Schema (using Malli) (#9)

* schema for input args * refac * schema for input args * schema for GA pheno population * schema for GA pheno population * schema for GA pheno population * schema for GA pheno mutation * schema for GA phenos * schema * schema failure case test * more schemas * more schemas * CLI args schema * eval schema / refac to ns * apply mods schema -> somewhat of a perf hit can be felt now * WIP failing to de-intrument all schema on command * log points * ns reorg * test style, coverage update * mods style, bias fewer * more schema
ogeagla · Feb 16, 2024 · c3d83a2 · c3d83a2
1 parent e94f071
commit c3d83a2
Show file tree

Hide file tree

Showing 23 changed files with 371 additions and 74 deletions.
diff --git a/README.md b/README.md
@@ -106,7 +106,7 @@ An example of using the GUI:
 
 Coverage looks like this if you run `lein cloverage`:
 
-![test_coverage_2024-02-15_10-13.png](screenshots%2Ftest_coverage_2024-02-15_10-13.png)
+![test_coverage_2024-02-16_11-15.png](screenshots%2Ftest_coverage_2024-02-16_11-15.png)
 
 ## How It Works
 

diff --git a/nomis-ns-graph.png b/nomis-ns-graph.png
diff --git a/project.clj b/project.clj
@@ -5,6 +5,7 @@
             :url  "https://www.eclipse.org/legal/epl-2.0/"}
   :dependencies [[org.clojure/clojure "1.11.1"]
                  [org.clojure/core.async "1.6.681"]
+                 [metosin/malli "0.14.0"]
 
                  [org.clojure/tools.cli "1.0.219"]
 

diff --git a/screenshots/test_coverage_2024-02-16_11-15.png b/screenshots/test_coverage_2024-02-16_11-15.png
diff --git a/src/closyr/core.clj b/src/closyr/core.clj
@@ -3,8 +3,8 @@
   (:require
     [clojure.string :as str]
     [clojure.tools.cli :as cli]
-    [closyr.dataset.csv :as input-csv]
-    [closyr.log :as log]
+    [closyr.util.csv :as input-csv]
+    [closyr.util.log :as log]
     [closyr.symbolic-regression :as symreg])
   (:import
     (java.io

diff --git a/src/closyr/ga.clj b/src/closyr/ga.clj
@@ -1,8 +1,8 @@
 (ns closyr.ga
   (:refer-clojure :exclude [rand rand-int rand-nth shuffle])
   (:require
-    [closyr.log :as log]
-    [closyr.dataset.prng :refer :all]))
+    [closyr.util.log :as log]
+    [closyr.util.prng :refer :all]))
 
 
 (set! *warn-on-reflection* true)

diff --git a/src/closyr/ops.clj b/src/closyr/ops.clj
@@ -3,11 +3,12 @@
   (:require
     [clojure.core.async :as async :refer [go go-loop timeout <!! >!! <! >! chan put! take! alts!! alts! close!]]
     [clojure.string :as str]
-    [closyr.dataset.prng :refer :all]
-    [closyr.log :as log]
+    [closyr.util.prng :refer :all]
+    [closyr.util.log :as log]
     [closyr.ops.common :as ops-common]
     [closyr.ops.eval :as ops-eval]
-    [closyr.ops.modify :as ops-modify])
+    [closyr.ops.modify :as ops-modify]
+    [closyr.util.spec :as specs])
   (:import
     (java.text
       DecimalFormat)
@@ -74,7 +75,9 @@
       (and (number? n) (Double/isNaN n))))
 
 
-(defn- compute-residual
+(defn compute-residual
+  "Compute the residual (difference) between 2 number (y-values)"
+  {:malli/schema [:=> [:cat number? number?] number?]}
   [expected actual]
   (let [res (if (not-finite? actual)
               max-resid
@@ -281,6 +284,7 @@
 
       (reset! test-timer* (Date.))
       (log/info i "-step pop size: " pop-size
+                " points: " (count input-ys-vec)
                 " max leafs: " max-leafs
                 " took secs: " took-s
                 " phenos/s: " (Math/round ^double (/ (* pop-size *log-steps*) took-s))
@@ -303,3 +307,6 @@
                              :best-p95-score        (:score best-p95-v)
                              :best-p90-score        (:score best-p90-v)}))))
   (reset! sim-stats* {}))
+
+
+(specs/instrument-all!)
diff --git a/src/closyr/ops/common.clj b/src/closyr/ops/common.clj
@@ -2,8 +2,9 @@
   (:refer-clojure :exclude [rand rand-int rand-nth shuffle])
   (:require
     [clojure.core.async :as async :refer [go go-loop timeout <!! >!! <! >! chan put! take! alts!! alt!! close!]]
-    [closyr.dataset.prng :refer :all]
-    [closyr.log :as log])
+    [closyr.util.prng :refer :all]
+    [closyr.util.log :as log]
+    [closyr.util.spec :as specs])
   (:import
     (java.util
       Date
@@ -257,6 +258,7 @@
 
 (defn extend-xs
   "Add extra xs on either side of the provided range"
+  {:malli/schema [:=> [:cat [:sequential number?]] map?]}
   [input-xs-vec]
   (let [x-min                (first input-xs-vec)
         x-max                (last input-xs-vec)
@@ -284,3 +286,6 @@
      :x-head-list x-head-list
      :x-tail      x-tail
      :x-tail-list x-tail-list}))
+
+
+(specs/instrument-all!)
diff --git a/src/closyr/ops/eval.clj b/src/closyr/ops/eval.clj
@@ -1,7 +1,8 @@
 (ns closyr.ops.eval
   (:require
-    [closyr.log :as log]
-    [closyr.ops.common :as ops-common])
+    [closyr.util.log :as log]
+    [closyr.ops.common :as ops-common]
+    [closyr.util.spec :as specs])
   (:import
     (org.matheclipse.core.eval
       EvalControlledCallable
@@ -38,6 +39,7 @@
 
 (defn ^IExpr eval-phenotype-on-expr-args
   "Eval an expr at every point in the args"
+  {:malli/schema [:=> [:cat #'specs/GAPhenotype some?] any?]}
   [{^IAST expr :expr ^ISymbol x-sym :sym ^ExprEvaluator util :util p-id :id :as pheno}
    ^"[Lorg.matheclipse.core.interfaces.IExpr;" expr-args]
   (try
@@ -90,8 +92,9 @@
 
 (defn eval-vec-pheno
   "Evaluate a phenotype's expr on input xs/ys vecs"
+  {:malli/schema [:=> [:cat #'specs/GAPhenotype #'specs/SolverEvalArgs] [:or [:vector number?] nil?]]}
   [p
-   {:keys [input-xs-list input-xs-count input-ys-vec]
+   {:keys [input-xs-list input-xs-count]
     :as   run-args}]
   (let [^IExpr new-expr (:expr p)
         ^IExpr eval-p   (eval-phenotype-on-expr-args p input-xs-list)]
@@ -164,3 +167,6 @@
 
     {:xs xs
      :ys evaluated-ys}))
+
+
+(specs/instrument-all!)
diff --git a/src/closyr/ops/initialize.clj b/src/closyr/ops/initialize.clj
@@ -1,7 +1,8 @@
 (ns closyr.ops.initialize
   (:require
-    [closyr.log :as log]
-    [closyr.ops.common :as ops-common])
+    [closyr.util.log :as log]
+    [closyr.ops.common :as ops-common]
+    [closyr.util.spec :as specs])
   (:import
     (org.matheclipse.core.expression
       AST
@@ -21,6 +22,7 @@
 
 (defn initial-phenotypes
   "Initial exprs scaled up in quantity to use in GA evolution"
+  {:malli/schema [:=> [:cat pos-int?] #'specs/GAPopulationPhenotypes]}
   [p-count]
   (let [^ISymbol x ops-common/sym-x]
     (->>
@@ -31,6 +33,7 @@
 
 (defn initial-mutations
   "All mutations to use in GA evolutions"
+  {:malli/schema [:=> [:cat] [:vector #'specs/GAMutation]]}
   []
   [{:op          :modify-fn
     :label       "Derivative"
@@ -61,8 +64,8 @@
     :modifier-fn (fn ^IExpr [{^IAST expr :expr ^ISymbol x-sym :sym :as pheno}]
                    (F/Plus expr (F/Divide 1 F/C100)))}
 
-   {:op         :modify-fn
-    :label      "-1/100"
+   {:op          :modify-fn
+    :label       "-1/100"
     :modifier-fn (fn ^IExpr [{^IAST expr :expr ^ISymbol x-sym :sym :as pheno}]
                    (F/Subtract expr (F/Divide 1 F/C100)))}
 
@@ -659,3 +662,6 @@
                         (if (ops-common/should-modify-branch leaf-count pheno)
                           (F/Subtract ie (F/num 0.1))
                           ie))}])
+
+
+(specs/instrument-all!)
diff --git a/src/closyr/ops/modify.clj b/src/closyr/ops/modify.clj
@@ -3,9 +3,10 @@
   (:require
     [clojure.core.async :as async :refer [go go-loop timeout <!! >!! <! >! chan put! take! alts!! alt!! close!]]
     [clojure.string :as str]
-    [closyr.dataset.prng :refer :all]
-    [closyr.log :as log]
-    [closyr.ops.common :as ops-common])
+    [closyr.ops.common :as ops-common]
+    [closyr.util.log :as log]
+    [closyr.util.prng :refer :all]
+    [closyr.util.spec :as specs])
   (:import
     (java.util.function
       Function)
@@ -139,6 +140,17 @@
 
 (defn crossover
   "Do phenotype crossover on their expr AST"
+  {:malli/schema
+   [:=>
+
+    ;; inputs:
+    [:cat
+     pos-int?
+     #'specs/GAPhenotype
+     #'specs/GAPhenotype]
+
+    ;; outputs:
+    #'specs/GAPhenotype]}
   [max-leafs
    {^IAST e1 :expr ^ISymbol x-sym :sym ^ExprEvaluator util :util :as p}
    {^IAST e2 :expr :as p-discard}]
@@ -184,6 +196,23 @@
 
 (defn apply-modifications
   "Apply a sequence of modifications"
+  {:malli/schema
+   [:=>
+
+    ;; inputs:
+    [:cat
+     pos-int?
+     pos-int?
+     [:sequential #'specs/GAMutation]
+     #'specs/GAPhenotype
+     #'specs/GAPhenotype]
+
+    ;; outputs:
+    [:map {:closed true}
+     [:new-pheno #'specs/GAPhenotype]
+     [:iters int?]
+     [:mods [:sequential #'specs/GAMutation]]]]}
+
   [max-leafs mods-count initial-muts p-winner p-discard]
   (loop [iters              0
          mods-left-to-apply mods-count
@@ -237,26 +266,29 @@
     (concat (repeat 40 1))
     (concat (repeat 30 2))
     (concat (repeat 20 3))
-    (concat (repeat 18 4))
-    (concat (repeat 16 5))
-    (concat (repeat 15 6))
-    (concat (repeat 14 7))
-    (concat (repeat 13 8))
-    (concat (repeat 12 9))
-    (concat (repeat 11 10))
-    (concat (repeat 10 11))
-    (concat (repeat 9 12))
-    (concat (repeat 8 13))
-    (concat (repeat 7 14))
-    (concat (repeat 6 15))
-    (concat (repeat 5 16))
-    (concat (repeat 4 17))
-    (concat (repeat 3 18))
-    (concat (repeat 2 19))
-    (concat (repeat 1 20))
+    (concat (repeat 15 4))
+    (concat (repeat 12 5))
+    (concat (repeat 10 6))
+    (concat (repeat 9 7))
+    (concat (repeat 8 8))
+    (concat (repeat 7 9))
+    (concat (repeat 6 10))
+    (concat (repeat 5 11))
+    (concat (repeat 4 12))
+    (concat (repeat 3 13))
+    (concat (repeat 2 14))
+    (concat (repeat 1 15))
+    ;; (concat (repeat 5 16))
+    ;; (concat (repeat 4 17))
+    ;; (concat (repeat 3 18))
+    ;; (concat (repeat 2 19))
+    ;; (concat (repeat 1 20))
     ;; (concat (repeat 3 21))
     ;; (concat (repeat 2 22))
     ;; (concat (repeat 1 23))
     ;; (concat (repeat 1 24))
     ;; (concat (repeat 1 25))
     vec))
+
+
+(specs/instrument-all!)
diff --git a/src/closyr/symbolic_regression.clj b/src/closyr/symbolic_regression.clj
@@ -2,12 +2,14 @@
   (:require
     [clojure.core.async :as async :refer [go go-loop timeout <!! >!! <! >! chan put! take! alts!! alts! close!]]
     [closyr.ga :as ga]
-    [closyr.log :as log]
+    [closyr.util.log :as log]
     [closyr.ops :as ops]
     [closyr.ops.common :as ops-common]
     [closyr.ops.initialize :as ops-init]
+    [closyr.util.spec :as specs]
     [closyr.ui.gui :as gui]
     [flames.core :as flames]
+    [malli.core :as m]
     [seesaw.core :as ss])
   (:import
     (java.util
@@ -466,6 +468,8 @@
 
   (init
     [this]
+    (specs/validate! "SolverRunConfig" specs/SolverRunConfig run-config)
+    (specs/validate! "SolverRunArgs" specs/SolverRunArgs run-args)
     (let [{:keys [iters initial-phenos initial-muts use-gui?]} run-config
           start    (print-and-save-start-time iters initial-phenos)
           init-pop (ga/initialize
@@ -476,7 +480,8 @@
 
       (log/info "Running with logging every n steps: " (:log-steps run-config))
 
-      (assoc this :ga-result init-pop
+      (assoc this
+             :ga-result init-pop
              :iters-to-go iters
              :start-ms start)))
 
@@ -488,10 +493,13 @@
           iters-to-go (:iters-to-go this)]
       (binding [ops/*log-steps* log-steps]
         (if (zero? iters-to-go)
-          (assoc this :status :done :result {:iters-done       (- iters iters-to-go)
-                                             :final-population population
-                                             :next-step        :wait})
+          (assoc this
+                 :status :done
+                 :result {:iters-done       (- iters iters-to-go)
+                          :final-population population
+                          :next-step        :wait})
           (let [{scores :pop-scores :as ga-result} (ga/evolve population)]
+            (specs/validate! "GAPopulation" specs/GAPopulationPhenotypes (:pop ga-result))
             (ops/report-iteration iters-to-go iters ga-result run-args run-config)
             (assoc this :ga-result ga-result :iters-to-go (next-iters iters-to-go scores)))))))
 
@@ -534,11 +542,10 @@
     return-value))
 
 
-(defn- run-ga-iterations-using-record
+(defn run-ga-iterations-using-record
   "Run GA evolution iterations on initial population"
-  [{:keys [iters initial-phenos initial-muts use-gui?] :as run-config}
-   run-args]
-
+  {:malli/schema [:=> [:cat #'specs/SolverRunConfig #'specs/SolverRunArgs] #'specs/SolverRunResults]}
+  [run-config run-args]
   (loop [solver-state (init (map->SolverStateController {:run-config run-config :run-args run-args}))]
     (let [[recur? next-solver-state] (run-iteration solver-state)]
       (if recur?
@@ -700,8 +707,22 @@
     (System/exit 0)))
 
 
+(def ^:private CLIArgs
+  [:map
+   {:closed true}
+   [:log-level {:optional true} keyword?]
+   [:iterations any?]
+   [:population any?]
+   [:headless any?]
+   [:xs {:optional true} any?]
+   [:ys {:optional true} any?]
+   [:use-flamechart {:optional true} any?]
+   [:max-leafs {:optional true} any?]])
+
+
 (defn run-app-from-cli-args
   "Run app from CLI args"
+  {:malli/schema [:=> [:cat #'CLIArgs] #'specs/SolverRunResults]}
   [{:keys [iterations population headless xs ys use-flamechart max-leafs] :as cli-opts}]
   (log/info "CLI: run from options: " cli-opts)
   (let [run-config {:initial-phenos (ops-init/initial-phenotypes population)
@@ -722,6 +743,11 @@
     result))
 
 
+;; todo: feel kind of hacky to have to call this in ?every? ns that has defn schema
+(specs/instrument-all!)
+
+
+(comment (println "FN SCHEMAS: " (m/function-schemas)))
 (comment (macroexpand-1 `(log/info "Hello")))
 (comment (log/info "Hello"))
 (comment (run-app-without-gui))