Add simpson's rule (#246)

Author: Erennn7

mathematics/simpson_rule.r

@@ -0,0 +1,231 @@
+# Simpson's Rule for Numerical Integration
+#
+# Simpson's Rule is a method for numerical integration that approximates 
+# the definite integral of a function using quadratic polynomials.
+# It provides better accuracy than the trapezoidal rule by using parabolic 
+# arcs instead of straight line segments.
+#
+# Time Complexity: O(n) where n is the number of subintervals
+# Space Complexity: O(n) for storing the x values
+#
+# Applications:
+# - Numerical integration in scientific computing
+# - Physics and engineering calculations
+# - Signal processing and data analysis
+# - Probability and statistics (computing areas under curves)
+# - Computer graphics and visualization
+#
+# Formula: ∫[a,b] f(x)dx ≈ (h/3)[f(x₀) + 4f(x₁) + 2f(x₂) + 4f(x₃) + ... + f(xₙ)]
+# where h = (b-a)/n and n must be even
+
+simpson_rule <- function(f, a, b, n) {
+  #' Approximate the definite integral of f(x) from a to b using Simpson's Rule
+  #' @param f: Function to integrate
+  #' @param a: Lower limit of integration (numeric)
+  #' @param b: Upper limit of integration (numeric)
+  #' @param n: Number of subintervals (must be even)
+  #' @return: Approximation of the integral (numeric)
+  #' @details Simpson's Rule uses parabolic approximations to estimate the 
+  #'          area under a curve. The interval [a,b] is divided into n 
+  #'          subintervals, and quadratic interpolation is applied.
+  #' @references https://en.wikipedia.org/wiki/Simpson%27s_rule
+  
+  # Validate that n is even
+  if (n %% 2 != 0) {
+    stop("Error: Number of subintervals n must be even for Simpson's Rule.")
+  }
+  
+  # Validate inputs
+  if (!is.numeric(a) || !is.numeric(b) || !is.numeric(n)) {
+    stop("Error: a, b, and n must be numeric values.")
+  }
+  
+  if (n <= 0) {
+    stop("Error: Number of subintervals n must be positive.")
+  }
+  
+  if (a >= b) {
+    stop("Error: Lower limit a must be less than upper limit b.")
+  }
+  
+  # Calculate step size
+  h <- (b - a) / n
+  
+  # Generate x values
+  x <- seq(a, b, by = h)
+  
+  # Initialize result with endpoints
+  result <- f(x[1]) + f(x[n + 1])
+  
+  # Apply Simpson's coefficients
+  # Odd indices (i = 3, 5, 7, ...) get coefficient 4
+  # Even indices (i = 2, 4, 6, ...) get coefficient 2
+  for (i in 2:n) {
+    if (i %% 2 != 0) {
+      # Odd index: multiply by 4
+      result <- result + 4 * f(x[i])
+    } else {
+      # Even index: multiply by 2
+      result <- result + 2 * f(x[i])
+    }
+  }
+  
+  # Multiply by h/3
+  result <- result * h / 3
+  
+  return(result)
+}
+
+# Vectorized version for better performance
+simpson_rule_vectorized <- function(f, a, b, n) {
+  #' Vectorized implementation of Simpson's Rule
+  #' @param f: Function to integrate (must support vectorized input)
+  #' @param a: Lower limit of integration
+  #' @param b: Upper limit of integration
+  #' @param n: Number of subintervals (must be even)
+  #' @return: Approximation of the integral
+  
+  if (n %% 2 != 0) {
+    stop("Error: Number of subintervals n must be even.")
+  }
+  
+  h <- (b - a) / n
+  x <- seq(a, b, by = h)
+  y <- f(x)
+  
+  # Create coefficient vector: [1, 4, 2, 4, 2, ..., 2, 4, 1]
+  coefficients <- rep(2, n + 1)
+  coefficients[1] <- 1
+  coefficients[n + 1] <- 1
+  coefficients[seq(2, n, by = 2)] <- 4
+  
+  result <- sum(coefficients * y) * h / 3
+  return(result)
+}
+
+# Helper function to print integration results
+print_integration_result <- function(f, a, b, n, result, exact = NULL) {
+  #' Print formatted integration results
+  #' @param f: Function that was integrated
+  #' @param a: Lower limit
+  #' @param b: Upper limit
+  #' @param n: Number of subintervals
+  #' @param result: Computed integral value
+  #' @param exact: Exact value if known (optional)
+  
+  cat("Simpson's Rule Integration:\n")
+  cat(sprintf("  Interval: [%.4f, %.4f]\n", a, b))
+  cat(sprintf("  Subintervals: %d\n", n))
+  cat(sprintf("  Approximate integral: %.10f\n", result))
+  
+  if (!is.null(exact)) {
+    error <- abs(result - exact)
+    rel_error <- error / abs(exact) * 100
+    cat(sprintf("  Exact value: %.10f\n", exact))
+    cat(sprintf("  Absolute error: %.2e\n", error))
+    cat(sprintf("  Relative error: %.4f%%\n", rel_error))
+  }
+  cat("\n")
+}
+
+# ========== Example Usage ==========
+
+cat("========== Example 1: Integral of sin(x) from 0 to π ==========\n\n")
+
+# Define the function to integrate
+func1 <- function(x) sin(x)
+
+# Integration limits
+a1 <- 0
+b1 <- pi
+
+# Number of subintervals (must be even)
+n1 <- 10
+
+# Calculate integral
+integral1 <- simpson_rule(func1, a1, b1, n1)
+exact1 <- 2.0  # Exact value of ∫sin(x)dx from 0 to π is 2
+
+print_integration_result(func1, a1, b1, n1, integral1, exact1)
+
+# ========== Example 2: Integral of x² from 0 to 1 ==========
+
+cat("========== Example 2: Integral of x² from 0 to 1 ==========\n\n")
+
+func2 <- function(x) x^2
+a2 <- 0
+b2 <- 1
+n2 <- 20
+
+integral2 <- simpson_rule(func2, a2, b2, n2)
+exact2 <- 1/3  # Exact value is 1/3
+
+print_integration_result(func2, a2, b2, n2, integral2, exact2)
+
+# ========== Example 3: Integral of e^x from 0 to 1 ==========
+
+cat("========== Example 3: Integral of e^x from 0 to 1 ==========\n\n")
+
+func3 <- function(x) exp(x)
+a3 <- 0
+b3 <- 1
+n3 <- 10
+
+integral3 <- simpson_rule(func3, a3, b3, n3)
+exact3 <- exp(1) - 1  # Exact value is e - 1
+
+print_integration_result(func3, a3, b3, n3, integral3, exact3)
+
+# ========== Example 4: Comparison with different n values ==========
+
+cat("========== Example 4: Convergence study for sin(x) ==========\n\n")
+
+n_values <- c(10, 20, 50, 100, 200)
+cat("Convergence of Simpson's Rule with increasing subintervals:\n")
+cat(sprintf("%-15s %-20s %-15s\n", "Subintervals", "Approximate", "Error"))
+cat(strrep("-", 50), "\n")
+
+for (n in n_values) {
+  result <- simpson_rule(sin, 0, pi, n)
+  error <- abs(result - 2.0)
+  cat(sprintf("%-15d %-20.12f %.4e\n", n, result, error))
+}
+cat("\n")
+
+# ========== Example 5: Using vectorized version ==========
+
+cat("========== Example 5: Vectorized implementation ==========\n\n")
+
+func5 <- function(x) 1 / (1 + x^2)  # Arctangent derivative
+a5 <- 0
+b5 <- 1
+n5 <- 100
+
+integral5 <- simpson_rule_vectorized(func5, a5, b5, n5)
+exact5 <- pi / 4  # ∫[0,1] 1/(1+x²)dx = arctan(1) - arctan(0) = π/4
+
+print_integration_result(func5, a5, b5, n5, integral5, exact5)
+
+# ========== Example 6: Gaussian function ==========
+
+cat("========== Example 6: Gaussian (Normal) distribution PDF ==========\n\n")
+
+# Standard normal distribution PDF from -3 to 3
+gaussian <- function(x) (1 / sqrt(2 * pi)) * exp(-x^2 / 2)
+a6 <- -3
+b6 <- 3
+n6 <- 100
+
+integral6 <- simpson_rule(gaussian, a6, b6, n6)
+# Exact value is approximately 0.9973 (99.73% within 3 standard deviations)
+
+cat(sprintf("Integral of standard normal PDF from %.1f to %.1f: %.6f\n", a6, b6, integral6))
+cat(sprintf("This represents %.2f%% of the total area under the curve.\n\n", integral6 * 100))
+
+# ========== Notes ==========
+cat("========== Important Notes ==========\n")
+cat("1. Simpson's Rule requires an EVEN number of subintervals.\n")
+cat("2. It provides O(h⁴) accuracy (fourth-order accurate).\n")
+cat("3. More accurate than Trapezoidal Rule for smooth functions.\n")
+cat("4. Works best for continuous, smooth functions.\n")
+cat("5. For higher accuracy, increase the number of subintervals.\n")