Spaces:
Running
Running
File size: 44,103 Bytes
e9e13d8 3c962e6 e9e13d8 48feff6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 215d7d0 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 48feff6 e9e13d8 e485eac 3c962e6 e485eac e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 662415d e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 48feff6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 662415d e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 48feff6 e9e13d8 e485eac e9e13d8 e485eac e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac 3c962e6 e485eac 3c962e6 e485eac 3c962e6 e485eac 3c962e6 e485eac 3c962e6 e485eac 3c962e6 e485eac e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac 48feff6 e9e13d8 3c962e6 e9e13d8 e485eac 3c962e6 e9e13d8 3c962e6 e9e13d8 c3b8d53 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 e485eac 3c962e6 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 48feff6 e9e13d8 e485eac e9e13d8 ccf4a01 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 e485eac e9e13d8 48feff6 e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 48feff6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 48feff6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 48feff6 e9e13d8 e485eac e9e13d8 e485eac e9e13d8 3c962e6 e9e13d8 662415d e9e13d8 3c962e6 e485eac e9e13d8 e485eac 3c962e6 e485eac 3c962e6 e485eac e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 3c962e6 e9e13d8 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 |
# /// script
# requires-python = ">=3.12"
# dependencies = [
# "marimo",
# ]
# ///
import marimo
__generated_with = "0.12.9"
app = marimo.App(app_title="Applicative programming with effects")
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
# Applicative programming with effects
`Applicative Functor` encapsulates certain sorts of *effectful* computations in a functionally pure way, and encourages an *applicative* programming style.
Applicative is a functor with application, providing operations to
+ embed pure expressions (`pure`), and
+ sequence computations and combine their results (`apply`).
In this notebook, you will learn:
1. How to view `Applicative` as multi-functor intuitively.
2. How to use `lift` to simplify chaining application.
3. How to bring *effects* to the functional pure world.
4. How to view `Applicative` as a lax monoidal functor.
5. How to use `Alternative` to amalgamate multiple computations into a single computation.
/// details | Notebook metadata
type: info
version: 0.1.3 | last modified: 2025-04-16 | author: [mΓ©taboulie](https://github.com/metaboulie)<br/>
reviewer: [Haleshot](https://github.com/Haleshot)
///
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
# The intuition: [Multifunctor](https://arxiv.org/pdf/2401.14286)
## Limitations of functor
Recall that functors abstract the idea of mapping a function over each element of a structure.
Suppose now that we wish to generalise this idea to allow functions with any number of arguments to be mapped, rather than being restricted to functions with a single argument. More precisely, suppose that we wish to define a hierarchy of `fmap` functions with the following types:
```haskell
fmap0 :: a -> f a
fmap1 :: (a -> b) -> f a -> f b
fmap2 :: (a -> b -> c) -> f a -> f b -> f c
fmap3 :: (a -> b -> c -> d) -> f a -> f b -> f c -> f d
```
And we have to declare a special version of the functor class for each case.
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Defining Multifunctor
/// admonition
we use prefix `f` rather than `ap` to indicate *Applicative Functor*
///
As a result, we may want to define a single `Multifunctor` such that:
1. Lift a regular n-argument function into the context of functors
```python
# lift a regular 3-argument function `g`
g: Callable[[A, B, C], D]
# into the context of functors
fg: Callable[[Functor[A], Functor[B], Functor[C]], Functor[D]]
```
3. Apply it to n functor-wrapped values
```python
# fa: Functor[A], fb: Functor[B], fc: Functor[C]
fg(fa, fb, fc)
```
5. Get a single functor-wrapped result
```python
fd: Functor[D]
```
We will define a function `lift` such that
```python
fd = lift(g, fa, fb, fc)
```
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Pure, apply and lift
Traditionally, applicative functors are presented through two core operations:
1. `pure`: embeds an object (value or function) into the applicative functor
```python
# a -> F a
pure: Callable[[A], Applicative[A]]
# for example, if `a` is
a: A
# then we can have `fa` as
fa: Applicative[A] = pure(a)
# or if we have a regular function `g`
g: Callable[[A], B]
# then we can have `fg` as
fg: Applicative[Callable[[A], B]] = pure(g)
```
2. `apply`: applies a function inside an applicative functor to a value inside an applicative functor
```python
# F (a -> b) -> F a -> F b
apply: Callable[[Applicative[Callable[[A], B]], Applicative[A]], Applicative[B]]
# and we can have
fd = apply(apply(apply(fg, fa), fb), fc)
```
As a result,
```python
lift(g, fa, fb, fc) = apply(apply(apply(pure(g), fa), fb), fc)
```
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
/// admonition | How to use *Applicative* in the manner of *Multifunctor*
1. Define `pure` and `apply` for an `Applicative` subclass
- We can define them much easier compared with `lift`.
2. Use the `lift` method
- We can use it much more convenient compared with the combination of `pure` and `apply`.
///
/// attention | You can suppress the chaining application of `apply` and `pure` as:
```python
apply(pure(g), fa) -> lift(g, fa)
apply(apply(pure(g), fa), fb) -> lift(g, fa, fb)
apply(apply(apply(pure(g), fa), fb), fc) -> lift(g, fa, fb, fc)
```
///
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Abstracting applicatives
We can now provide an initial abstraction definition of applicatives:
```python
@dataclass
class Applicative[A](Functor, ABC):
@classmethod
@abstractmethod
def pure(cls, a: A) -> "Applicative[A]":
raise NotImplementedError("Subclasses must implement pure")
@classmethod
@abstractmethod
def apply(
cls, fg: "Applicative[Callable[[A], B]]", fa: "Applicative[A]"
) -> "Applicative[B]":
raise NotImplementedError("Subclasses must implement apply")
@classmethod
def lift(cls, f: Callable, *args: "Applicative") -> "Applicative":
curr = cls.pure(f)
if not args:
return curr
for arg in args:
curr = cls.apply(curr, arg)
return curr
```
/// attention | minimal implementation requirement
- `pure`
- `apply`
///
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""# Instances, laws and utility functions""")
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Applicative instances
When we are actually implementing an *Applicative* instance, we can keep in mind that `pure` and `apply` fundamentally:
- embed an object (value or function) to the computational context
- apply a function inside the computation context to a value inside the computational context
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
### The Wrapper Applicative
- `pure` should simply *wrap* an object, in the sense that:
```haskell
Wrapper.pure(1) => Wrapper(value=1)
```
- `apply` should apply a *wrapped* function to a *wrapped* value
The implementation is:
"""
)
@app.cell
def _(Applicative, dataclass):
@dataclass
class Wrapper[A](Applicative):
value: A
@classmethod
def pure(cls, a: A) -> "Wrapper[A]":
return cls(a)
@classmethod
def apply(
cls, fg: "Wrapper[Callable[[A], B]]", fa: "Wrapper[A]"
) -> "Wrapper[B]":
return cls(fg.value(fa.value))
return (Wrapper,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""> try with Wrapper below""")
@app.cell
def _(Wrapper) -> None:
Wrapper.lift(
lambda a: lambda b: lambda c: a + b * c,
Wrapper(1),
Wrapper(2),
Wrapper(3),
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
### The List Applicative
- `pure` should wrap the object in a list, in the sense that:
```haskell
List.pure(1) => List(value=[1])
```
- `apply` should apply a list of functions to a list of values
- you can think of this as cartesian product, concatenating the result of applying every function to every value
The implementation is:
"""
)
@app.cell
def _(Applicative, dataclass, product):
@dataclass
class List[A](Applicative):
value: list[A]
@classmethod
def pure(cls, a: A) -> "List[A]":
return cls([a])
@classmethod
def apply(cls, fg: "List[Callable[[A], B]]", fa: "List[A]") -> "List[B]":
return cls([g(a) for g, a in product(fg.value, fa.value)])
return (List,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""> try with List below""")
@app.cell
def _(List) -> None:
List.apply(
List([lambda a: a + 1, lambda a: a * 2]),
List([1, 2]),
)
@app.cell
def _(List) -> None:
List.lift(lambda a: lambda b: a + b, List([1, 2]), List([3, 4, 5]))
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
### The Maybe Applicative
- `pure` should wrap the object in a Maybe, in the sense that:
```haskell
Maybe.pure(1) => "Just 1"
Maybe.pure(None) => "Nothing"
```
- `apply` should apply a function maybe exist to a value maybe exist
- if the function is `None` or the value is `None`, simply returns `None`
- else apply the function to the value and wrap the result in `Just`
The implementation is:
"""
)
@app.cell
def _(Applicative, dataclass):
@dataclass
class Maybe[A](Applicative):
value: None | A
@classmethod
def pure(cls, a: A) -> "Maybe[A]":
return cls(a)
@classmethod
def apply(
cls, fg: "Maybe[Callable[[A], B]]", fa: "Maybe[A]"
) -> "Maybe[B]":
if fg.value is None or fa.value is None:
return cls(None)
return cls(fg.value(fa.value))
def __repr__(self):
return "Nothing" if self.value is None else f"Just({self.value!r})"
return (Maybe,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""> try with Maybe below""")
@app.cell
def _(Maybe) -> None:
Maybe.lift(
lambda a: lambda b: a + b,
Maybe(1),
Maybe(2),
)
@app.cell
def _(Maybe) -> None:
Maybe.lift(
lambda a: lambda b: None,
Maybe(1),
Maybe(2),
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
### The Either Applicative
- `pure` should wrap the object in `Right`, in the sense that:
```haskell
Either.pure(1) => Right(1)
```
- `apply` should apply a function that is either on Left or Right to a value that is either on Left or Right
- if the function is `Left`, simply returns the `Left` of the function
- else `fmap` the `Right` of the function to the value
The implementation is:
"""
)
@app.cell
def _(Applicative, B, Callable, Union, dataclass):
@dataclass
class Either[A](Applicative):
left: A = None
right: A = None
def __post_init__(self):
if (self.left is not None and self.right is not None) or (
self.left is None and self.right is None
):
msg = "Provide either the value of the left or the value of the right."
raise TypeError(
msg
)
@classmethod
def pure(cls, a: A) -> "Either[A]":
return cls(right=a)
@classmethod
def apply(
cls, fg: "Either[Callable[[A], B]]", fa: "Either[A]"
) -> "Either[B]":
if fg.left is not None:
return cls(left=fg.left)
return cls.fmap(fg.right, fa)
@classmethod
def fmap(
cls, g: Callable[[A], B], fa: "Either[A]"
) -> Union["Either[A]", "Either[B]"]:
if fa.left is not None:
return cls(left=fa.left)
return cls(right=g(fa.right))
def __repr__(self):
if self.left is not None:
return f"Left({self.left!r})"
return f"Right({self.right!r})"
return (Either,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""> try with `Either` below""")
@app.cell
def _(Either) -> None:
Either.apply(Either(left=TypeError("Parse Error")), Either(right=2))
@app.cell
def _(Either) -> None:
Either.apply(
Either(right=lambda x: x + 1), Either(left=TypeError("Parse Error"))
)
@app.cell
def _(Either) -> None:
Either.apply(Either(right=lambda x: x + 1), Either(right=1))
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Collect the list of response with sequenceL
One often wants to execute a list of commands and collect the list of their response, and we can define a function `sequenceL` for this
/// admonition
In a further notebook about `Traversable`, we will have a more generic `sequence` that execute a **sequence** of commands and collect the **sequence** of their response, which is not limited to `list`.
///
```python
@classmethod
def sequenceL(cls, fas: list["Applicative[A]"]) -> "Applicative[list[A]]":
if not fas:
return cls.pure([])
return cls.apply(
cls.fmap(lambda v: lambda vs: [v] + vs, fas[0]),
cls.sequenceL(fas[1:]),
)
```
Let's try `sequenceL` with the instances.
"""
)
@app.cell
def _(Wrapper) -> None:
Wrapper.sequenceL([Wrapper(1), Wrapper(2), Wrapper(3)])
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
/// attention
For the `Maybe` Applicative, the presence of any `Nothing` causes the entire computation to return Nothing.
///
"""
)
@app.cell
def _(Maybe) -> None:
Maybe.sequenceL([Maybe(1), Maybe(2), Maybe(None), Maybe(3)])
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""The result of `sequenceL` for `List Applicative` is the Cartesian product of the input lists, yielding all possible ordered combinations of elements from each list.""")
@app.cell
def _(List) -> None:
List.sequenceL([List([1, 2]), List([3]), List([5, 6, 7])])
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Applicative laws
/// admonition | id and compose
Remember that
- `id = lambda x: x`
- `compose = lambda f: lambda g: lambda x: f(g(x))`
///
Traditionally, there are four laws that `Applicative` instances should satisfy. In some sense, they are all concerned with making sure that `pure` deserves its name:
- The identity law:
```python
# fa: Applicative[A]
apply(pure(id), fa) = fa
```
- Homomorphism:
```python
# a: A
# g: Callable[[A], B]
apply(pure(g), pure(a)) = pure(g(a))
```
Intuitively, applying a non-effectful function to a non-effectful argument in an effectful context is the same as just applying the function to the argument and then injecting the result into the context with pure.
- Interchange:
```python
# a: A
# fg: Applicative[Callable[[A], B]]
apply(fg, pure(a)) = apply(pure(lambda g: g(a)), fg)
```
Intuitively, this says that when evaluating the application of an effectful function to a pure argument, the order in which we evaluate the function and its argument doesn't matter.
- Composition:
```python
# fg: Applicative[Callable[[B], C]]
# fh: Applicative[Callable[[A], B]]
# fa: Applicative[A]
apply(fg, apply(fh, fa)) = lift(compose, fg, fh, fa)
```
This one is the trickiest law to gain intuition for. In some sense it is expressing a sort of associativity property of `apply`.
We can add 4 helper functions to `Applicative` to check whether an instance respects the laws or not:
```python
@dataclass
class Applicative[A](Functor, ABC):
@classmethod
def check_identity(cls, fa: "Applicative[A]"):
if cls.lift(id, fa) != fa:
raise ValueError("Instance violates identity law")
return True
@classmethod
def check_homomorphism(cls, a: A, f: Callable[[A], B]):
if cls.lift(f, cls.pure(a)) != cls.pure(f(a)):
raise ValueError("Instance violates homomorphism law")
return True
@classmethod
def check_interchange(cls, a: A, fg: "Applicative[Callable[[A], B]]"):
if cls.apply(fg, cls.pure(a)) != cls.lift(lambda g: g(a), fg):
raise ValueError("Instance violates interchange law")
return True
@classmethod
def check_composition(
cls,
fg: "Applicative[Callable[[B], C]]",
fh: "Applicative[Callable[[A], B]]",
fa: "Applicative[A]",
):
if cls.apply(fg, cls.apply(fh, fa)) != cls.lift(compose, fg, fh, fa):
raise ValueError("Instance violates composition law")
return True
```
> Try to validate applicative laws below
"""
)
@app.cell
def _():
id = lambda x: x
compose = lambda f: lambda g: lambda x: f(g(x))
const = lambda a: lambda _: a
return compose, const, id
@app.cell
def _(List, Wrapper) -> None:
print("Checking Wrapper")
print(Wrapper.check_identity(Wrapper.pure(1)))
print(Wrapper.check_homomorphism(1, lambda x: x + 1))
print(Wrapper.check_interchange(1, Wrapper.pure(lambda x: x + 1)))
print(
Wrapper.check_composition(
Wrapper.pure(lambda x: x * 2),
Wrapper.pure(lambda x: x + 0.1),
Wrapper.pure(1),
)
)
print("\nChecking List")
print(List.check_identity(List.pure(1)))
print(List.check_homomorphism(1, lambda x: x + 1))
print(List.check_interchange(1, List.pure(lambda x: x + 1)))
print(
List.check_composition(
List.pure(lambda x: x * 2), List.pure(lambda x: x + 0.1), List.pure(1)
)
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Utility functions
/// attention | using `fmap`
`fmap` is defined automatically using `pure` and `apply`, so you can use `fmap` with any `Applicative`
///
```python
@dataclass
class Applicative[A](Functor, ABC):
@classmethod
def skip(
cls, fa: "Applicative[A]", fb: "Applicative[B]"
) -> "Applicative[B]":
'''
Sequences the effects of two Applicative computations,
but discards the result of the first.
'''
return cls.apply(cls.const(fa, id), fb)
@classmethod
def keep(
cls, fa: "Applicative[A]", fb: "Applicative[B]"
) -> "Applicative[B]":
'''
Sequences the effects of two Applicative computations,
but discard the result of the second.
'''
return cls.lift(const, fa, fb)
@classmethod
def revapp(
cls, fa: "Applicative[A]", fg: "Applicative[Callable[[A], [B]]]"
) -> "Applicative[B]":
'''
The first computation produces values which are provided
as input to the function(s) produced by the second computation.
'''
return cls.lift(lambda a: lambda f: f(a), fa, fg)
```
- `skip` sequences the effects of two Applicative computations, but **discards the result of the first**. For example, if `m1` and `m2` are instances of type `Maybe[Int]`, then `Maybe.skip(m1, m2)` is `Nothing` whenever either `m1` or `m2` is `Nothing`; but if not, it will have the same value as `m2`.
- Likewise, `keep` sequences the effects of two computations, but **keeps only the result of the first**.
- `revapp` is similar to `apply`, but where the first computation produces value(s) which are provided as input to the function(s) produced by the second computation.
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
/// admonition | Exercise
Try to use utility functions with different instances
///
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
# Formal implementation of Applicative
Now, we can give the formal implementation of `Applicative`
"""
)
@app.cell
def _(
ABC,
B,
Callable,
Functor,
abstractmethod,
compose,
const,
dataclass,
id,
):
@dataclass
class Applicative[A](Functor, ABC):
@classmethod
@abstractmethod
def pure(cls, a: A) -> "Applicative[A]":
"""Lift a value into the Structure."""
msg = "Subclasses must implement pure"
raise NotImplementedError(msg)
@classmethod
@abstractmethod
def apply(
cls, fg: "Applicative[Callable[[A], B]]", fa: "Applicative[A]"
) -> "Applicative[B]":
"""Sequential application."""
msg = "Subclasses must implement apply"
raise NotImplementedError(msg)
@classmethod
def lift(cls, f: Callable, *args: "Applicative") -> "Applicative":
"""Lift a function of arbitrary arity to work with values in applicative context."""
curr = cls.pure(f)
if not args:
return curr
for arg in args:
curr = cls.apply(curr, arg)
return curr
@classmethod
def fmap(
cls, f: Callable[[A], B], fa: "Applicative[A]"
) -> "Applicative[B]":
return cls.lift(f, fa)
@classmethod
def sequenceL(cls, fas: list["Applicative[A]"]) -> "Applicative[list[A]]":
"""
Execute a list of commands and collect the list of their response.
"""
if not fas:
return cls.pure([])
return cls.apply(
cls.fmap(lambda v: lambda vs: [v, *vs], fas[0]),
cls.sequenceL(fas[1:]),
)
@classmethod
def skip(
cls, fa: "Applicative[A]", fb: "Applicative[B]"
) -> "Applicative[B]":
"""
Sequences the effects of two Applicative computations,
but discards the result of the first.
"""
return cls.apply(cls.const(fa, id), fb)
@classmethod
def keep(
cls, fa: "Applicative[A]", fb: "Applicative[B]"
) -> "Applicative[B]":
"""
Sequences the effects of two Applicative computations,
but discard the result of the second.
"""
return cls.lift(const, fa, fb)
@classmethod
def revapp(
cls, fa: "Applicative[A]", fg: "Applicative[Callable[[A], [B]]]"
) -> "Applicative[B]":
"""
The first computation produces values which are provided
as input to the function(s) produced by the second computation.
"""
return cls.lift(lambda a: lambda f: f(a), fa, fg)
@classmethod
def check_identity(cls, fa: "Applicative[A]") -> bool:
if cls.lift(id, fa) != fa:
msg = "Instance violates identity law"
raise ValueError(msg)
return True
@classmethod
def check_homomorphism(cls, a: A, f: Callable[[A], B]) -> bool:
if cls.lift(f, cls.pure(a)) != cls.pure(f(a)):
msg = "Instance violates homomorphism law"
raise ValueError(msg)
return True
@classmethod
def check_interchange(cls, a: A, fg: "Applicative[Callable[[A], B]]") -> bool:
if cls.apply(fg, cls.pure(a)) != cls.lift(lambda g: g(a), fg):
msg = "Instance violates interchange law"
raise ValueError(msg)
return True
@classmethod
def check_composition(
cls,
fg: "Applicative[Callable[[B], C]]",
fh: "Applicative[Callable[[A], B]]",
fa: "Applicative[A]",
) -> bool:
if cls.apply(fg, cls.apply(fh, fa)) != cls.lift(compose, fg, fh, fa):
msg = "Instance violates composition law"
raise ValueError(msg)
return True
return (Applicative,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
# Effectful programming
Our original motivation for applicatives was the desire to generalise the idea of mapping to functions with multiple arguments. This is a valid interpretation of the concept of applicatives, but from the three instances we have seen it becomes clear that there is also another, more abstract view.
The arguments are no longer just plain values but may also have effects, such as the possibility of failure, having many ways to succeed, or performing input/output actions. In this manner, applicative functors can also be viewed as abstracting the idea of **applying pure functions to effectful arguments**, with the precise form of effects that are permitted depending on the nature of the underlying functor.
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## The IO Applicative
We will try to define an `IO` applicative here.
As before, we first abstract how `pure` and `apply` should function.
- `pure` should wrap the object in an IO action, and make the object *callable* if it's not because we want to perform the action later:
```haskell
IO.pure(1) => IO(effect=lambda: 1)
IO.pure(f) => IO(effect=f)
```
- `apply` should perform an action that produces a value, then apply the function with the value
The implementation is:
"""
)
@app.cell
def _(Applicative, Callable, dataclass):
@dataclass
class IO(Applicative):
effect: Callable
def __call__(self):
return self.effect()
@classmethod
def pure(cls, a):
return cls(a) if isinstance(a, Callable) else IO(lambda: a)
@classmethod
def apply(cls, fg, fa):
return cls.pure(fg.effect(fa.effect()))
return (IO,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""For example, a function that reads a given number of lines from the keyboard can be defined in applicative style as follows:""")
@app.cell
def _(IO):
def get_chars(n: int = 3):
return IO.sequenceL([
IO.pure(input(f"input the {i}th str")) for i in range(1, n + 1)
])
return (get_chars,)
@app.cell
def _() -> None:
# get_chars()()
return
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""# From the perspective of category theory""")
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Lax Monoidal Functor
An alternative, equivalent formulation of `Applicative` is given by
"""
)
@app.cell
def _(ABC, Functor, abstractmethod, dataclass):
@dataclass
class Monoidal[A](Functor, ABC):
@classmethod
@abstractmethod
def unit(cls) -> "Monoidal[Tuple[()]]":
pass
@classmethod
@abstractmethod
def tensor(
cls, this: "Monoidal[A]", other: "Monoidal[B]"
) -> "Monoidal[Tuple[A, B]]":
pass
return (Monoidal,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
Intuitively, this states that a *monoidal functor* is one which has some sort of "default shape" and which supports some sort of "combining" operation.
- `unit` provides the identity element
- `tensor` combines two contexts into a product context
More technically, the idea is that `monoidal functor` preserves the "monoidal structure" given by the pairing constructor `(,)` and unit type `()`.
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
Furthermore, to deserve the name "monoidal", instances of Monoidal ought to satisfy the following laws, which seem much more straightforward than the traditional Applicative laws:
- Left identity
`tensor(unit, v) β
v`
- Right identity
`tensor(u, unit) β
u`
- Associativity
`tensor(u, tensor(v, w)) β
tensor(tensor(u, v), w)`
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
/// admonition | β
indicates isomorphism
`β
` refers to *isomorphism* rather than equality.
In particular we consider `(x, ()) β
x β
((), x)` and `((x, y), z) β
(x, (y, z))`
///
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Mutual definability of Monoidal and Applicative
We can implement `pure` and `apply` in terms of `unit` and `tensor`, and vice versa.
```python
pure(a) = fmap((lambda _: a), unit)
apply(fg, fa) = fmap((lambda pair: pair[0](pair[1])), tensor(fg, fa))
```
```python
unit() = pure(())
tensor(fa, fb) = lift(lambda fa: lambda fb: (fa, fb), fa, fb)
```
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Instance: ListMonoidal
- `unit` should simply return a empty tuple wrapper in a list
```haskell
ListMonoidal.unit() => [()]
```
- `tensor` should return the *cartesian product* of the items of 2 ListMonoidal instances
The implementation is:
"""
)
@app.cell
def _(B, Callable, Monoidal, dataclass, product):
@dataclass
class ListMonoidal[A](Monoidal):
items: list[A]
@classmethod
def unit(cls) -> "ListMonoidal[Tuple[()]]":
return cls([()])
@classmethod
def tensor(
cls, this: "ListMonoidal[A]", other: "ListMonoidal[B]"
) -> "ListMonoidal[Tuple[A, B]]":
return cls(list(product(this.items, other.items)))
@classmethod
def fmap(
cls, f: Callable[[A], B], ma: "ListMonoidal[A]"
) -> "ListMonoidal[B]":
return cls([f(a) for a in ma.items])
return (ListMonoidal,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""> try with `ListMonoidal` below""")
@app.cell
def _(ListMonoidal):
xs = ListMonoidal([1, 2])
ys = ListMonoidal(["a", "b"])
ListMonoidal.tensor(xs, ys)
return xs, ys
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""and we can prove that `tensor(fa, fb) = lift(lambda fa: lambda fb: (fa, fb), fa, fb)`:""")
@app.cell
def _(List, xs, ys) -> None:
List.lift(lambda fa: lambda fb: (fa, fb), List(xs.items), List(ys.items))
@app.cell(hide_code=True)
def _(ABC, B, Callable, abstractmethod, dataclass):
@dataclass
class Functor[A](ABC):
@classmethod
@abstractmethod
def fmap(cls, f: Callable[[A], B], a: "Functor[A]") -> "Functor[B]":
msg = "Subclasses must implement fmap"
raise NotImplementedError(msg)
@classmethod
def const(cls, a: "Functor[A]", b: B) -> "Functor[B]":
return cls.fmap(lambda _: b, a)
@classmethod
def void(cls, a: "Functor[A]") -> "Functor[None]":
return cls.const_fmap(a, None)
return (Functor,)
@app.cell(hide_code=True)
def _():
import marimo as mo
return (mo,)
@app.cell(hide_code=True)
def _():
from abc import ABC, abstractmethod
from collections.abc import Callable
from dataclasses import dataclass
from typing import TypeVar, Union
return ABC, Callable, TypeVar, Union, abstractmethod, dataclass
@app.cell(hide_code=True)
def _():
from itertools import product
return (product,)
@app.cell(hide_code=True)
def _(TypeVar):
A = TypeVar("A")
B = TypeVar("B")
C = TypeVar("C")
return A, B, C
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
# From Applicative to Alternative
## Abstracting Alternative
In our studies so far, we saw that both `Maybe` and `List` can represent computations with a varying number of results.
We use `Maybe` to indicate a computation can fail somehow and `List` for computations that can have many possible results. In both of these cases, one useful operation is amalgamating all possible results from multiple computations into a single computation.
`Alternative` formalizes computations that support:
- **Failure** (empty result)
- **Choice** (combination of results)
- **Repetition** (multiple results)
It extends `Applicative` with monoidal structure, where:
```python
@dataclass
class Alternative[A](Applicative, ABC):
@classmethod
@abstractmethod
def empty(cls) -> "Alternative[A]":
'''Identity element for alternative computations'''
@classmethod
@abstractmethod
def alt(
cls, fa: "Alternative[A]", fb: "Alternative[A]"
) -> "Alternative[A]":
'''Binary operation combining computations'''
```
- `empty` is the identity element (e.g., `Maybe(None)`, `List([])`)
- `alt` is a combination operator (e.g., `Maybe` fallback, list concatenation)
`empty` and `alt` should satisfy the following **laws**:
```python
# Left identity
alt(empty, fa) == fa
# Right identity
alt(fa, empty) == fa
# Associativity
alt(fa, alt(fb, fc)) == alt(alt(fa, fb), fc)
```
/// admonition
Actually, `Alternative` is a *monoid* on `Applicative Functors`. We will talk about *monoid* and review these laws in the next notebook about `Monads`.
///
/// attention | minimal implementation requirement
- `empty`
- `alt`
///
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## Instances of Alternative
### The Maybe Alternative
- `empty`: the identity element of `Maybe` is `Maybe(None)`
- `alt`: return the first element if it's not `None`, else return the second element
"""
)
@app.cell
def _(Alternative, Maybe, dataclass):
@dataclass
class AltMaybe[A](Maybe, Alternative):
@classmethod
def empty(cls) -> "AltMaybe[A]":
return cls(None)
@classmethod
def alt(cls, fa: "AltMaybe[A]", fb: "AltMaybe[A]") -> "AltMaybe[A]":
if fa.value is not None:
return cls(fa.value)
return cls(fb.value)
def __repr__(self):
return "Nothing" if self.value is None else f"Just({self.value!r})"
return (AltMaybe,)
@app.cell
def _(AltMaybe) -> None:
print(AltMaybe.empty())
print(AltMaybe.alt(AltMaybe(None), AltMaybe(1)))
print(AltMaybe.alt(AltMaybe(None), AltMaybe(None)))
print(AltMaybe.alt(AltMaybe(1), AltMaybe(None)))
print(AltMaybe.alt(AltMaybe(1), AltMaybe(2)))
@app.cell
def _(AltMaybe) -> None:
print(AltMaybe.check_left_identity(AltMaybe(1)))
print(AltMaybe.check_right_identity(AltMaybe(1)))
print(AltMaybe.check_associativity(AltMaybe(1), AltMaybe(2), AltMaybe(None)))
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
### The List Alternative
- `empty`: the identity element of `List` is `List([])`
- `alt`: return the concatenation of 2 input lists
"""
)
@app.cell
def _(Alternative, List, dataclass):
@dataclass
class AltList[A](List, Alternative):
@classmethod
def empty(cls) -> "AltList[A]":
return cls([])
@classmethod
def alt(cls, fa: "AltList[A]", fb: "AltList[A]") -> "AltList[A]":
return cls(fa.value + fb.value)
return (AltList,)
@app.cell
def _(AltList) -> None:
print(AltList.empty())
print(AltList.alt(AltList([1, 2, 3]), AltList([4, 5])))
@app.cell
def _(AltList) -> None:
AltList([1])
@app.cell
def _(AltList) -> None:
AltList([1])
@app.cell
def _(AltList) -> None:
print(AltList.check_left_identity(AltList([1, 2, 3])))
print(AltList.check_right_identity(AltList([1, 2, 3])))
print(
AltList.check_associativity(
AltList([1, 2]), AltList([3, 4, 5]), AltList([6])
)
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
## some and many
/// admonition | This section mainly refers to
- https://stackoverflow.com/questions/7671009/some-and-many-functions-from-the-alternative-type-class/7681283#7681283
///
First let's have a look at the implementation of `some` and `many`:
```python
@classmethod
def some(cls, fa: "Alternative[A]") -> "Alternative[list[A]]":
# Short-circuit if input is empty
if fa == cls.empty():
return cls.empty()
return cls.apply(
cls.fmap(lambda a: lambda b: [a] + b, fa), cls.many(fa)
)
@classmethod
def many(cls, fa: "Alternative[A]") -> "Alternative[list[A]]":
# Directly return empty list if input is empty
if fa == cls.empty():
return cls.pure([])
return cls.alt(cls.some(fa), cls.pure([]))
```
So `some f` runs `f` once, then *many* times, and conses the results. `many f` runs f *some* times, or *alternatively* just returns the empty list.
The idea is that they both run `f` as often as possible until it **fails**, collecting the results in a list. The difference is that `some f` immediately fails if `f` fails, while `many f` will still succeed and *return* the empty list in such a case. But what all this exactly means depends on how `alt` is defined.
Let's see what it does for the instances `AltMaybe` and `AltList`.
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""For `AltMaybe`. `None` means failure, so some `None` fails as well and evaluates to `None` while many `None` succeeds and evaluates to `Just []`. Both `some (Just ())` and `many (Just ())` never return, because `Just ()` never fails.""")
@app.cell
def _(AltMaybe) -> None:
print(AltMaybe.some(AltMaybe.empty()))
print(AltMaybe.many(AltMaybe.empty()))
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""For `AltList`, `[]` means failure, so `some []` evaluates to `[]` (no answers) while `many []` evaluates to `[[]]` (there's one answer and it is the empty list). Again `some [()]` and `many [()]` don't return.""")
@app.cell
def _(AltList) -> None:
print(AltList.some(AltList.empty()))
print(AltList.many(AltList.empty()))
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(r"""## Formal implementation of Alternative""")
@app.cell
def _(ABC, Applicative, abstractmethod, dataclass):
@dataclass
class Alternative[A](Applicative, ABC):
"""A monoid on applicative functors."""
@classmethod
@abstractmethod
def empty(cls) -> "Alternative[A]":
msg = "Subclasses must implement empty"
raise NotImplementedError(msg)
@classmethod
@abstractmethod
def alt(
cls, fa: "Alternative[A]", fb: "Alternative[A]"
) -> "Alternative[A]":
msg = "Subclasses must implement alt"
raise NotImplementedError(msg)
@classmethod
def some(cls, fa: "Alternative[A]") -> "Alternative[list[A]]":
# Short-circuit if input is empty
if fa == cls.empty():
return cls.empty()
return cls.apply(
cls.fmap(lambda a: lambda b: [a, *b], fa), cls.many(fa)
)
@classmethod
def many(cls, fa: "Alternative[A]") -> "Alternative[list[A]]":
# Directly return empty list if input is empty
if fa == cls.empty():
return cls.pure([])
return cls.alt(cls.some(fa), cls.pure([]))
@classmethod
def check_left_identity(cls, fa: "Alternative[A]") -> bool:
return cls.alt(cls.empty(), fa) == fa
@classmethod
def check_right_identity(cls, fa: "Alternative[A]") -> bool:
return cls.alt(fa, cls.empty()) == fa
@classmethod
def check_associativity(
cls, fa: "Alternative[A]", fb: "Alternative[A]", fc: "Alternative[A]"
) -> bool:
return cls.alt(fa, cls.alt(fb, fc)) == cls.alt(cls.alt(fa, fb), fc)
return (Alternative,)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
/// admonition
We will explore more about `Alternative` in a future notebooks about [Monadic Parsing](https://www.cambridge.org/core/journals/journal-of-functional-programming/article/monadic-parsing-in-haskell/E557DFCCE00E0D4B6ED02F3FB0466093)
///
"""
)
@app.cell(hide_code=True)
def _(mo) -> None:
mo.md(
r"""
# Further reading
Notice that these reading sources are optional and non-trivial
- [Applicaive Programming with Effects](https://www.staff.city.ac.uk/~ross/papers/Applicative.html)
- [Equivalence of Applicative Functors and
Multifunctors](https://arxiv.org/pdf/2401.14286)
- [Applicative functor](https://wiki.haskell.org/index.php?title=Applicative_functor)
- [Control.Applicative](https://hackage.haskell.org/package/base-4.21.0.0/docs/Control-Applicative.html#t:Applicative)
- [Typeclassopedia#Applicative](https://wiki.haskell.org/index.php?title=Typeclassopedia#Applicative)
- [Notions of computation as monoids](https://www.cambridge.org/core/journals/journal-of-functional-programming/article/notions-of-computation-as-monoids/70019FC0F2384270E9F41B9719042528)
- [Free Applicative Functors](https://arxiv.org/abs/1403.0749)
- [The basics of applicative functors, put to practical work](http://www.serpentine.com/blog/2008/02/06/the-basics-of-applicative-functors-put-to-practical-work/)
- [Abstracting with Applicatives](http://comonad.com/reader/2012/abstracting-with-applicatives/)
- [Static analysis with Applicatives](https://gergo.erdi.hu/blog/2012-12-01-static_analysis_with_applicatives/)
- [Explaining Applicative functor in categorical terms - monoidal functors](https://cstheory.stackexchange.com/questions/12412/explaining-applicative-functor-in-categorical-terms-monoidal-functors)
- [Applicative, A Strong Lax Monoidal Functor](https://beuke.org/applicative/)
- [Applicative Functors](https://bartoszmilewski.com/2017/02/06/applicative-functors/)
"""
)
if __name__ == "__main__":
app.run()
|