Serizawa

[knightlife]

What about this 3-cell rake parent?

x = 3, y = 3, rule = serizawa
2.A2$B.B!

That one is the same as the 3-cell pattern I posted yesterday (rotated 180 degrees).

[p46beth]

A slight modification of the best previous known quadratic growth pattern fits in the same bounding box (29 by 15) but has 4 fewer cells

Code: Select all

x = 29, y = 15, rule = Serizawa
5.A2.A$.B2.AB2ABA2.B$A.A8.A.A$4.B.2A.B8$20.A2.A$16.B2.AB2ABA2.B$15.A.
A8.A.A$19.B.2A.B!

I have thought of another way to compare pattern size: total cell value. The quadratic pattern I just posted is also the smallest known in total cell value.

For infinite linear growth, the smallest known is 5 (the 3x3 pattern with 3 cells).

Of course, we have arbitrarily selected one state to be "1" and the other to be "2", rendering this almost meaningless. It has slightly more meaning for totalistic rules, or any rule in which the numerical value of the cell is used in the step function.

[knightlife]

Further reduction of quadratic growth pattern:

Code: Select all

x = 20, y = 10, rule = Serizawa
.B2.B$A.2B.A7$15.B2.B$14.A.2B.A!

It's only 12 cells in a 20x10 bounding box.

[knightlife]

10-cell quadratic, 12 x 3 bounding box:

Code: Select all

x = 12, y = 3, rule = Serizawa
9.2A$A2BA4.B2.B$9.2A!

one cell wide quadratic is 12 cells, 19 x 1 bounding box:

Code: Select all

x = 19, y = 1, rule = Serizawa
A2BA3.A2BA4.A2BA!

[cloudy197]

Some interesting results with a collision between a p3 oscillator and a glider:

Code: Select all

x = 27, y = 15, rule = Serizawa
A.A$.B22.A.A$25.B8$3.A18.A$2.A.A16.A.A$.A18.A$2.A.A16.A.A$3.A18.A!

Some patterns and syntheses:

Code: Select all

x = 92, y = 38, rule = Serizawa
20.A$19.A.A$18.A3.A$3.A.A11.A.A.A.A13.A51.A$2.A.B.A9.A.A3.A.A11.3A16.
A2.A2.A2.A2.A2.A17.2AB$3.B.B9.A9.A12.2A14.AB2AB2AB2AB2AB2ABA17.A.B$2.
A.B.A9.A.A3.A.A11.3A14.A18.A17.B$3.A.A11.A.A.A.A13.A16.A.2A.8A.2A.A
18.B$18.A3.A68.B$19.A.A68.A$20.A3$89.A$19.A.A68.B$20.B15.A.A50.A$3.A.
A31.B$3.ABA$4.A14.A.A$A7.A5.A3.A.B.A3.A7.A$.B5.B7.B3.B.B3.B6.A2.A11.A
7.A2BA2.A2BA2.A2BA7.A$A7.A5.A3.A.B.A3.A$4.A14.A.A67.A$3.ABA82.2AB$3.A
.A83.A.B$20.B69.B$19.A.A68.B$89.A.B$90.B$90.B$91.B$90.B$90.B$91.B$90.
B$90.B$88.A2.B$89.2A!

EDIT: Spaceship collections:

Code: Select all

x = 92, y = 5, rule = Serizawa
.B5.A8.B7.B3.B6.B2.B7.B2.B2.B7.B2.B2.B2.B7.B2.B2.B2.B2.B$A.A3.ABA5.AB
.BA4.A.B.B.A4.B.2B.B5.B.2B.2B.B5.B.2B.2B.2B.B5.B.2B.2B.2B.2B.B$6.A2.A
3.A5.A4.A.B.A4.3A2.3A3.3A5.3A3.3A8.3A3.3A11.3A$8.A5.A3.A6.A.A6.A4.A5.
A7.A5.A10.A5.A13.A$26.A!

[knightlife]

Other spaceships including two that output "puffs of smoke":

Code: Select all

x = 72, y = 33, rule = Serizawa
10.A3.B13.B3.B3.B$9.ABA.A.A11.B.A.A.A.A.B$9.A.A.A12.3A2.A.A2.3A$13.A
13.A3.A.A3.A7$6.B3.A3.B15.A3.A3.B16.A5.B2.B5.A$5.A.A.ABA.A.A10.B2.ABA
.ABA.B.A14.ABA3.B.2B.B3.ABA$7.A.A.A.A11.A.A3.A.A.A2.A15.A2.A2.A4.A2.A
2.A$7.A5.A42.A12.A$61.A2.A9$.A15.A$ABA2.A10.ABA2.A$A.A.ABA9.A.A.ABA$
3.A3.A11.A2.A$6.A13.2BA12.B3.A3.B$34.A.A.ABA.A.A3$37.A.A.A$38.BAB!

EDIT:

A nice double backrake that produces clean big gliders:

Code: Select all

x = 18, y = 8, rule = Serizawa
.A5.A2.A5.A$ABA3.AB2ABA3.ABA$A2.A.A6.A.A2.A$2.A3.A.2A.A3.A$6.B4.B2$4.
A8.A$5.A6.A!

[Wojowu]

p12 agar

Code: Select all

x = 31, y = 31, rule = Serizawa
.B3.B3.B3.B3.B3.B3.B3.B$A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A$.B3.B3.B3.B3.
B3.B3.B3.B2$.B3.B3.B3.B3.B3.B3.B3.B$A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A$.
B3.B3.B3.B3.B3.B3.B3.B2$.B3.B3.B3.B3.B3.B3.B3.B$A.A.A.A.A.A.A.A.A.A.A
.A.A.A.A.A$.B3.B3.B3.B3.B3.B3.B3.B2$.B3.B3.B3.B3.B3.B3.B3.B$A.A.A.A.A
.A.A.A.A.A.A.A.A.A.A.A$.B3.B3.B3.B3.B3.B3.B3.B2$.B3.B3.B3.B3.B3.B3.B
3.B$A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A$.B3.B3.B3.B3.B3.B3.B3.B2$.B3.B3.B
3.B3.B3.B3.B3.B$A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A$.B3.B3.B3.B3.B3.B3.B
3.B2$.B3.B3.B3.B3.B3.B3.B3.B$A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A$.B3.B3.B
3.B3.B3.B3.B3.B2$.B3.B3.B3.B3.B3.B3.B3.B$A.A.A.A.A.A.A.A.A.A.A.A.A.A.
A.A$.B3.B3.B3.B3.B3.B3.B3.B!

Unit cell on torus

Code: Select all

x = 4, y = 4, rule = Serizawa:T4,4
.B$A.A$.B!

[knightlife]

This quadratic is not chaotic, it is a nice triangular array of gliders.

Code: Select all

x = 34, y = 21, rule = Serizawa
16.A5.A2.A5.A$15.ABA3.AB2ABA3.ABA$15.A2.A.A6.A.A2.A$17.A3.A.2A.A3.A$
21.B4.B2$19.A8.A$20.A6.A6$32.B$31.A.A2$.A5.B2.B16.A$ABA3.B.2B.B$A2.A
2.A4.A$2.A$7.A2.A!

Further optimization is possible, to clean up extra "dots".

EDIT:
This one has glider arrays moving both left and right:

Code: Select all

x = 34, y = 17, rule = Serizawa
16.A5.A2.A5.A$15.ABA3.AB2ABA3.ABA$15.A2.A.A6.A.A2.A$17.A3.A.2A.A3.A$
21.B4.B2$19.A8.A$20.A6.A5$.A5.B2.B$ABA3.B.2B.B20.B$A2.A2.A4.A19.A.A$
2.A$7.A2.A16.A!

[simsim314]

Single Directional Side Rake:

A p24 side-rake taken from the double rake (by 137ben) and one side cleaned with glider.

Code: Select all

x = 24, y = 30, rule = Serizawa
5$12.B3.B$11.A.3A.A$14.A$13.3A$8.B11.B$7.A.A9.A.A$14.A2$14.A$13.A.A$
14.A$13.A.A$13.A.A$12.A3.A$11.2A3.2A$6.B3.AB5.BA$5.A.A3.7A$13.3A$14.A!

A splitter that sits on rake

Code: Select all

x = 60, y = 64, rule = Serizawa
7$13.B3.B$12.A.3A.A$15.A$14.3A$9.B11.B$8.A.A.ABA.ABA.A.A$14.3A$15.B$
14.BAB$15.B3$15.A3$7.B$6.A.A4$11.A11.2A$10.ABA11.BA$9.ABA11.2A$10.A9$
35.2A$36.BA$35.2A10$47.2A$48.BA$47.2A2$53.B$52.A.A!

As mentioned before one can eliminate all the side tracks (or any specific by choise):

Code: Select all

x = 75, y = 73, rule = Serizawa
3$14.B3.B$13.A.A.A.A2$14.A3BA$10.B5.A5.B$9.A.A3.A.A3.A.A$15.ABA2$16.A
7$8.B$7.A.A.2A7.2A$11.B9.BA$11.2A7.2A10$32.2A$33.BA$32.2A10$44.2A$45.
BA$44.2A$49.B$48.A.A2$41.A$40.ABA$40.A.A6$55.A$56.B$43.A11.A6.A$42.B
18.ABA$36.A6.A17.A.A$35.ABA$35.A.A2$47.A.A$48.B!

And finally two puffers that clean each other to create a side-rake

Code: Select all

x = 112, y = 200, rule = Serizawa
23$45.B$41.B2.A.A$40.A.BA$41.B.A$40.A2.A$40.3A$39.ABA.A52.B$40.AB50.B
2.A.A$40.A.BA47.A.BA$41.ABA48.B.A$39.A2.A48.A2.A$38.A4.A47.3A$90.ABA.
A$91.AB$91.A.BA$39.A.A5.A44.ABA$40.A.A5.B41.A2.A$38.A2.A.A3.A41.A4.A$
40.A.A$41.A2$53.A36.A.A5.A$54.B36.A.A5.B$38.A14.A35.A2.A.A3.A$40.A50.
A.A$92.A2$59.A44.A$60.B44.B$38.A20.A29.A14.A$40.A50.A3$65.A44.A$66.B
44.B$38.A26.A23.A20.A$40.A50.A3$71.A$72.B$38.A32.A17.A$40.A50.A3$77.A
$78.B$38.A38.A11.A$40.A50.A3$83.A$84.B$38.A44.A5.A$40.A50.A4$87.2A$
38.A47.AB.A$40.A46.ABA.A$88.A3$81.2A$38.A41.AB$40.A40.2A8.A4$75.2A$
38.A35.AB11.A.A$40.A34.2A10.ABA.A$88.A3$69.2A$38.A29.AB$40.A28.2A20.A
4$63.2A$38.A23.AB23.A.A$40.A22.2A22.ABA.A$88.A3$57.2A$38.A17.AB$40.A
16.2A32.A4$51.2A$38.A11.AB35.A.A$40.A10.2A34.ABA.A$88.A3$45.2A$38.A5.
AB$40.A4.2A44.A5$38.A3.A44.A.A$43.B43.ABA.A$40.A.A45.A$41.B3$38.A9.A$
49.B41.A$48.A$89.B$88.A.A$91.B$38.A15.A35.A$55.B$40.A.A11.A$41.B$96.A
$97.B$38.A21.A35.A$61.B$60.A2$102.A$103.B$38.A27.A35.A6.A$67.B40.ABA$
40.A.A23.A41.A.A$41.B$105.A$104.A$38.A33.A32.A.A$73.B32.A$72.A4$38.A
39.A$79.B$40.A.A35.A$41.B3$38.A45.A$85.B$84.A4$38.A51.A$91.B$40.A.A
47.A6.A$41.B54.ABA$96.A.A$40.A$41.A51.A$41.A50.A$93.A.A$36.2A56.A$35.
AB$36.2A$29.B$28.A.A2$32.A$31.2AB$32.B2$33.A$33.2A$35.A!

And here is large glider rake:

Code: Select all

x = 24, y = 24, rule = Serizawa
3$10.B3.B$9.B.B.A.A$8.3A$9.A$12.A2.A$11.A2.ABA$11.2B.A.A$10.AB$11.ABA
$12.A2$11.A$17.A$10.3A3.A.A$18.BA$4.B3.A8.2A$3.A.A.B$8.A!

[simsim314]

Here is 5 cells quadratic:

Code: Select all

x = 0, y = 0, rule = Serizawa
B2$3.B2$B15$52.AB!

[simsim314]

And here is a collection of 3 cells infinite growth:

Code: Select all

x = 35, y = 1219, rule = Serizawa
B.B$7.B85$3.B$6.B$3.B75$5.B2$8.B2$5.B116$4.B$7.B$3.B113$9.B2$9.B.A
109$6.B$2.B3$6.A149$10.B$7.B7$9.A130$18.B$21.A2$17.B156$17.A3$17.B$
13.B123$32.B.A3$31.A124$34.A3$29.B2$29.A!

[dvgrn]

Here is 5 cells quadratic...

I take it you've done some searching for 4-cell quadratic-growth patterns, without finding anything? It looks like there's some hope for such a search, but it isn't trivial to decide which patterns will end up with quadratic growth.

For example, here's a 4-cell infinite-growth pattern that doesn't stop producing novel reactions until about 2.185 million ticks:

Code: Select all

x = 30, y = 16, rule = Serizawa
B2$3.B2$B11$29.A!

Any of probably hundreds of similar slow-to-settle patterns might end up in a stable rake-building orbit; no way to tell for sure without running each one until the novelty works its way out. Another 4-cell with good novelty potential early on is

Code: Select all

x = 30, y = 28, rule = Serizawa
B2$3.B2$B23$29.A!

EDIT: Yikes! Here's one with a series of false alarms early on, but then it starts doing really interesting quadratic-growth stuff around 3.5M ticks. It's hard to run this one far enough to see if the expanding region will be able to escape from the box that it starts in:

Code: Select all

#C 4-cell Serizawa quadratic-growth candidate --
#C moving the key cell by (2,1) appears to have the same effect
x = 32, y = 24, rule = Serizawa
B2$3.B2$B19$31.A!

By 3.74 million ticks the rake-making region seems to be strictly bounded -- as many rakes are being destroyed as created, so no long-term quadratic growth after all. Of course it's quite possible that another quadratic-growth region will start up somewhere in this monster -- novelty clearly isn't exhausted yet -- but it may make sense to look for a simpler example of 4-cell quadratic growth...!

[dvgrn]

... it may make sense to look for a simpler example of 4-cell quadratic growth.

Found one, with a slightly-revised search script -- originally I was only looking 16K ticks ahead, not nearly far enough to catch late novelty appearances. Now I'm waiting to see if it looks quadratic at 2 million ticks. Even that may not be enough in a few cases, but for the most part Hashlife is happily running away with the pattern by T=2M.

Code: Select all

x = 35, y = 20, rule = Serizawa
B2$3.B2$B15$34.A!

Don't expect anything interesting to happen until 44K ticks, when the rake-making puffer appears more or less out of nowhere. Unfortunately this is still not an entirely predictable quadratic growth rate, because of a secondary puffer that shows up at 77K and starts transforming the output of the first puffer in unpredictable ways. The search for simplicity continues...

[simsim314]

Found one

WOW! Yes I was definitely searching for 4 cell quadratics in two ways

1. Randomized small box.
2. The way you did, adding single stable cell to the 3 cell infinite linear growth.

And yes I totally missed the 2M novelty case. I run up to 8K and suspected it to be enough...

Well every search has to make assumptions...

-----

For now I suspect it would be pretty hard to prove that the quadratic pattern you found is actually a quadratic one, due to the chaos introduced later (as you mentioned). My 5 cell pattern is a clear cut.

Well let's hope to find some more clear cut 4 cell quadratic.

[simsim314]

Found it! Very clear 4 cell quadratic:

Code: Select all

x = 125, y = 18, rule = Serizawa
A13$121.B2$124.B2$121.B!

Here is another one with late ignition (starts after 600K):

Code: Select all

x = 373, y = 17, rule = Serizawa
372.A12$B2$3.B2$B!

It has "twin" moving the static cell six units to the left.

[dvgrn]

Found it! Very clear 4 cell quadratic...

Ah, good -- congratulations! Now I won't be distracted by this search any longer.

I looked at lots of false positives, mostly within 60 cells of the 3-cell. The odds of interesting novelty seemed to go down significantly, about 40 cells from the 3-cell, and cells to the right seemed to give better results on average than cells to the left, so I hadn't tried running a search that far out yet.

The complex maybe-quadratic "novelty generators" that I posted above remind me of some of Bill Gosper's experiments in Conway's Life from a few years ago. Different types of puffers and rakes have different large-scale signatures in a long-running pattern. It might be possible to find a few new rakes or puffers in Serizawa by looking for unfamiliar fingerprints in these patterns, in the same way that you can recognize switch engines in a zoomed-out Life pattern.

Here is another one with late ignition (starts after 600K):...

This seems to be a copy of the previous pattern.

[simsim314]

This seems to be a copy of the previous pattern.

Oops copy-paste bug...now fixed.

[simsim314]

Yes there is a lot of "quadratic mess". Here is an example of something which I couldn't even figure out if it quadratic or not, because it's so messy, and it's simply stuck golly before the growth rate is clear:

Code: Select all

x = 464, y = 14, rule = Serizawa
463.A9$B2$3.B2$B!

Anyway we also should prove that quadratic is impossible with 3. Probably some kind of brute force.

EDIT

I was thinking of GOL quadratic search, and the problem is that linear growth is so much rear in life unlike in Serizawa where it's pretty common. This makes the quadratic patterns in Life even more rear, so the current 26 cells record can't be so simply beaten. The current search in Serizawa with minor optimization took me an hour or two, in GOL it would probably take weeks even months.

[dvgrn]

Yes there is a lot of "quadratic mess". Here is an example of something which I couldn't even figure out if it quadratic or not, because it's so messy, and it's simply stuck golly before the growth rate is clear...

Wow, that's an impressive one. By 1.5M ticks it seems to have decided to go quadratic in the first and third quadrants, anyway.

The second quadrant is prolific enough that it will also contribute significantly to the population count. Probably the vertical bars will continue to appear irregularly, or they might settle into periodicity eventually... there's some fairly complicated recursive math happening there, with gliders from lower down bubbling up unpredictably to start new northward puffers. Anyway, it's certainly not clear that even an approximate growth rate can be calculated without running this one a lot farther.

Not bad for a four-cell starting pattern!

[knightlife]

I have been searching for a reaction to make a simple "big glider" gun.
This p64 "big glider" gun uses a 4-glider reaction I found:

Code: Select all

x = 60, y = 120, rule = Serizawa
34.A$42.A3$34.B6.A.A$30.A2.B13.A$34.B7$43.A$29.A16.A$33.A.A7.A3$34.A
12$41.A.A$42.B8$5.A$14.A3$4.A8.A.A13.2A$.A28.BA$4.A24.2A6$4.A.A8.A$A
3.ABA11.A$4.3A8.A3$5.A40.B$14.A30.A.A6$36.A$35.ABA$35.A.A11$28.A3$27.
A.A7.A$23.A16.A$37.A7$27.3A16.B$24.A2.AB12.A3.A.A$27.3A5.A.A3$36.A$
28.A7$54.A3$45.A7.A.A$42.A16.A$45.A7$54.B$41.A13.B2.A$45.A.A6.B3$46.A
$54.A!

I havn't found a good 3-glider reaction that doesn't require excessive cleanup.
Extraneous gliders tend to interfere with the next synthesis.

[simsim314]

This p64 "big glider" gun uses a 4-glider reaction I found:

It's possible to reduce it to three "glider gun loops":

Code: Select all

x = 109, y = 167, rule = Serizawa
11$60.A$88.A3$59.A11.A15.A.A$56.A13.B22.A$59.A11.A6$59.A.A$60.B28.A$
55.A36.A$59.A.A27.A3$60.A20$32.A$23.A63.A.A$87.ABA$88.A$22.A.A2.A5.A$
18.A9.B7.A$27.A5.A$23.A$22.ABA$23.A5$23.A$22.ABA$22.A.A3$31.A.A53.A.A
$31.ABA54.B$32.A7$83.A$92.A$85.A2$22.A35.A13.A$19.A39.B13.B$22.A8.A.A
24.A13.A3$32.A53.A$23.A24$92.A$91.ABA$91.A.A21$64.A3$63.A.A27.A$59.A
36.A$93.A7$63.A11.A$60.A13.B22.A$63.A11.A15.A.A3$92.A$64.A!

Although it contains more circulating gliders it has only three gun loops.

[Antonin Duda]

just casually posting this so the thread appears on top and maybe, more interesting things, like smaller self constructing spaceships, could be posted here much more easily

Like this small novelty generator, which stabilizes after ~3.2M generations

Code: Select all

x = 122, y = 70, rule = Serizawa
B2$3.B2$B62$119.B3$117.B3.B!

And this alternating puffer:

Code: Select all

x = 72, y = 111, rule = Serizawa
27.B3.B$26.A.B.B.A$27.A3.A2$27.A3.A$26.ABA.ABA$27.A.A.A$27.5A$28.B.B
2$28.B.B$27.B3.B$27.2A.2A$28.ABA$27.2B.2B2$17.A23.A$16.B5.A.2A7.2A.A
5.B$17.A6.4A3.4A6.A$23.3A2.B.B2.3A$25.A.A3.A.A$26.B2.A2.B2$29.A6$29.A
2$27.A3.A3$8.A20.A20.A$7.B43.B$8.A20.A20.A4$65.A$66.B$65.A2$24.2A7.2A
$25.BA5.AB$5.A19.A7.A19.A$4.B49.B$5.A23.A23.A3$27.A3.A2$15.A6.B.2B7.
2B.B6.A$14.B6.B2.B9.B2.B6.B$15.A8.3B5.3B8.A$23.B11.B2$21.B7.A$20.A.A$
29.A5$25.A.A3.A.A$23.A2.B5.B8$21.A8.A2$18.A21.A$26.A31.2A7.A$16.A22.A
19.BA7.B$2.2A54.2A7.A$.AB11.A7.A$2.2A20.A9.A$29.A2$29.A4$10.A.A$10.AB
A$11.A23.A8$37.A$21.A.A2.A$18.A2.ABA16.A$22.A9.A$19.A22.A2$36.A7.A$
24.A9.A$29.A2$29.A!

[FWKnightship]

P15 gun:

Code: Select all

x = 27, y = 27, rule = Serizawa
5.A15.A4$4.A4.A7.A2.A.A$A7.B7.B$4.A4.A7.A2$21.B$4.A.A13.A.A$5.B6$21.B
$4.A.A13.A.A$5.B2$9.A7.A4.A$A9.B7.B7.A$4.A.A2.A7.A4.A4$5.A15.A!

[eRroR_6o6]

pure chaos

Code: Select all

x = 33, y = 39, rule = Serizawa
3$16.B2.B$4.A2.A7.B.2B.A5.B2.B$3.AB2ABA5.3A3.A4.A.2B.B$2.A4.BA6.A3.A
9.3A$3.A.2A23.A$10.A$9.2AB$10.A.B$11.B$11.B$12.B$11.A5$11.2A$10.A2.B$
12.B$12.B$11.A.B$10.2AB$11.A6$10.2A$10.2BA$9.A.A$9.A.A$11.BA$9.A.A$
10.A!

Edit: 8-cell novelty generator thing

Code: Select all

x = 219, y = 246, rule = Serizawa
2$70.A13$191.B2$194.B2$191.B203$195.B2$198.B2$195.B13$74.A!

[confocaloid]

The supplied pattern "Delay27.rle" doesn't work with the supplied ruletable "Serizawa.rule" (the device is damaged after one use, and doesn't work second time):

Code: Select all

# A delay of 27 timesteps.
#
x = 30, y = 27, rule = Serizawa
8.A$15.A7.A4$7.A.A6.A7.A$3.A25.A$16.A7.A6$A8.A$.B13.A$A8.A12.A.A3$23.
A$8.A5$A$.B$A!

I double-checked whether the supplied ruletable defines the exact same rules as Figure 1 in the paper in "Serizawa86.pdf". The ruletable appears to match the paper correctly.

Figure 8b in the same paper shows the delay device, and that looks roughly the same way as the above pattern, but there are no visible gridlines, so it is hard to tell whether all the cells are placed correctly.

One possible way to fix the device is to translate the rightmost 11 nonzero cells to the right:

Code: Select all

x = 31, y = 27, rule = Serizawa
8.A$16.A7.A4$7.A.A7.A7.A$3.A26.A$17.A7.A6$A8.A$.B14.A$A8.A13.A.A3$24.
A$8.A5$A$.B$A!

Figure 1 from "Serizawa86.pdf"

serizawa-figure-1.png (98.78 KiB) Viewed 782 times

Figure 8b from "Serizawa86.pdf"

serizawa-figure-8b.png (37.82 KiB) Viewed 782 times