.    15.8KB        0

cumprod()

#include <vector>

// [[Rcpp::export]]
std::vector<double> cumprodC(const std::vector<double> &x)
{
    std::vector<double> out{x};

    for (std::size_t i = 1; i < x.size(); i++)
    {
        out[i] = out[i - 1] * x[i];
    }

    return out;
}

v1 <- c(10, 4, 6, 8)

cumprod(v1)
#> [1]   10   40  240 1920
cumprodC(v1)
#> [1]   10   40  240 1920

# performance benefits?
bench::mark(
  cumprod(v1),
  cumprodC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression        min   median `itr/sec` mem_alloc
#>   <bch:expr>   <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 cumprod(v1)    90.1ns    101ns  8302502.        0B
#> 2 cumprodC(v1)  692.1ns    721ns  1263257.    4.12KB
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

cumminC()

#include <vector>
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::vector<double> cumminC(const std::vector<double> &x)
{
    std::vector<double> out{x};

    for (std::size_t i = 1; i < x.size(); i++)
    {
        out[i] = (out[i] < out[i - 1]) ? out[i] : out[i - 1];
    }

    return out;
}

v1 <- c(3:1, 2:0, 4:2)

cummin(v1)
#> [1] 3 2 1 1 1 0 0 0 0
cumminC(v1)
#> [1] 3 2 1 1 1 0 0 0 0

# performance benefits?
bench::mark(
  cummin(v1),
  cumminC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression       min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr>  <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 cummin(v1)     110ns    165ns  5353852.        0B        0
#> 2 cumminC(v1)    771ns    947ns   991420.    4.12KB        0

cummaxC()

#include <vector>
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::vector<double> cummaxC(const std::vector<double> &x)
{
    std::vector<double> out{x};

    for (std::size_t i = 1; i < x.size(); i++)
    {
        out[i] = (out[i] > out[i - 1]) ? out[i] : out[i - 1];
    }
    
    return out;
}

v1 <- c(3:1, 2:0, 4:2)

cummax(v1)
#> [1] 3 3 3 3 3 3 4 4 4
cummaxC(v1)
#> [1] 3 3 3 3 3 3 4 4 4

# performance benefits?
bench::mark(
  cummax(v1),
  cummaxC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression       min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr>  <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 cummax(v1)     100ns    120ns  6895724.        0B        0
#> 2 cummaxC(v1)    741ns    792ns  1107069.    4.12KB        0

diff()

#include <vector>
#include <functional>
#include <algorithm>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::vector<double> diffC(const std::vector<double> &x, int lag)
{
    std::vector<double> vec_start;
    std::vector<double> vec_lagged;
    std::vector<double> vec_diff;

    for (std::size_t i = lag; i < x.size(); i++)
    {
        vec_lagged.push_back(x[i]);
    }

    for (std::size_t i = 0; i < (x.size() - lag); i++)
    {
        vec_start.push_back(x[i]);
    }

    std::transform(
        vec_lagged.begin(), vec_lagged.end(),
        vec_start.begin(), std::back_inserter(vec_diff),
        std::minus<double>());

    return vec_diff;
}

v1 <- c(1, 2, 4, 8, 13)
v2 <- c(1, 2, NA, 8, 13)

diff(v1, 2)
#> [1] 3 6 9
diffC(v1, 2)
#> [1] 3 6 9

diff(v2, 2)
#> [1] NA  6 NA
diffC(v2, 2)
#> [1] NA  6 NA

# performance benefits?
bench::mark(
  diff(v1, 2),
  diffC(v1, 2),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression        min   median `itr/sec` mem_alloc
#>   <bch:expr>   <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 diff(v1, 2)    3.95µs   4.39µs   212450.        0B
#> 2 diffC(v1, 2)   1.05µs   1.12µs   824596.        0B
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

range()

#include <iostream>
#include <vector>
#include <algorithm>
using namespace std;

// [[Rcpp::export]]
std::vector<double> rangeC(std::vector<double> x)
{
    std::vector<double> rangeVec{0.0, 0.0};

    rangeVec.at(0) = *std::min_element(x.begin(), x.end());
    rangeVec.at(1) = *std::max_element(x.begin(), x.end());

    return rangeVec;
}

v1 <- c(10, 4, 6, 8)

range(v1)
#> [1]  4 10
rangeC(v1)
#> [1]  4 10

# performance benefits?
bench::mark(
  range(v1),
  rangeC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 range(v1)    2.52µs   2.73µs   352181.        0B        0
#> 2 rangeC(v1) 701.98ns 782.08ns  1200231.    4.12KB        0

var()

#include <vector>
#include <cmath>
#include <numeric>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
double variance(std::vector<double> x)
{
    double sumSquared{0};

    double mean = std::accumulate(x.begin(), x.end(), 0.0) / x.size();

    for (const auto& xElement : x)
    {
        sumSquared += pow(xElement - mean, 2.0);
    }

    return sumSquared / (x.size() - 1);
}

v1 <- c(1, 4, 7, 8)

var(v1)
#> [1] 10
variance(v1)
#> [1] 10

# performance benefits?
bench::mark(
  var(v1),
  variance(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression        min   median `itr/sec` mem_alloc
#>   <bch:expr>   <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 var(v1)        5.36µs   5.86µs   159660.        0B
#> 2 variance(v1) 672.07ns 762.05ns  1146229.    4.12KB
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

25.2 Missing values (Exercises 25.4.5)

Q1. Rewrite any of the functions from Exercise 25.2.6 to deal with missing values. If na.rm is true, ignore the missing values. If na.rm is false, return a missing value if the input contains any missing values. Some good functions to practice with are min(), max(), range(), mean(), and var().

A1. We will only create a version of range() that deals with missing values. The same principle applies to others:

#include <iostream>
#include <vector>
#include <algorithm>
#include <math.h>
#include <Rcpp.h>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::vector<double> rangeC_NA(std::vector<double> x, bool removeNA = true)
{
    std::vector<double> rangeVec{0.0, 0.0};

    bool naPresent = std::any_of(
        x.begin(),
        x.end(),
        [](double d)
        { return isnan(d); });

    if (naPresent)
    {
        if (removeNA)
        {
            std::remove(x.begin(), x.end(), NAN);
        }
        else
        {
            rangeVec.at(0) = NA_REAL; // NAN;
            rangeVec.at(1) = NA_REAL; // NAN;

            return rangeVec;
        }
    }

    rangeVec.at(0) = *std::min_element(x.begin(), x.end());
    rangeVec.at(1) = *std::max_element(x.begin(), x.end());

    return rangeVec;
}

v1 <- c(10, 4, NA, 6, 8)

range(v1, na.rm = FALSE)
#> [1] NA NA
rangeC_NA(v1, FALSE)
#> [1] NA NA

range(v1, na.rm = TRUE)
#> [1]  4 10
rangeC_NA(v1, TRUE)
#> [1]  4 10

Q2. Rewrite cumsum() and diff() so they can handle missing values. Note that these functions have slightly more complicated behaviour.

A2. The cumsum() docs say:

An NA value in x causes the corresponding and following elements of the return value to be NA, as does integer overflow in cumsum (with a warning).

Similarly, diff() docs say:

NA’s propagate.

Therefore, both of these functions don’t allow removing missing values and the NAs propagate.

As seen from the examples above, diffC() already behaves this way.

Similarly, cumsumC() propagates NAs as well.

#include <Rcpp.h>
using namespace Rcpp;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
NumericVector cumsumC(NumericVector x) {
  int n = x.size();
  NumericVector out(n);
  
  out[0] = x[0];
  for(int i = 1; i < n; ++i) {
    out[i] = out[i - 1] + x[i];
  }
  
  return out;
}

v1 <- c(1, 2, 3, 4)
v2 <- c(1, 2, NA, 4)

cumsum(v1)
#> [1]  1  3  6 10
cumsumC(v1)
#> [1]  1  3  6 10

cumsum(v2)
#> [1]  1  3 NA NA
cumsumC(v2)
#> [1]  1  3 NA NA

25.3 Standard Template Library (Exercises 25.5.7)

Q1. To practice using the STL algorithms and data structures, implement the following using R functions in C++, using the hints provided:

A1.

median.default() using partial_sort.

#include <iostream>
#include <vector>
#include <algorithm>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
double medianC(std::vector<double> &x)
{
    int middleIndex = static_cast<int>(x.size() / 2);

    std::partial_sort(x.begin(), x.begin() + middleIndex, x.end());

    // for even number of observations
    if (x.size() % 2 == 0)
    {
        return (x[middleIndex - 1] + x[middleIndex]) / 2;
    }

    return x[middleIndex];
}

v1 <- c(1, 3, 3, 6, 7, 8, 9)
v2 <- c(1, 2, 3, 4, 5, 6, 8, 9)

median.default(v1)
#> [1] 6
medianC(v1)
#> [1] 6

median.default(v2)
#> [1] 4.5
medianC(v2)
#> [1] 4.5

# performance benefits?
bench::mark(
  median.default(v2),
  medianC(v2),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression              min   median `itr/sec` mem_alloc
#>   <bch:expr>         <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 median.default(v2)   19.6µs   21.7µs    44295.        0B
#> 2 medianC(v2)         692.1ns    731ns  1251690.        0B
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

%in% using unordered_set and the find() or count() methods.

#include <vector>
#include <unordered_set>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::vector<bool> matchC(const std::vector<double> &x, const std::vector<double> &table)
{
    std::unordered_set<double> tableUnique(table.begin(), table.end());
    std::vector<bool> out;

    for (const auto &xElem : x)
    {
        out.push_back(tableUnique.find(xElem) != tableUnique.end() ? true : false);
    }

    return out;
}

x1 <- c(3, 4, 8)
x2 <- c(1, 2, 3, 3, 4, 4, 5, 6)

x1 %in% x2
#> [1]  TRUE  TRUE FALSE
matchC(x1, x2)
#> [1]  TRUE  TRUE FALSE

# performance benefits?
bench::mark(
  x1 %in% x2,
  matchC(x1, x2),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression          min   median `itr/sec` mem_alloc
#>   <bch:expr>     <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 x1 %in% x2     862.05ns   1.07µs   889456.        0B
#> 2 matchC(x1, x2)   1.22µs   1.27µs   713125.    4.12KB
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

unique() using an unordered_set (challenge: do it in one line!).

#include <unordered_set>
#include <vector>
#include <iostream>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::unordered_set<double> uniqueC(const std::vector<double> &x)
{
    std::unordered_set<double> xSet(x.begin(), x.end());

    return xSet;
}

Note that these functions are not comparable. As far as I can see, there is no way to get the same output as the R version of the function using the unordered_set data structure.

v1 <- c(1, 3, 3, 6, 7, 8, 9)

unique(v1)
#> [1] 1 3 6 7 8 9
uniqueC(v1)
#> [1] 9 8 7 6 3 1

We can make comparable version using set data structure:

#include <set>
#include <vector>
#include <iostream>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::set<double> uniqueC2(const std::vector<double> &x)
{
    std::set<double> xSet(x.begin(), x.end());

    return xSet;
}

v1 <- c(1, 3, 3, 6, 7, 8, 9)

unique(v1)
#> [1] 1 3 6 7 8 9
uniqueC2(v1)
#> [1] 1 3 6 7 8 9

# performance benefits?
bench::mark(
  unique(v1),
  uniqueC2(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression        min   median `itr/sec` mem_alloc
#>   <bch:expr>   <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 unique(v1)     2.11µs    2.5µs   350572.        0B
#> 2 uniqueC2(v1) 832.02ns  952.6ns   972023.    4.12KB
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

min() using std::min(), or max() using std::max().

#include <iostream>
#include <vector>
#include <algorithm>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
const double minC(const std::vector<double> &x)
{
     return *std::min_element(x.begin(), x.end());
}

// [[Rcpp::export]]
const double maxC(std::vector<double> x)
{
     return *std::max_element(x.begin(), x.end());
}

v1 <- c(3, 3, 6, 1, 9, 7, 8)

min(v1)
#> [1] 1
minC(v1)
#> [1] 1

# performance benefits?
bench::mark(
  min(v1),
  minC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 min(v1)       210ns    231ns  3652276.        0B        0
#> 2 minC(v1)      671ns    732ns  1221424.    4.12KB        0

max(v1)
#> [1] 9
maxC(v1)
#> [1] 9

# performance benefits?
bench::mark(
  max(v1),
  maxC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression      min   median `itr/sec` mem_alloc `gc/sec`
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl>
#> 1 max(v1)       210ns    221ns  3988269.        0B        0
#> 2 maxC(v1)      681ns    712ns  1270236.    4.12KB        0

which.min() using min_element, or which.max() using max_element.

#include <vector>
#include <algorithm>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
int which_maxC(std::vector<double> &x)
{
    int maxIndex = std::distance(x.begin(), std::max_element(x.begin(), x.end()));
  
    // R is 1-index based, while C++ is 0-index based
    return maxIndex + 1;
}

// [[Rcpp::export]]
int which_minC(std::vector<double> &x)
{
    int minIndex = std::distance(x.begin(), std::min_element(x.begin(), x.end()));
  
    // R is 1-index based, while C++ is 0-index based
    return minIndex + 1;
}

v1 <- c(3, 3, 6, 1, 9, 7, 8)

which.min(v1)
#> [1] 4
which_minC(v1)
#> [1] 4

# performance benefits?
bench::mark(
  which.min(v1),
  which_minC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression          min   median `itr/sec` mem_alloc
#>   <bch:expr>     <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 which.min(v1)     381ns    401ns  2336850.        0B
#> 2 which_minC(v1)    651ns    691ns  1290918.    4.12KB
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

which.max(v1)
#> [1] 5
which_maxC(v1)
#> [1] 5

# performance benefits?
bench::mark(
  which.max(v1),
  which_maxC(v1),
  iterations = 100
)
#> # A tibble: 2 × 6
#>   expression          min   median `itr/sec` mem_alloc
#>   <bch:expr>     <bch:tm> <bch:tm>     <dbl> <bch:byt>
#> 1 which.max(v1)     391ns    421ns  2072346.        0B
#> 2 which_maxC(v1)    662ns    697ns  1284518.    4.12KB
#>   `gc/sec`
#>      <dbl>
#> 1        0
#> 2        0

setdiff(), union(), and intersect() for integers using sorted ranges and set_union, set_intersection and set_difference.

Note that the following C++ implementations of given functions are not strictly equivalent to their R versions. As far as I can see, there is no way for them to be identical while satisfying the specifications mentioned in the question.

union()

#include <algorithm>
#include <iostream>
#include <vector>
#include <set>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::set<int> unionC(std::vector<int> &v1, std::vector<int> &v2)
{
    std::sort(v1.begin(), v1.end());
    std::sort(v2.begin(), v2.end());

    std::vector<int> union_vec(v1.size() + v2.size());
    auto it = std::set_union(v1.begin(), v1.end(), v2.begin(), v2.end(), union_vec.begin());

    union_vec.resize(it - union_vec.begin());
    std::set<int> union_set(union_vec.begin(), union_vec.end());

    return union_set;
}

v1 <- c(1, 4, 5, 5, 5, 6, 2)
v2 <- c(4, 1, 6, 8)

union(v1, v2)
#> [1] 1 4 5 6 2 8
unionC(v1, v2)
#> [1] 1 2 4 5 6 8

intersect()

#include <algorithm>
#include <iostream>
#include <vector>
#include <set>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::set<int> intersectC(std::vector<int> &v1, std::vector<int> &v2)
{
    std::sort(v1.begin(), v1.end());
    std::sort(v2.begin(), v2.end());

    std::vector<int> union_vec(v1.size() + v2.size());
    auto it = std::set_intersection(v1.begin(), v1.end(), v2.begin(), v2.end(), union_vec.begin());

    union_vec.resize(it - union_vec.begin());
    std::set<int> union_set(union_vec.begin(), union_vec.end());

    return union_set;
}

v1 <- c(1, 4, 5, 5, 5, 6, 2)
v2 <- c(4, 1, 6, 8)

intersect(v1, v2)
#> [1] 1 4 6
intersectC(v1, v2)
#> [1] 1 4 6

setdiff()

#include <algorithm>
#include <iostream>
#include <vector>
#include <set>
using namespace std;
// [[Rcpp::plugins(cpp11)]]

// [[Rcpp::export]]
std::set<int> setdiffC(std::vector<int> &v1, std::vector<int> &v2)
{
    std::sort(v1.begin(), v1.end());
    std::sort(v2.begin(), v2.end());

    std::vector<int> union_vec(v1.size() + v2.size());
    auto it = std::set_difference(v1.begin(), v1.end(), v2.begin(), v2.end(), union_vec.begin());

    union_vec.resize(it - union_vec.begin());
    std::set<int> union_set(union_vec.begin(), union_vec.end());

    return union_set;
}

v1 <- c(1, 4, 5, 5, 5, 6, 2)
v2 <- c(4, 1, 6, 8)

setdiff(v1, v2)
#> [1] 5 2
setdiffC(v1, v2)
#> [1] 2 5

25.4 Session information

sessioninfo::session_info(include_base = TRUE)
#> ─ Session info ───────────────────────────────────────────
#>  setting  value
#>  version  R version 4.5.2 (2025-10-31)
#>  os       Ubuntu 24.04.3 LTS
#>  system   x86_64, linux-gnu
#>  ui       X11
#>  language (EN)
#>  collate  C.UTF-8
#>  ctype    C.UTF-8
#>  tz       UTC
#>  date     2026-01-13
#>  pandoc   3.8.3 @ /opt/hostedtoolcache/pandoc/3.8.3/x64/ (via rmarkdown)
#>  quarto   NA
#> 
#> ─ Packages ───────────────────────────────────────────────
#>  package     * version date (UTC) lib source
#>  base        * 4.5.2   2025-10-31 [3] local
#>  bench         1.1.4   2025-01-16 [1] RSPM
#>  bookdown      0.46    2025-12-05 [1] RSPM
#>  bslib         0.9.0   2025-01-30 [1] RSPM
#>  cachem        1.1.0   2024-05-16 [1] RSPM
#>  cli           3.6.5   2025-04-23 [1] RSPM
#>  compiler      4.5.2   2025-10-31 [3] local
#>  datasets    * 4.5.2   2025-10-31 [3] local
#>  digest        0.6.39  2025-11-19 [1] RSPM
#>  downlit       0.4.5   2025-11-14 [1] RSPM
#>  emoji         16.0.0  2024-10-28 [1] RSPM
#>  evaluate      1.0.5   2025-08-27 [1] RSPM
#>  fastmap       1.2.0   2024-05-15 [1] RSPM
#>  fs            1.6.6   2025-04-12 [1] RSPM
#>  glue          1.8.0   2024-09-30 [1] RSPM
#>  graphics    * 4.5.2   2025-10-31 [3] local
#>  grDevices   * 4.5.2   2025-10-31 [3] local
#>  htmltools     0.5.9   2025-12-04 [1] RSPM
#>  jquerylib     0.1.4   2021-04-26 [1] RSPM
#>  jsonlite      2.0.0   2025-03-27 [1] RSPM
#>  knitr         1.51    2025-12-20 [1] RSPM
#>  lifecycle     1.0.5   2026-01-08 [1] RSPM
#>  magrittr    * 2.0.4   2025-09-12 [1] RSPM
#>  memoise       2.0.1   2021-11-26 [1] RSPM
#>  methods     * 4.5.2   2025-10-31 [3] local
#>  pillar        1.11.1  2025-09-17 [1] RSPM
#>  pkgconfig     2.0.3   2019-09-22 [1] RSPM
#>  profmem       0.7.0   2025-05-02 [1] RSPM
#>  R6            2.6.1   2025-02-15 [1] RSPM
#>  Rcpp        * 1.1.1   2026-01-10 [1] RSPM
#>  rlang         1.1.7   2026-01-09 [1] RSPM
#>  rmarkdown     2.30    2025-09-28 [1] RSPM
#>  sass          0.4.10  2025-04-11 [1] RSPM
#>  sessioninfo   1.2.3   2025-02-05 [1] RSPM
#>  stats       * 4.5.2   2025-10-31 [3] local
#>  stringi       1.8.7   2025-03-27 [1] RSPM
#>  stringr       1.6.0   2025-11-04 [1] RSPM
#>  tibble        3.3.1   2026-01-11 [1] RSPM
#>  tools         4.5.2   2025-10-31 [3] local
#>  utf8          1.2.6   2025-06-08 [1] RSPM
#>  utils       * 4.5.2   2025-10-31 [3] local
#>  vctrs         0.6.5   2023-12-01 [1] RSPM
#>  withr         3.0.2   2024-10-28 [1] RSPM
#>  xfun          0.55    2025-12-16 [1] RSPM
#>  xml2          1.5.1   2025-12-01 [1] RSPM
#>  yaml          2.3.12  2025-12-10 [1] RSPM
#> 
#>  [1] /home/runner/work/_temp/Library
#>  [2] /opt/R/4.5.2/lib/R/site-library
#>  [3] /opt/R/4.5.2/lib/R/library
#>  * ── Packages attached to the search path.
#> 
#> ──────────────────────────────────────────────────────────

24 Improving performance