Search This Blog

Sunday, May 29, 2011

Numerical Integration: Better Refinements?

;; Numerical Integration: Better Refinements?

;; Here are some very simple functions which we might want to test integration
;; methods on:
(defn square  [x] (* x x))
(defn sine    [x] (Math/sin x))
(defn step    [x] (if (< x 1/2) 0.0 1.0))
(defn inverse [x] (/ x))

;; Here are some Newton-Cotes formulae for approximate integration:

(defn trapezium-rule [f a b]
  (* 1/2 (- b a) (+ (f a) (f b))))

(defn simpson-rule [f a b]
  (let [midpoint (+ a (/ (- b a) 2))]
    (* 1/6 (- b a) (+ (f a) (* 4 (f midpoint)) (f b)))))

(defn simpson38-rule [f a b]
  (let [midpoint1 (/ (+ a a b) 3)
        midpoint2 (/ (+ a b b) 3)]
    (* 1/8 (- b a) (+ (f a) (* 3 (f midpoint1)) (* 3 (f midpoint2)) (f b)))))

(defn booles-rule [f a b]
  (let [midpoint1 (/ (+ a a a b) 4)
        midpoint2 (/ (+ a a b b) 4)
        midpoint3 (/ (+ a b b b) 4)]
    (* 1/90 (- b a) (+ (* 7 (f a)) (* 32 (f midpoint1)) (* 12 (f midpoint2)) (* 32 (f midpoint3)) (* 7 (f b))))))

;; We can use any of these rules to get estimate of the integral of a function over an interval:
(simpson-rule inverse 1 3) ; 10/9

;; If we halve the interval and use the rule over both halves, then we can use the rule to
;; get a better estimate by adding the estimates for the half-intervals
(+
 (simpson-rule inverse 1 2)
 (simpson-rule inverse 2 3)) ; 11/10

;; We can guess at the error involved in the estimate by taking the difference
;; between these two estimates, on the basis that splitting the interval usually
;; makes most of the error go away.

(- (simpson-rule inverse 1 3)
   (+
    (simpson-rule inverse 1 2)
    (simpson-rule inverse 2 3))) ; 1/90

;; So we'd expect that the first estimate is out by a bit more than 1/90, and
;; that the second is out by rather less than 1/90

;; For the inverse function, which can be integrated symbolically, we know the
;; true answer:
(- (Math/log 3) (Math/log 1)) ; 1.0986122886681098
(/ 10.0 9) ; 1.1111111111111112
(/ 11.0 10) ; 1.1

;; So the errors are really:
(- 1.0986122 10/9)  ; -0.0124989111111109  ; which is ~ 1/90
(- 1.0986122 11/10) ; -0.00138780000000005 ; which is ~ 1/900

;; This method of guessing the error is deeply suspect, and can go wrong, but I
;; won't go into details.

;; I think it's good enough for our purposes as long as the functions we want to
;; integrate are reasonably well behaved and we take small enough intervals.

;; So we can easily make a function which gives us the more refined of the two
;; estimates, together with a guess as to how close it is to the truth.
(defn approx-with-error[rule f a b]
  (let [guess (rule f a b)
        midpoint (/ (+ a b) 2)
        better-guess (+ (rule f a midpoint) (rule f midpoint b))
        error-estimate (- guess better-guess)
        abs-error-estimate (if (> error-estimate 0) error-estimate (- error-estimate))]
    [better-guess abs-error-estimate]))


;; Let's try it out on a few cases, on the particularly nasty integral of 1/x over [0.01,100]

;; This is the true answer
(- (Math/log 100) (Math/log 0.01)) ; 9.210340371976184

(approx-with-error trapezium-rule inverse 0.01 100) ; [2500.999775019998 2499.000174980002]
;; We guess 2500, and we think we're out by at most 2499, which is just true
(approx-with-error simpson-rule inverse 0.01 100) ; [835.4437770204454 832.5559396728856]
(approx-with-error simpson38-rule inverse 0.01 100) ; [627.4427811912234 624.2442845054817]
(approx-with-error booles-rule inverse 0.01 100) ; [391.7824297523125 388.1576179566068]

;; When we split the interval into two halves [0.01, 50.05] [50.05,100]
(approx-with-error trapezium-rule inverse 0.01 50.05) ; [1252.2495505293746 1250.2503495705255]
(approx-with-error trapezium-rule inverse 50.05 100) ; [0.7072645364881702 0.041486462512828726]

;; Our guess tells us that the great majority of the error is in the first sub interval
;; We might want to refine that first, before bothering with the other one:
;; We'll now split [0.01, 25.025][25.025,50.05][50.05,100]
(approx-with-error trapezium-rule inverse 0.01 25.025) ; [626.6241012183343 624.6256989814656]
(approx-with-error trapezium-rule inverse 25.025 50.05) ; [0.7083333333333333 0.04166666666666663]
(approx-with-error trapezium-rule inverse 50.05 100) ; [0.7072645364881702 0.041486462512828726]

;; Again, one subinterval seems to be responsible for the majority of our errors.

;; We could keep a list of intervals, sorted by the estimated error, and always refine the one
;; with the largest guessed error.

(defn interval->errorstruct [rule f [a b]]
  (let [[guess error-guess] (approx-with-error rule f a b)]
    [error-guess, guess, [a,b]]))

(def errorstructs (map (partial interval->errorstruct trapezium-rule inverse)
                       [[0.01,25.025][25.025 50.05][50.05 100]]))

errorstructs
;; ([624.6256989814656 626.6241012183343     [0.01   25.025]]
;;  [0.04166666666666663 0.7083333333333333  [25.025 50.05]]
;;  [0.041486462512828726 0.7072645364881702 [50.05  100]])

;; And now we need a function to refine the interval with the largest error

(defn refine[rule f errorstructs]
  (let [sortedstructs (reverse (sort errorstructs))
        [_ _ [a b]] (first sortedstructs)
        remains (rest sortedstructs)
        midpoint (/ (+ a b) 2)
        subinterval1 (interval->errorstruct rule f [a midpoint])
        subinterval2 (interval->errorstruct rule f [midpoint b])] 
        (cons subinterval1 (cons subinterval2 remains))))

;; Now with every call to refine, we refine the interval with the largest error estimate
(refine trapezium-rule inverse errorstructs)
;; ([311.93889676733556 313.93570379188156 [0.01 12.5175]]
;;  [0.04159457274964806 0.7079060863675477 [12.5175 25.025]]
;;  [0.04166666666666663 0.7083333333333333 [25.025 50.05]]
;;  [0.041486462512828726 0.7072645364881702 [50.05 100]])


(def successive-trapezium-refinements (iterate (partial refine trapezium-rule inverse) errorstructs))


;; Here's what it looks like after a few iterations
(nth successive-trapezium-refinements 5)
;; ([18.81475746721543 20.764864658796014 [0.01 0.7917187499999999]]
;;  [0.040533241059784286 0.7015625070966475 [0.7917187499999999 1.5734374999999998]]
;;  [0.04109463731966401 0.7049306354620377 [1.5734374999999998 3.136875]]
;;  [0.041379306898238544 0.7066276281534741 [3.136875 6.26375]]
;;  [0.04152264848816445 0.7074793872568832 [6.26375 12.5175]]
;;  [0.04159457274964806 0.7079060863675477 [12.5175 25.025]]
;;  [0.04166666666666663 0.7083333333333333 [25.025 50.05]]
;;  [0.041486462512828726 0.7072645364881702 [50.05 100]])

        
;; We can get our best guess for the whole thing and our total error estimate
;; by reducing this list

(reduce + (map first  (nth successive-trapezium-refinements 5))) ; 19.104035002910425
(reduce + (map second (nth successive-trapezium-refinements 5))) ; 25.708968772954105


;; After a hundred refinements..
(reduce + (map first  (nth successive-trapezium-refinements 100))) ; 0.010431101535137086
(reduce + (map second (nth successive-trapezium-refinements 100))) ; 9.213824736866899

;; After a thousand refinements..
(reduce + (map first  (nth successive-trapezium-refinements 1000))) ; 1.0913238861095381E-4
(reduce + (map second (nth successive-trapezium-refinements 1000))) ; 9.210376750235199

;; That's not bad, (the real answer is 9.210340371976184), but it's running very slowly.

;; We could try with a higher order rule

(def successive-boole-refinements (iterate (partial refine booles-rule inverse) errorstructs))
(reduce + (map first   (nth successive-boole-refinements 1000))) ; 4.420942778526893E-15
(reduce + (map second  (nth successive-boole-refinements 1000))) ; 9.210340371976176

;; In this case, that seems to work very well, but the run time is appalling.

;; The problem is that we have a longer and longer list of intervals at every
;; step, and every step, we have to sort this list. That's an n^2 algorithm,
;; which won't scale well.

;; What we should do here is use a priority queue. Clojure doesn't have an
;; immutable version, although it's possible to fake one with a sorted map.

;; But rather than do that, I'm going to drop out of the functional paradigm
;; altogther, and use the heap implementation from Java in a mutable fashion,
;; looping and popping and adding.

(defn improve-loop [rule f a b count]
  (let [pq (java.util.PriorityQueue. count (comparator (fn[a b](> (first a)(first b)))))]
    (.add pq (interval->errorstruct rule f [a b]))
    (loop [pq pq count count]
      (if (zero? count) pq
          (let [[err val [a b]] (.poll pq)
                midpoint (/ (+ a b) 2)
                aa (interval->errorstruct rule f [a midpoint])
                bb (interval->errorstruct rule f [midpoint b])]
            (doto pq
              (.add aa)
              (.add bb))
            (recur pq (dec count)))))))

;; Now we can do our calculation much faster
(defn integrate [rule f a b count]
  (let [pq (improve-loop rule f a b count)]
    [(reduce + (map first pq))
     (reduce + (map second pq))]))

;; We'll ask for a thousand refinements, and get back the error estimate, and the answer.
(integrate booles-rule inverse 0.01 100 1000) ; [4.455637248046429E-15 9.21034037197618]


;; Let's try the same integral over the very nasty range [0.0000001, 10000000] which caused serious
;; problems for our previous methods.
;; The real answer is
(- (Math/log 10000000) (Math/log 0.00000001)) ; 34.538776394910684

;; And our approximations are:
(integrate booles-rule inverse 0.00000001 10000000 10) ; [3.797743055486256E10 3.797743056542089E10]
(integrate booles-rule inverse 0.00000001 10000000 100) ; [3.3430724324184924E-5 34.53877704296225]
(integrate booles-rule inverse 0.00000001 10000000 1000) ; [4.549938203979309E-11 34.53877639491147]
(integrate booles-rule inverse 0.00000001 10000000 10000) ; [9.361001557239845E-16 34.53877639491065]

;; For the non-stiff integrals that we started playing with, Boole's rule is great:
;; It's exact for quadratics, and several higher powers
(integrate booles-rule square 0 2 10) ; [0 8/3]
(integrate booles-rule (fn[x] (* x x x x)) 0 2 10) ; [0 32/5]
(integrate booles-rule (fn[x] (* x x x x x)) 0 2 10) ; [0 32/3]

;; and very good for higher powers, even with very few refinements
(integrate booles-rule (fn[x] (* x x x x x x)) 0 2 10) ; [969/8589934592 471219269093/25769803776]
(integrate booles-rule (fn[x] (* x x x x x x)) 0 2 20) ; [2127/1099511627776 60316066438099/3298534883328]

;; convergence is great for sine
(integrate booles-rule sine 0 Math/PI 10) ; [1.7383848804897184E-9 1.9999999999725113]
(integrate booles-rule sine 0 Math/PI 100) ; [3.1931922384043077E-15 1.9999999999999991]
(integrate booles-rule sine 0 Math/PI 1000) ; [2.526233413538588E-17 1.999999999999999]
(integrate booles-rule sine 0 Math/PI 10000) ; [6.32722651455846E-18 2.0]


;; But I'm still quite worried about the error estimate that we made. It's only
;; a guess, and it can be a bad guess.  Here are some functions that are
;; deliberately designed to screw things up.

;; This function is extremely vibratey near the origin.
(defn sineinverse[x] (Math/sin (/ x)))

;; The error estimates are clearly wrong here, but the answers seem to settle down to something that looks plausible.

(integrate booles-rule sineinverse 0.001 10 1) ; [0.09752245288534744 3.189170812427795]
(integrate booles-rule sineinverse 0.001 10 10) ; [0.014802407142066881 2.725700351059874]
(integrate booles-rule sineinverse 0.001 10 100) ; [2.666579898821515E-4 2.7262259059929814]
(integrate booles-rule sineinverse 0.001 10 1000) ; [7.84363268117651E-9 2.7262019887881457]
(integrate booles-rule sineinverse 0.001 10 10000) ; [8.311387750656713E-15 2.726201989096135]

;; I'm slightly reassured that if we use the trapezium rule, which should be
;; slower converging but less sensitive to high derivatives, we seem to settle down to the same thing:

(integrate trapezium-rule sineinverse 0.001 10 1) ; [0.3493754261290617 2.9603246206316127]
(integrate trapezium-rule sineinverse 0.001 10 10) ; [0.12037643221528535 2.759621850911819]
(integrate trapezium-rule sineinverse 0.001 10 100) ; [0.011584111090323689 2.728470051290524]
(integrate trapezium-rule sineinverse 0.001 10 1000) ; [7.174438961790802E-4 2.7265014884997414]
(integrate trapezium-rule sineinverse 0.001 10 10000) ; [1.0854830311799172E-5 2.7262023437746654]

;; Since I don't actually know what the integral of sin(1/x) is, I've no idea
;; whether this answer is correct. Since both rules seem to settle down to the
;; same answer, I tend to believe that it is.

;; Here's a weird function, which looks like it should be even worse that sin(1/x) on its own
(defn strange[x] (- (Math/sin (/ x)) (/ (Math/cos (/ x)) x)))

;; In fact it's the derivative of x sin(1/x), so we can calculate the real answer over [0.001, 10]
;; which should be:
(- (* 10 (Math/sin 1/10)) (* 0.001 (Math/sin 1000))) ; 0.9975072869277496

;; Interestingly, the error estimates look sound for this one:
(integrate booles-rule strange 0.001 10 1) ; [109.91706304856582 -108.12753277035351]
(integrate booles-rule strange 0.001 10 10) ; [0.07641821305025362 -1.0276123964492345]
(integrate booles-rule strange 0.001 10 100) ; [0.0798435700032961 1.0469088424961843]
(integrate booles-rule strange 0.001 10 1000) ; [2.0359056110968434E-6 0.9975072871949854]
(integrate booles-rule strange 0.001 10 10000) ; [1.9224976990340685E-12 0.997507286927752]

;; Since we seem to have dealt well with some nasty functions, we might be
;; getting confident in our rule. I know I was!

;; But this innocuous looking function
(defn sine80squared[x] (square (Math/sin (* x 80))))

;; Is a complete nightmare. It looks as though the method is converging well:
(integrate booles-rule sine80squared 0 Math/PI 1) ; [1.1091279850485843E-28 3.7074689566598855E-28]
(integrate booles-rule sine80squared 0 Math/PI 10) ; [0.013089969389960716 0.7853981633974437]
(integrate booles-rule sine80squared 0 Math/PI 100) ; [1.7991469747360833E-12 0.7853981633974478]
(integrate booles-rule sine80squared 0 Math/PI 1000) ; [3.207733520089899E-13 0.7853981633974484]
(integrate booles-rule sine80squared 0 Math/PI 10000) ; [3.1095566849572033E-15 0.7853981633974481]

;; But if we use a different rule, it also seems to converge, but to a completely different answer
(integrate trapezium-rule sine80squared 0 Math/PI 1) ; [3.226319244612108E-28 4.262878793991289E-28]
(integrate trapezium-rule sine80squared 0 Math/PI 10) ; [0.01573134053904405 0.19774924859401588]
(integrate trapezium-rule sine80squared 0 Math/PI 100) ; [5.4528883422580115E-5 0.19634904414812832]
(integrate trapezium-rule sine80squared 0 Math/PI 1000) ; [4.6327740574637914E-7 0.19634954102557595]
(integrate trapezium-rule sine80squared 0 Math/PI 10000) ; [5.200068066397881E-9 0.19634954084967793]

;; In fact both answers are wrong. We can calculate the real integral of this
;; function over the interval [0,pi] which should be:

(/ (Math/PI) 2) ; 1.5707963267948966

;; In fact if we use our much earlier 'divide every interval evenly' algorithm:
(defn iterated-rule [rule f a b N]
  (if (= N 0)
    (rule f a b)
    (let [midpoint (+ a (/ (- b a) 2))]
      (+ (iterated-rule rule f a midpoint (dec N))
         (iterated-rule rule f midpoint b (dec N))))))

;; We very quickly get surprisingly good answers:
(iterated-rule trapezium-rule sine80squared 0 Math/PI 1) ; 1.13079223568638E-28
(iterated-rule trapezium-rule sine80squared 0 Math/PI 2) ; 4.262878793991289E-28
(iterated-rule trapezium-rule sine80squared 0 Math/PI 3) ; 3.5626606693542546E-28
(iterated-rule trapezium-rule sine80squared 0 Math/PI 4) ; 3.6535380881685736E-28
(iterated-rule trapezium-rule sine80squared 0 Math/PI 5) ; 1.5707963267948966
(iterated-rule trapezium-rule sine80squared 0 Math/PI 6) ; 1.570796326794897
(iterated-rule trapezium-rule sine80squared 0 Math/PI 7) ; 1.5707963267948972
(iterated-rule trapezium-rule sine80squared 0 Math/PI 8) ; 1.5707963267948966
(iterated-rule trapezium-rule sine80squared 0 Math/PI 9) ; 1.5707963267948957
(iterated-rule trapezium-rule sine80squared 0 Math/PI 10) ; 1.5707963267948963

;; With Booles' rule the story is the same:
(iterated-rule booles-rule sine80squared 0 Math/PI 1) ; 3.1974110932118413E-28
(iterated-rule booles-rule sine80squared 0 Math/PI 2) ; 3.7074689566598855E-28
(iterated-rule booles-rule sine80squared 0 Math/PI 3) ; 2.2340214425527414
(iterated-rule booles-rule sine80squared 0 Math/PI 4) ; 1.5358897417550046
(iterated-rule booles-rule sine80squared 0 Math/PI 5) ; 1.570796326794895
(iterated-rule booles-rule sine80squared 0 Math/PI 6) ; 1.5707963267948961
(iterated-rule booles-rule sine80squared 0 Math/PI 7) ; 1.5707963267948966
(iterated-rule booles-rule sine80squared 0 Math/PI 8) ; 1.5707963267948968
(iterated-rule booles-rule sine80squared 0 Math/PI 9) ; 1.5707963267948963


;; So the nice rule that we'd come up with, which worked so well for the stiff problem
;; of integrating 1/x near the origin, is completely broken on something that the obvious
;; recursion integrates (suspiciously) well.

;; The problem is that we used an error estimate that we can't trust to control
;; the refinement process.

;; If we guess that a certain interval contains hardly any error, then it will
;; never get refined at all, so we'll never find out that our guess is wrong.





Thursday, May 26, 2011

Numerical Integration: Better Refinements?

;; Numerical Integration: Better Refinements?

;; Here are some Newton-Cotes formulae for approximate integration:

(defn trapezium-rule [f a b]
  (* 1/2 (- b a) (+ (f a) (f b))))

(defn simpson-rule [f a b]
  (let [midpoint (+ a (/ (- b a) 2))]
    (* 1/6 (- b a) (+ (f a) (* 4 (f midpoint)) (f b)))))

(defn simpson38-rule [f a b]
  (let [midpoint1 (/ (+ a a b) 3)
        midpoint2 (/ (+ a b b) 3)]
    (* 1/8 (- b a) (+ (f a) (* 3 (f midpoint1)) (* 3 (f midpoint2)) (f b)))))

(defn booles-rule [f a b]
  (let [midpoint1 (/ (+ a a a b) 4)
        midpoint2 (/ (+ a a b b) 4)
        midpoint3 (/ (+ a b b b) 4)]
    (* 1/90 (- b a) (+ (* 7 (f a)) (* 32 (f midpoint1)) (* 12 (f midpoint2)) (* 32 (f midpoint3)) (* 7 (f b))))))


;; And here is a way to apply them to (power 2 N) subintervals
(defn iterated-rule [rule f a b N]
  (if (= N 0)
    (rule f a b)
    (let [midpoint (+ a (/ (- b a) 2))]
      (+ (iterated-rule rule f a midpoint (dec N))
         (iterated-rule rule f midpoint b (dec N))))))

;; Here are some very simple functions which we might want to test integration
;; methods on:
(defn square  [x] (* x x))
(defn sine    [x] (Math/sin x))
(defn step    [x] (if (< x 1/2) 0.0 1.0))
(defn inverse [x] (/ x))

;; We might notice that with our Newton-Cotes formulae, the change in the
;; estimate when we make a refinement is often around the same size as the
;; actual error.

;; Let's try using that to produce a method where we can say what error we'd
;; like, and if the change from refining the guess is too big, we should refine
;; more. And if we do refine more, we'll ask for half the desired error on each
;; of the two subintervals, which should mean that the total error is at least
;; comparable to the error we asked for!

(defn adaptive-rule-recurse [rule f a b desired-error]
  (let [guess (rule f a b)
        midpoint (/ (+ a b) 2)
        better-guess (+ (rule f a midpoint) (rule f midpoint b))
        error-estimate (- guess better-guess)
        abs-error-estimate (if (> error-estimate 0) error-estimate (- error-estimate))]
    (if (< abs-error-estimate desired-error) better-guess
        (let [half-desired-error (/ desired-error 2)]
          (+ (adaptive-rule-recurse rule f a midpoint half-desired-error)
             (adaptive-rule-recurse rule f midpoint b half-desired-error))))))

(adaptive-rule-recurse trapezium-rule square 0. 2 0.1) ; 2.6875
(adaptive-rule-recurse trapezium-rule square 0. 2 0.01) ; 2.66796875
(adaptive-rule-recurse trapezium-rule square 0. 2 0.001) ; 2.6669921875

(adaptive-rule-recurse trapezium-rule step 0.0001 1 0.1) ; 0.5
(adaptive-rule-recurse trapezium-rule step 0.0001 1 0.01) ; 0.5
(adaptive-rule-recurse trapezium-rule step 0.0001 1 0.001) ; 0.5
(adaptive-rule-recurse trapezium-rule step 0.0001 1 0.0001) ; 0.5

;; Here we're using the trapezium rule on the integral that we were previously unable to get
;; a good answer for: remember that the correct answer is 9.210340371976182

(adaptive-rule-recurse trapezium-rule inverse 0.0001 1 0.1)    ; 9.234002964342716
(adaptive-rule-recurse trapezium-rule inverse 0.0001 1 0.01)   ; 9.211881820961814
(adaptive-rule-recurse trapezium-rule inverse 0.0001 1 0.001)  ; 9.210518109406467
(adaptive-rule-recurse trapezium-rule inverse 0.0001 1 0.0001) ; 9.210358164670637

;; At this point, we're getting much better answers than we ever got before, but they've started
;; taking noticeable time to compute.

;; However, we still retain the option of using the higher order formulae:

(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.1) ; 9.210347324857782
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.01) ; 9.210345994304586
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.001) ; 9.210344014376869
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.0001) ; 9.21034116413936
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.00001) ; 9.210340404907965
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.000001) ; 9.210340376142819
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.0000001) ; 9.210340372376441
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.00000001) ; 9.210340372016345
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.000000001) ; 9.21034037198001
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.0000000001) ; 9.210340371976589
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.00000000001) ; 9.210340371976214
(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.000000000001) ; 9.210340371976185

;; These answers come back instantaneously, even when we're calculating the
;; previously impossible answer to the question to 12 decimal places.

;; But beware! If we ask for too much accuracy, then the effect of floating
;; point noise means that our recursion may never terminate:

(adaptive-rule-recurse booles-rule inverse 0.0001 1 0.0000000000001) ; freezes REPL!


;; At this point, I'm very tempted to say:

(defn integrate [f a b]
  (adaptive-rule-recurse booles-rule f a b 0.00000001))

;; Famously, the integral of 1/x diverges (very slowly) as x gets large
;; which makes it quite difficult to calculate for large intervals without using special tricks.

;; The integral from 1 to 1000000 of 1/x is log 1000000 - log 1

(- (Math/log 1000000) (Math/log 1)) ; 13.815510557964274

;; Let's try calculating that with our new integrate function:

(time
 (integrate (fn[x] (/ 1.0 x)) 1 1000000))

;; "Elapsed time: 293.219062 msecs"
;; 13.81551055800455

;; I think that's pretty good.

;; Unfortunately, we can still lock up the computer by asking the wrong question

(- (Math/log 100000000) (Math/log 1)) ; 18.420680743952367
;; but:
(time
 (integrate (fn[x] (/ 1.0 x)) 1 100000000)) ; Freezes REPL


;; Can you work out what's going on and what we could do about it?

Numerical Integration: Better Rules?

;; Numerical Integration: Better Rules?

;; So far we've found a couple of approximations to 'the area under a graph'.

(defn trapezium-rule [f a b]
  (* 1/2 (- b a) (+ (f a) (f b))))

(defn simpson-rule [f a b]
  (let [midpoint (+ a (/ (- b a) 2))]
    (* 1/6 (- b a) (+ (f a) (* 4 (f midpoint)) (f b)))))

;; And a way of repeatedly splitting intervals and applying the rules to the sub-intervals
;; to produce better and better approximations.
(defn iterated-rule [rule f a b N]
  (if (= N 0)
    (rule f a b)
    (let [midpoint (+ a (/ (- b a) 2))]
      (+ (iterated-rule rule f a midpoint (dec N))
         (iterated-rule rule f midpoint b (dec N))))))


;; For these two functions, which are nice and smooth, with derivatives that
;; don't get too large, these rules produce good approximations
(defn square[x] (* x x))
(defn sine [x] (Math/sin x))

;; For this one, which isn't smooth, the approximations converge slowly:
(defn step [x]  (if (< x 1/2) 0.0 1.0))

;; For this, which is smooth, but which blows up near 0, the approximations are bad
(defn inverse [x] (/ x))


;; One approach which is often taken is to use 'more accurate' rules.

;; The trapezium rule and Simpson's rule are members of a family called 'Newton-Cotes' formulas.
;; The more complicated a Newton-Cotes formula is, the more polynomials it can integrate exactly.

;; But in fact, for nice smooth functions, they tend to be more accurate the
;; more polynomials they can integrate exactly.

;; Here are two more examples, and their estimates for the integral of sine over
;; a half circle (which should be exactly 2):

(defn simpson38-rule [f a b]
  (let [midpoint1 (/ (+ a a b) 3)
        midpoint2 (/ (+ a b b) 3)]
    (* 1/8 (- b a) (+ (f a) (* 3 (f midpoint1)) (* 3 (f midpoint2)) (f b)))))

(simpson38-rule sine 0 Math/PI) ; 2.040524284763495

(defn booles-rule [f a b]
  (let [midpoint1 (/ (+ a a a b) 4)
        midpoint2 (/ (+ a a b b) 4)
        midpoint3 (/ (+ a b b b) 4)]
    (* 1/90 (- b a) (+ (* 7 (f a)) (* 32 (f midpoint1)) (* 12 (f midpoint2)) (* 32 (f midpoint3)) (* 7 (f b))))))

(booles-rule sine 0 Math/PI) ; 1.9985707318238355

;; We can use the same approach to getting better estimates by subdividing: For
;; sine, as well as getting better estimates to start with, these rules have
;; high rates of convergence. It doesn't take many subdivisions before we start
;; to run into the limits of floating point accuracy.

(iterated-rule booles-rule sine 0 Math/PI 0) ; 1.9985707318238355
(iterated-rule booles-rule sine 0 Math/PI 1) ; 1.9999831309459855
(iterated-rule booles-rule sine 0 Math/PI 2) ; 1.9999997524545716
(iterated-rule booles-rule sine 0 Math/PI 3) ; 1.9999999961908446
(iterated-rule booles-rule sine 0 Math/PI 4) ; 1.9999999999407074
(iterated-rule booles-rule sine 0 Math/PI 5) ; 1.999999999999074
(iterated-rule booles-rule sine 0 Math/PI 6) ; 1.9999999999999853
(iterated-rule booles-rule sine 0 Math/PI 7) ; 1.9999999999999993
(iterated-rule booles-rule sine 0 Math/PI 8) ; 1.9999999999999998
(iterated-rule booles-rule sine 0 Math/PI 9) ; 1.9999999999999998


(iterated-rule simpson38-rule sine 0 Math/PI 0) ; 2.040524284763495
(iterated-rule simpson38-rule sine 0 Math/PI 1) ; 2.002009846628558
(iterated-rule simpson38-rule sine 0 Math/PI 2) ; 2.000119386415226
(iterated-rule simpson38-rule sine 0 Math/PI 3) ; 2.000007370036249
(iterated-rule simpson38-rule sine 0 Math/PI 4) ; 2.000000459216732
(iterated-rule simpson38-rule sine 0 Math/PI 5) ; 2.0000000286790867
(iterated-rule simpson38-rule sine 0 Math/PI 6) ; 2.0000000017921
(iterated-rule simpson38-rule sine 0 Math/PI 7) ; 2.000000000112001
(iterated-rule simpson38-rule sine 0 Math/PI 8) ; 2.0000000000069997
(iterated-rule simpson38-rule sine 0 Math/PI 9) ; 2.000000000000438


;; For the step function, however, boole's rule isn't really that much better
;; than the trapezium rule

(iterated-rule booles-rule step 0 1 0) ; 0.5666666666666667
(iterated-rule booles-rule step 0 1 1) ; 0.5388888888888889
(iterated-rule booles-rule step 0 1 2) ; 0.5194444444444445
(iterated-rule booles-rule step 0 1 3) ; 0.5097222222222222
(iterated-rule booles-rule step 0 1 4) ; 0.5048611111111111
(iterated-rule booles-rule step 0 1 5) ; 0.5024305555555555
(iterated-rule booles-rule step 0 1 6) ; 0.5012152777777777
(iterated-rule booles-rule step 0 1 7) ; 0.5006076388888889
(iterated-rule booles-rule step 0 1 8) ; 0.5003038194444445
(iterated-rule booles-rule step 0 1 9) ; 0.5001519097222222

;; Both seem to need double the number of points to halve their error.

(iterated-rule trapezium-rule step 0 1 0) ; 0.5
(iterated-rule trapezium-rule step 0 1 1) ; 0.75
(iterated-rule trapezium-rule step 0 1 2) ; 0.625
(iterated-rule trapezium-rule step 0 1 3) ; 0.5625
(iterated-rule trapezium-rule step 0 1 4) ; 0.53125
(iterated-rule trapezium-rule step 0 1 5) ; 0.515625
(iterated-rule trapezium-rule step 0 1 6) ; 0.5078125
(iterated-rule trapezium-rule step 0 1 7) ; 0.50390625
(iterated-rule trapezium-rule step 0 1 8) ; 0.501953125
(iterated-rule trapezium-rule step 0 1 9) ; 0.5009765625

;; Performance is similarly bad for boole's rule on the inverse function

(iterated-rule booles-rule inverse 0.0001 1 0) ; 779.9400477089192
(iterated-rule booles-rule inverse 0.0001 1 1) ; 391.7824297523124
(iterated-rule booles-rule inverse 0.0001 1 2) ; 198.04899407170583
(iterated-rule booles-rule inverse 0.0001 1 3) ; 101.52648013049377
(iterated-rule booles-rule inverse 0.0001 1 4) ; 53.607079025964396
(iterated-rule booles-rule inverse 0.0001 1 5) ; 29.984599583713347
(iterated-rule booles-rule inverse 0.0001 1 6) ; 18.501553032592827
(iterated-rule booles-rule inverse 0.0001 1 7) ; 13.071085113384171
(iterated-rule booles-rule inverse 0.0001 1 8) ; 10.635953348374398
(iterated-rule booles-rule inverse 0.0001 1 9) ; 9.647557264415854


(iterated-rule trapezium-rule inverse 0.0001 1 0) ; 4999.99995
(iterated-rule trapezium-rule inverse 0.0001 1 1) ; 2500.9997750199977
(iterated-rule trapezium-rule inverse 0.0001 1 2) ; 1251.8327765203333
(iterated-rule trapezium-rule inverse 0.0001 1 3) ; 627.5916420975113
(iterated-rule trapezium-rule inverse 0.0001 1 4) ; 315.8156999790102
(iterated-rule trapezium-rule inverse 0.0001 1 5) ; 160.27209317742054
(iterated-rule trapezium-rule inverse 0.0001 1 6) ; 82.84288624590312
(iterated-rule trapezium-rule inverse 0.0001 1 7) ; 44.46707207824598
(iterated-rule trapezium-rule inverse 0.0001 1 8) ; 25.61044440290146
(iterated-rule trapezium-rule inverse 0.0001 1 9) ; 16.499072595032356


;; So although it seems these higher order Newton-Cotes formulae are much more
;; accurate and faster converging on well behaved functions, they don't seem to
;; help much in integrating anything tricky.

Wednesday, May 25, 2011

Numerical Integration: Harder Functions

;; Numerical Integration: Harder Functions

;; So far we've found a couple of approximations to 'the area under a graph'.

(defn trapezium-rule [f a b]
  (* 1/2 (- b a) (+ (f a) (f b))))

(defn simpson-rule [f a b]
  (let [midpoint (+ a (/ (- b a) 2))]
    (* 1/6 (- b a) (+ (f a) (* 4 (f midpoint)) (f b)))))

;; And a way of repeatedly splitting intervals and applying the rules to the sub-intervals
;; to produce better and better approximations.
(defn iterated-rule [rule f a b N]
  (if (= N 0)
    (rule f a b)
    (let [midpoint (+ a (/ (- b a) 2))]
      (+ (iterated-rule rule f a midpoint (dec N))
         (iterated-rule rule f midpoint b (dec N))))))

;; We know that Simpson's rule is exact for our original function

(defn square[x] (* x x))

(simpson-rule square 0 2) ; 8/3
(iterated-rule simpson-rule square 0 2 1) ; 8/3
(iterated-rule simpson-rule square 0 2 2) ; 8/3

;; How do we do with some other functions?

;; Another function where we can calculate the answer directly is sine
(defn sine [x] (Math/sin x))

;; The integral of sine over a half-turn (i.e. over the interval [0,pi]) is exactly 2.

;; here are the first few approximations with the trapezium rule:
(take 10 (map (partial iterated-rule trapezium-rule sine 0 Math/PI) (iterate inc 0)))
;; (1.9236706937217898E-16 1.5707963267948968 1.8961188979370398
;; 1.9742316019455508 1.993570343772339 1.9983933609701445 1.9995983886400377
;; 1.9998996001842024 1.9999749002350526 1.9999937250705755)

;; After an inauspicious start, the answer is clearly settling down to 2.

;; Simpson's rule also settles down to 2, but it's much quicker.
(take 10 (map (partial iterated-rule simpson-rule sine 0 Math/PI) (iterate inc 0)))
;; (2.094395102393196 2.0045597549844216 2.000269169948388 2.0000165910479364
;; 2.000001033369414 2.0000000645300027 2.000000004032258 2.000000000252003
;; 2.000000000015751 2.000000000000985)

;; So it looks like we're onto a winner so far.

;; Another type of function that doesn't do so well is:

(defn step [x]  (if (< x 1/2) 0 1))

;; It should be fairly obvious that the integral of this over [0,1] is 0.5, but
;; in fact the convergence of the methods is very slow compared to what we had
;; for sine or square.

(take 10 (map (partial iterated-rule trapezium-rule step 0. 1) (iterate inc 0)))
;; (0.5 0.75 0.625 0.5625 0.53125 0.515625 0.5078125 0.50390625 0.501953125 ...

(take 10 (map (partial iterated-rule simpson-rule step 0. 1) (iterate inc 0)))
;; (0.8333333333333336 0.5833333333333335 0.5416666666666667 0.5208333333333335
;; 0.5104166666666667 0.5052083333333335 0.5026041666666667 0.5013020833333335
;; 0.5006510416666667 0.5003255208333335 )

;; Notice how if we make the tiniest possible change to the function, which doesn't change the integral at all:
(defn evil-step [x]  (if (<= x 1/2) 0 1))

;; the approximations are all changed quite a lot, although they do seem to
;; still converge to the correct answer.

(take 10 (map (partial iterated-rule trapezium-rule evil-step 0. 1) (iterate inc 0)))
;; (0.5 0.25 0.375 0.4375 0.46875 0.484375 0.4921875 0.49609375 0.498046875 0.4990234375)
(take 10 (map (partial iterated-rule simpson-rule evil-step 0. 1) (iterate inc 0)))
;; (0.1666666666666667 0.4166666666666668 0.4583333333333335 0.4791666666666668
;; 0.4895833333333335 0.4947916666666668 0.4973958333333335 0.4986979166666668
;; 0.4993489583333335 0.4996744791666668)



;; What about if we're integrating a (moderately) badly behaved function?:

(defn inverse [x] (/ x))

;; the real answer, by devious mathematical trickery is (log a) - (log b)

;; So if we integrate over the region [0.0001, 1], where the values of 1/x are
;; sometimes very large, we should get:

(- (Math/log 1) (Math/log 0.0001)) ; 9.210340371976182

(take 10 (map (partial iterated-rule trapezium-rule inverse 0.0001 1) (iterate inc 0)))
;; (4999.99995 2500.9997750199977 1251.8327765203333 627.5916420975113
;; 315.8156999790102 160.27209317742054 82.84288624590312 44.46707207824598
;; 25.61044440290146 16.499072595032356)

(take 10 (map (partial iterated-rule simpson-rule inverse 0.0001 1) (iterate inc 0)))
;; (1667.9997166933313 835.4437770204453 419.5112639565708 211.89038593950997
;; 108.42422424355732 57.03315060206397 31.67513402236028 19.324901844453297
;; 13.461948659075995 10.812578055293251)

;; It looks like both the rules, after starting off with appalling first
;; guesses, are converging to something, and maybe even to the right answer, but
;; the convergence is very slow.  Remember that to get the tenth element of this
;; sequence we're splitting the interval up into 1024 pieces, and we haven't
;; even got the answer right to one significant figure.

;; Numerical Integration can give very misleading answers if it's used naively.
;; Quite often, the results of a careless numerical simulation will just 'look
;; wrong' to a trained eye, and in those cases it's usually the eye that's in
;; the right.

;; One way to be reassured that your answer is roughly correct is to alter
;; expected degree of accuracy of the method, and to check that the answer
;; doesn't change by much!






























Numerical Integration: Better Approximations

;; Numerical Integration: Better Approximations

;; Remember our example function:

(defn square[x] (* x x))

;; We're trying to find ways to calculate the 'area under its graph between 0
;; and 2', or the 'integral over the interval [0,2]'

;; A better estimate of this area under the graph is to imagine a trapezium
;; which has its base corners at (0,0) and (2,0), and the top corners at
;; (0,(square 0)) (2, (square 2)), and calculate the area of that.

;; More generally, if the interval we're interested is [a,b], and the function's
;; values there are fa and fb, then the area of the trapezium will just be:

(defn trapezium [a fa b fb]
  (* 1/2 (- b a) (+ fa fb)))

;; Another way to think about that is that it's the length of the interval
;; multiplied by the average of the values of the function at the ends.

;; So another approximation to the integral of f over [a,b] is:

(defn trapezium-rule [f a b]
  (trapezium a (f a) b (f b)))

(trapezium-rule square 0 2) ; 4

;; We can make another approximation by using the trapezium rule on the two
;; subintervals [0,1] and [1,2] and adding the results

(+ (trapezium-rule square 0 1)
   (trapezium-rule square 1 2)) ; 3

;; And an even better one by splitting those subintervals in half

(+ (trapezium-rule square 0 1/2)
   (trapezium-rule square 1/2 1)
   (trapezium-rule square 1 3/2)
   (trapezium-rule square 3/2 2)) ; 11/4

;; And so on...
(defn iterated-rule [rule f a b N]
  (if (= N 0)
    (rule f a b)
    (let [midpoint (+ a (/ (- b a) 2))]
      (+ (iterated-rule rule f a midpoint (dec N))
         (iterated-rule rule f midpoint b (dec N))))))

;; This converges fairly nicely:
(map (partial iterated-rule trapezium-rule square 0 2) (range 10))
;; (4 3 11/4 43/16 171/64 683/256 2731/1024 10923/4096 43691/16384 174763/65536)

(map (partial - 8/3)
     (map (partial iterated-rule trapezium-rule square 0 2) (range 10)))
;; (-4/3 -1/3 -1/12 -1/48 -1/192 -1/768 -1/3072 -1/12288 -1/49152 -1/196608)

;; We now only need a thousand samples of the function to get the answer
;; accurate to one part in 100000.

;; But an even nicer approximation is Simpson's rule, which involves fitting a
;; parabola to the two end points and the midpoint, and calculating the area
;; under that.

;; That's equivalent to taking a weighted sum of the values of f at the
;; beginning, midpoint, and end of the interval, with weights 1:4:1

(defn simpson-rule [f a b]
  (let [midpoint (+ a (/ (- b a) 2))]
    (* 1/6 (- b a) (+ (f a) (* 4 (f midpoint)) (f b)))))

;; For the square function, which is itself a parabola, this rule actually
;; calculates the area exactly!
(simpson-rule square 0 2) ; 8/3

;; That about wraps it up for numerical approximations to the integrals of
;; quadratics, but they are easy to calculate exactly anyway.





























Numerical Integration: What is an Integral?

I've found myself needing to take a few difficult integrals recently, as part of some machine learning algorithms, and I ended up calculating them with numerical approximations.

Frustration at the amount of time they were taking lead me to progressively improve my methods, which got very much faster as I played with them.

Weirdly, my initial instincts had been to translate my programs into either C or Java, which might well have resulted in 100x speed ups over the obvious implementations in Clojure, but in fact the clarity of expression afforded by writing in a language where recursion is a natural thing to do lead me, by making fairly obvious changes, to some very fast algorithms.

Of course, I could still translate them to C or Java, or optimized Clojure, and they'd still get a lot faster, but it hardly seems worth the bother now.


;; Numerical Integration I: What is an integral, anyway?

;; What does it mean to 'evaluate the integral' of a function over an interval?

;; Let's take an example function

(defn square[x] (* x x))

;; We could evaluate it at a few input values:
(map square [0 1 2]) ; (0 1 4)

;; The integral of a function over an interval is a sort of
;; 'average sum over all the possible values in the interval'.

;; A first approximation to the integral of square over the interval [0,2]
;; would be to say:

;; Let's pretend that the value of square between 0 and 1 is like (square 0)
;; and the value between 1 and 2 is like (square 1)

;; So to find our approximation we add (square 0) to (square 1):

(reduce + (map square [0 1])) ; 1

;; Another approximation, equally valid, would be to say that the value between
;; 0 and 1 is like (square 1) so we could equally well add the value at 1 to the
;; value at 2

(reduce + (map square [1 2])) ; 5

;; These answers are quite different! One is too low, because we paid too much
;; attention to the numbers at the start of the sub-ranges. One is too high,
;; because we added up numbers from the ends.

;; We can make both answers more accurate by sampling the function at more
;; points, and dividing the sum by an appropriate factor to make up for having
;; more points.

(/ (reduce + (map square [0 1/2 1 3/2]  )) 2) ; 7/4
(/ (reduce + (map square   [1/2 1 3/2 2])) 2) ; 15/4

;; We can continue the procedure, sampling the function at twice as many points again:

(/ (reduce + (map square [0 1/4 1/2 3/4 1 5/4 6/4 7/4]  )) 4) ; 35/16
(/ (reduce + (map square   [1/4 1/2 3/4 1 5/4 6/4 7/4 2])) 4) ; 51/16

;; It's clear that these values will get closer to one another the more points
;; we sample at

;; If we continue the procedure indefinitely:
(defn riemann-lower-sum [f a b N]
  (let [points (range a b (/ N))]
    (/ (reduce + (map f points)) N)))

(map (partial riemann-lower-sum square 0 2) (iterate inc 1)) ; (1 7/4 55/27 35/16 57/25 253/108 ...)
;; We find that the answers settle down to a particular number, and in this case
;; it looks like they're settling down to 8/3

;; However we do need to take quite a lot of samples to make the approximation get close!
(take 15 (map (partial - 8/3)
              (map (partial riemann-lower-sum square 0 2)
                   (iterate (partial * 2) 1))))
;; (5/3 11/12 23/48 47/192 95/768 191/3072 383/12288 767/49152 1535/196608
;; 3071/786432 6143/3145728 12287/12582912 24575/50331648 49151/201326592
;; 98303/805306368)

;; The last number here is what we get if we keep splitting our sub-intervals in
;; half fifteen times, and that's quite a lot of sub-intervals. (1, 2, 4, 8, 16 ...)

;; Strictly speaking, the riemann lower sum is a sum of the lowest values of
;; the function on the subintervals, but because the square function is
;; monotonically increasing over our range, the function as we defined it has
;; the same value.

;; If this procedure has a limit, then the integral is defined as this limit.
;; For well behaved functions, this limit will exist.

;; The limit can be thought of as the area under the graph of f from a to b.

;; At school we learn to calculate this limit by finding an antiderivative to
;; the original function, and calculating how much it changes over the interval.

;; An antiderivative of (fn[x] (* x x)) is (fn[x] (* 1/3 x x x))

(let [anti (fn[x] (* 1/3 x x x))]
  (- (anti 2) (anti 0))) ; 8/3

;; Although this is clearly a much better way to calculate integrals than by
;; taking millions of function samples and adding them up, it's often quite hard
;; to find antiderivatives even for quite simple functions, and for some quite
;; simple functions, no easily expressible antiderivative exists.

;; So we quite often find ourselves needing to calculate approximations by computer,
;; and this is called numerical integration.



Followers