• Title: Open Mapping Theorem

  • Series: Functional Analysis

  • YouTube-Title: Functional Analysis 26 | Open Mapping Theorem

  • Bright video: https://youtu.be/o0vAbQFZnIo

  • Dark video: https://youtu.be/4K_iatmYi4o

  • PDF: Download PDF version of the bright video

  • Dark-PDF: Download PDF version of the dark video

  • Print-PDF: Download printable PDF version

  • Thumbnail (bright): Download PNG

  • Thumbnail (dark): Download PNG

  • Subtitle on GitHub: fa26_sub_eng.srt

  • Timestamps

    00:00 Introduction

    01:15 General example

    02:47 Examples

    03:47 Theorem

  • Subtitle in English

    1 00:00:00,379 –> 00:00:02,240 Hello and welcome back to

    2 00:00:02,250 –> 00:00:03,619 functional analysis.

    3 00:00:03,779 –> 00:00:05,480 And as always, many thanks

    4 00:00:05,489 –> 00:00:06,760 to all the nice people that

    5 00:00:06,769 –> 00:00:08,279 support this channel on Steady

    6 00:00:08,289 –> 00:00:09,119 or paypal.

    7 00:00:09,630 –> 00:00:11,199 This is part 26.

    8 00:00:11,319 –> 00:00:12,420 And I still want to show

    9 00:00:12,430 –> 00:00:13,960 you more main results in

    10 00:00:13,970 –> 00:00:15,680 the topic of functional analysis.

    11 00:00:16,459 –> 00:00:17,790 Today, we will talk about

    12 00:00:17,799 –> 00:00:19,450 the so called open mapping

    13 00:00:19,459 –> 00:00:21,290 theorem, also known as the

    14 00:00:21,299 –> 00:00:22,709 Banach Schauder theorem.

    15 00:00:23,469 –> 00:00:25,010 As the name suggests, the

    16 00:00:25,020 –> 00:00:26,620 theorem is about open

    17 00:00:26,629 –> 00:00:27,290 maps.

    18 00:00:27,299 –> 00:00:28,530 Therefore, I first want to

    19 00:00:28,540 –> 00:00:30,170 start explaining what an

    20 00:00:30,180 –> 00:00:31,920 open map is the

    21 00:00:31,930 –> 00:00:33,549 notion openness we define

    22 00:00:33,560 –> 00:00:34,860 for metric spaces.

    23 00:00:34,869 –> 00:00:36,360 So therefore, let’s choose

    24 00:00:36,369 –> 00:00:38,099 two metric spaces X and

    25 00:00:38,110 –> 00:00:38,520 Y.

    26 00:00:39,299 –> 00:00:40,549 Of course, now we want to

    27 00:00:40,560 –> 00:00:42,430 consider a map from X

    28 00:00:42,439 –> 00:00:43,020 to Y.

    29 00:00:43,849 –> 00:00:45,630 Now such a map F gets the

    30 00:00:45,639 –> 00:00:47,540 name open if it sends

    31 00:00:47,549 –> 00:00:49,290 open sets in X to

    32 00:00:49,299 –> 00:00:50,509 open sets in Y.

    33 00:00:51,520 –> 00:00:53,369 In other words, if we take

    34 00:00:53,380 –> 00:00:55,290 any open set A in

    35 00:00:55,299 –> 00:00:56,959 X, then the

    36 00:00:56,970 –> 00:00:58,759 image F of A should

    37 00:00:58,770 –> 00:01:00,360 also be an open set.

    38 00:01:00,919 –> 00:01:02,599 However, please don’t forget

    39 00:01:02,610 –> 00:01:04,540 the notion openness is always

    40 00:01:04,550 –> 00:01:05,919 given with respect to the

    41 00:01:05,930 –> 00:01:07,480 corresponding metric space.

    42 00:01:08,169 –> 00:01:08,589 OK.

    43 00:01:08,599 –> 00:01:10,220 So this looks like a natural

    44 00:01:10,230 –> 00:01:10,819 notion.

    45 00:01:10,949 –> 00:01:12,290 Therefore, let’s immediately

    46 00:01:12,300 –> 00:01:13,669 look at some examples.

    47 00:01:14,720 –> 00:01:16,660 First, a very general example

    48 00:01:16,669 –> 00:01:18,120 because you might already

    49 00:01:18,129 –> 00:01:19,699 recognize the connection

    50 00:01:19,709 –> 00:01:21,400 to the definition of continuity

    51 00:01:21,410 –> 00:01:21,720 here.

    52 00:01:22,500 –> 00:01:23,860 So let’s fix a bijective

    53 00:01:23,870 –> 00:01:25,370 map, which means we have

    54 00:01:25,379 –> 00:01:27,120 an inverse map F to the

    55 00:01:27,129 –> 00:01:28,410 power minus one.

    56 00:01:28,940 –> 00:01:30,449 Of course, this one now goes

    57 00:01:30,459 –> 00:01:32,220 from Y to X and we

    58 00:01:32,230 –> 00:01:34,110 also choose it to be a continuous

    59 00:01:34,120 –> 00:01:34,519 map.

    60 00:01:35,300 –> 00:01:36,970 Having these two assumptions.

    61 00:01:36,980 –> 00:01:38,830 The map F from X to Y

    62 00:01:38,839 –> 00:01:40,610 is indeed an open map.

    63 00:01:41,440 –> 00:01:43,029 This is easy to see when

    64 00:01:43,040 –> 00:01:44,269 you look at the definition

    65 00:01:44,279 –> 00:01:45,930 of the continuity of the

    66 00:01:45,940 –> 00:01:46,709 inverse map.

    67 00:01:47,720 –> 00:01:49,069 It simply means when you

    68 00:01:49,080 –> 00:01:50,489 have an open set on the right

    69 00:01:50,500 –> 00:01:52,239 hand side, the pre image

    70 00:01:52,250 –> 00:01:53,779 on the left hand side is

    71 00:01:53,790 –> 00:01:55,180 also an open set.

    72 00:01:55,559 –> 00:01:56,860 Of course, everything looks

    73 00:01:56,870 –> 00:01:58,300 very similar here.

    74 00:01:58,309 –> 00:01:59,589 You see we have the inverse

    75 00:01:59,599 –> 00:02:01,519 map from Y to X.

    76 00:02:02,019 –> 00:02:03,800 Then we take a subset A in

    77 00:02:03,809 –> 00:02:05,419 the space X which is on the

    78 00:02:05,430 –> 00:02:06,739 right hand side of the inverse

    79 00:02:06,750 –> 00:02:07,220 map.

    80 00:02:07,230 –> 00:02:08,339 But on the left hand side

    81 00:02:08,350 –> 00:02:09,820 in the picture, so don’t

    82 00:02:09,830 –> 00:02:10,800 get confused here.

    83 00:02:11,380 –> 00:02:13,270 Then we go to the pre image

    84 00:02:13,279 –> 00:02:14,860 which is here again denoted

    85 00:02:14,869 –> 00:02:16,419 by the power minus one.

    86 00:02:17,190 –> 00:02:17,539 OK.

    87 00:02:17,550 –> 00:02:18,520 You could think this is an

    88 00:02:18,529 –> 00:02:20,220 overload for a notation but

    89 00:02:20,229 –> 00:02:21,880 indeed it fits together

    90 00:02:21,889 –> 00:02:23,660 because this is simply

    91 00:02:23,669 –> 00:02:25,410 the image of A and

    92 00:02:25,419 –> 00:02:25,839 F.

    93 00:02:26,440 –> 00:02:27,759 This comes from the 1 to

    94 00:02:27,770 –> 00:02:29,399 1 correspondence we have

    95 00:02:29,410 –> 00:02:30,759 by our bijective map

    96 00:02:30,770 –> 00:02:31,789 F.

    97 00:02:31,860 –> 00:02:33,210 Now with this example, you

    98 00:02:33,220 –> 00:02:34,919 should see that the definition

    99 00:02:34,929 –> 00:02:36,440 of an open map looks

    100 00:02:36,449 –> 00:02:37,770 similar to the one of a

    101 00:02:37,779 –> 00:02:39,119 continuous map.

    102 00:02:39,130 –> 00:02:40,380 However, it’s completely

    103 00:02:40,389 –> 00:02:41,970 different because it goes

    104 00:02:41,979 –> 00:02:42,960 the other way around.

    105 00:02:43,720 –> 00:02:44,220 OK.

    106 00:02:44,229 –> 00:02:45,500 Then let’s write down some

    107 00:02:45,509 –> 00:02:47,039 concrete examples of open

    108 00:02:47,050 –> 00:02:47,419 maps.

    109 00:02:48,199 –> 00:02:49,639 Maybe a simple example would

    110 00:02:49,649 –> 00:02:51,600 be to look at a normal function

    111 00:02:51,639 –> 00:02:53,110 between the real numbers.

    112 00:02:53,669 –> 00:02:55,320 Of course R is just given

    113 00:02:55,330 –> 00:02:56,559 with the standard metric,

    114 00:02:56,570 –> 00:02:57,699 which means the absolute

    115 00:02:57,710 –> 00:02:58,309 value.

    116 00:02:58,440 –> 00:03:00,279 And I want to send X to

    117 00:03:00,289 –> 00:03:01,160 X cubed.

    118 00:03:02,009 –> 00:03:03,639 This has to be an open map

    119 00:03:03,649 –> 00:03:05,300 because F is bijective

    120 00:03:05,470 –> 00:03:06,669 and the inverse is

    121 00:03:06,679 –> 00:03:07,520 continuous.

    122 00:03:08,410 –> 00:03:10,000 In the same way, we can construct

    123 00:03:10,009 –> 00:03:11,130 a counter example.

    124 00:03:11,139 –> 00:03:12,570 Just look what happens when

    125 00:03:12,580 –> 00:03:14,100 we send X to X

    126 00:03:14,110 –> 00:03:14,720 squared.

    127 00:03:15,179 –> 00:03:16,490 Of course, we can’t use the

    128 00:03:16,500 –> 00:03:18,169 general example here because

    129 00:03:18,179 –> 00:03:19,690 it’s simply not bijective

    130 00:03:20,250 –> 00:03:21,529 but still it could be an

    131 00:03:21,539 –> 00:03:22,289 open map.

    132 00:03:22,809 –> 00:03:24,009 Of course, this is not the

    133 00:03:24,020 –> 00:03:25,899 case, it’s not an open map.

    134 00:03:25,910 –> 00:03:27,130 As you can see when

    135 00:03:27,139 –> 00:03:28,850 considering the open

    136 00:03:28,860 –> 00:03:30,679 set as the open interval

    137 00:03:30,690 –> 00:03:32,169 minus 2 to 2.

    138 00:03:32,839 –> 00:03:34,369 Then the image is simply

    139 00:03:34,380 –> 00:03:36,100 the interval 0 to

    140 00:03:36,110 –> 00:03:36,550 4.

    141 00:03:37,360 –> 00:03:38,779 And there you see this is

    142 00:03:38,789 –> 00:03:40,710 not an open set in R

    143 00:03:41,369 –> 00:03:42,740 Now with your new knowledge,

    144 00:03:42,750 –> 00:03:44,460 what an open map is, we can

    145 00:03:44,470 –> 00:03:46,440 finally state the open mapping

    146 00:03:46,449 –> 00:03:47,039 theorem.

    147 00:03:47,600 –> 00:03:48,899 The important ingredient

    148 00:03:48,910 –> 00:03:50,039 you always should remember

    149 00:03:50,050 –> 00:03:51,699 here is that we need Banach

    150 00:03:51,710 –> 00:03:53,300 spaces on the left and the

    151 00:03:53,309 –> 00:03:54,179 right hand side.

    152 00:03:54,699 –> 00:03:55,949 So the completeness of the

    153 00:03:55,960 –> 00:03:57,699 spaces is essential here.

    154 00:03:58,500 –> 00:04:00,119 Now for bounded linear

    155 00:04:00,130 –> 00:04:02,100 operator T, where I use the

    156 00:04:02,110 –> 00:04:03,380 notation with the curved

    157 00:04:03,389 –> 00:04:05,070 B again, we

    158 00:04:05,080 –> 00:04:06,289 have that it is

    159 00:04:06,300 –> 00:04:07,270 surjective

    160 00:04:08,039 –> 00:04:09,699 if and only if

    161 00:04:09,710 –> 00:04:11,259 T is an open map.

    162 00:04:12,050 –> 00:04:13,889 And that is the famous

    163 00:04:13,899 –> 00:04:15,350 open mapping theorem.

    164 00:04:16,098 –> 00:04:17,759 It just tells you that

    165 00:04:17,908 –> 00:04:19,778 surjectivity for a linear

    166 00:04:19,829 –> 00:04:21,368 bounded operator between

    167 00:04:21,378 –> 00:04:23,229 Banach spaces is a strong

    168 00:04:23,239 –> 00:04:25,158 property because

    169 00:04:25,169 –> 00:04:26,928 it implies the openness.

    170 00:04:27,670 –> 00:04:29,399 Indeed, the implication from

    171 00:04:29,410 –> 00:04:31,040 right to left is not so

    172 00:04:31,049 –> 00:04:32,600 surprising and easy to

    173 00:04:32,609 –> 00:04:34,549 prove the actual work

    174 00:04:34,559 –> 00:04:36,109 one has to do from the left

    175 00:04:36,119 –> 00:04:37,279 to the right hand side.

    176 00:04:37,910 –> 00:04:39,130 Therefore, the proof of the

    177 00:04:39,140 –> 00:04:40,779 first implication is a good

    178 00:04:40,790 –> 00:04:41,880 exercise for you.

    179 00:04:42,000 –> 00:04:43,450 But for the other one, we

    180 00:04:43,459 –> 00:04:44,790 need a whole other video.

    181 00:04:45,410 –> 00:04:46,970 However, before we do that,

    182 00:04:46,980 –> 00:04:48,670 I want to use the next video

    183 00:04:48,679 –> 00:04:50,170 to show you an important

    184 00:04:50,179 –> 00:04:51,910 consequence of the open mapping

    185 00:04:51,920 –> 00:04:52,570 theorem.

    186 00:04:53,040 –> 00:04:54,410 It’s called the bounded

    187 00:04:54,420 –> 00:04:55,709 inverse theorem.

    188 00:04:56,209 –> 00:04:57,730 Hence please watch the next

    189 00:04:57,739 –> 00:04:59,609 video because this consequence

    190 00:04:59,619 –> 00:05:01,429 is very important in a lot

    191 00:05:01,440 –> 00:05:02,359 of applications.

    192 00:05:03,170 –> 00:05:04,510 So have a nice day and see

    193 00:05:04,519 –> 00:05:04,989 you then.

    194 00:05:05,000 –> 00:05:05,750 Bye.

  • Quiz Content (n/a)
  • Back to overview page