Christophe Somewhat Techy Website

My life new project - First iteration

2025-12-15T22:00:00+00:00

Inspiration to create a life calendar

Drawing inspiration from the following Buster Benson life in weeks which in turn inspired this page and this one and also from this Wait but Why article; I am trying to create my own version of it.

In particular I like this version from Gina Trapani, and the existing github repo

First iteration

Hard coded with colored decade, each square is a week.

How I work (2025)

2025-12-05T01:20:00+00:00

Location: Dublin, Ireland

Current Gig: Physical Digital Implementation Engineer at Synopsys.

One word that best describes how you work: juggling

Current mobile device: Pixel 9

Current personal computer: MacBook Air

Work Computer: Windows 11 laptop, No machine to unix workstation (vpn) and Grid engine to submit various tools jobs.

Updating from previous article 2019. Same job different company, different vibe.

I have now 2 type of working day: - Working from home - Office

On a Typical Work Day in the office

The big change from 2019, is the commute which is now a lot longer.

Wake up at 6:00AM:
get myself dressed.
prepare lunch box for the kid.
Take the bus 11 to city centre then 40D about 1 hour commute, up to 1h30 if bus do not link properly.
- During the trip, I would be either listening to podcast/music or reading a book
- Also check Work Email/Teams message and start replying/sorting.
- List/review tasks and priorities for the day on my work notebook.
Once I arrive in the office:
- Start on first task of the day and move on. Usually start with analysing results from overnight runs / review logs and reports.
- Loop over
- Multiple breaks (lunch/coffee) so i don't spend all day sit at my desk, I actually need to stand up and move every so often.
- 10 minutes before end of the day:
- Review tasks status, write next day task in notebook.
- Email progress and issues found.
- List additional tasks for following day.
On the way home, I may take the luas and pick up the kid from school. I also tend to leave early in the afternoon, in order to be at home for the first evening meeting.
Prepare dinner, family time, dinner all together.
3 days a week I have late meeting.
19:00 to 20:00 kid bedtime routine (brushing teeth, wash face, story time, quiet time).
Reading after that for me.

Appps, software, or tools I use

Notebook - draft paper.
Apps/Software:
- Google Keep for quick capture on my phone.
- vi/gvim still for me the best editor. NotePad++ on Windows, and been slowly converted to VScodium on Mac.
- Python whenever I can.

My workspace setup

In the office i have triple screen, with one always on the linux workstation connection, and the other alternating between Explorer (online documentation),outloo/Teams(email) and Onenote. Beside this I have my earphone, and a notebook open. Reduced to single external screen at home ( dual with the laptop one ).

My best time-saving shortcut

Reduce and Prioritize and focus on only one single task at a time for 10 minutes. Basically blocked focused session.

Be intentional with my time and the time of others.

Took me a long time to get it, and I still fall behind sometimes.

Besides my phone and computer, I could not live without a notebook and pen.

On good meetings

Good meetings are late those days, I am missing a clear outcome announced ahead of the meeting, however meetings do tend to last too long.

Listening habits while at work

Depending on the tasks:

mostly podcasts (Freakonomics, Planet Money, Rationally Speaking, ....)
if needing more concentration:
- Online Radio: BBC 6 recommends especially Mary Ann Hobbes.
- Jazz Playlist.

My Key Strength

Most other engineer are (much) better than me on very specific tasks/areas, but I have a broad array of knowledge in IC design, more fox than hedgehog.

Relaxing and recharging

Getting better at this through the years, I only need a simple walk in nature. What’s your sleep routine like? Are you a night owl or early-riser? Go to bed around 9PM-10PM and read a book. I am more of an early-riser, I feel much better when I can wake up and start the day early. I usually slump in the afternoon.

Best advice I have ever received

One received recently: Do not hesitate to voice your unhappiness about stuff at work, don't wait until you reach meltdown to ask for help or voice your concern and issues.

Analog Layout

2024-11-14T16:00:00+00:00

Analog Layout

As a profession

I have started my career in the semi-conductor industry as an analog physical designer. I wrote about it on my personal blog back in 2012 when I thought about leaving both the profession and Ireland. Although Analog layout is no longer direct part of my work, as I moved to digital physical implementation, I am still keeping an interest on the subject.

Analog Layout and AI

This video talking about the potential of the much talked AI/LLM use in analog layout.

Automation

From the very beginning, I started my career, I thought for sure this analog layout can be automated, and yet it is resisting it. Sure the digital part of IC layout is automated, yet human engineer still have to direct, correct and repair the automation.

Random thoughts on AI - July 2024

2024-07-08T20:00:00+01:00

So everyone is talking about AI, either panicking or fantazing about the rise of AI, once again the future is both arriving fast and very slowly at the same time. Here is a capture of my current thought on the subject of AI, actually more questions than answers. Personally I think this is another bubble like the crypto one, but with a bit more substance. However there is some inherent limitations.

Energy Training AI, especially with Nvidia GPU is power intensive. There is a rush to equipe data center with the latest GPU, which consumption keeps climbing (400W !!! ). The powering of all the promised AI will be limited by the amount of electricity we decide to contribute to it.
Training Data Soon we will run out of material to train AI, and even without the legal limitations of fair use. I do not think, we can train AI on data generated by AI, we will end up with the same problem we have with photocopier. On top of this, the current training data, is generated by human worker (not paid very well to say the least) which are used to label picture and videos. There is also the problem that the trained data might be stained by cultural bias (ie who is doing the training). Also can Ai expand beyond the training data, for example if we were to train an AI only with data generated before 1850, would AI be able to answer question about general relativity or quantum phenomenon. Actually can AI expand and imagine something "new" ? How will it react when faced with something never seen before ?
Diminishing Returns Even if we set aside the energy and data issues, it looks like we are going to hit the eternal law of diminishing returns. Idealist thinks we are about to reach an exponential return as we fed more data (which will be limited), a more realistic approach would suggest a linear relationship between amount of training and "intelligence" produced; but I think each percentage increment of "returned intelligence" will require always exponentiallly increasing amount of training data.

See on arxiv the Paper: No "Zero-Shot" Without Exponential Data and the Youtube video from Computerphile.

Most non-engineer believe AI will not be able to replace human in term of emotional interaction, I actually think the opposite. Emotional recognition might be easier to achieve than we think, and human are pretty bad at it ( we are not telepath ), so AI is likely to surpass humans.
We also have to recognise, we don't understand some serious things about AI and machine learning:
Why is the last LLM so stable ?
In particular why are they so resilient to overfitting ? In traditional machine learning, overfitting is always an issue, yet here the more parameters, the better the model is.... This is so counter-intuitive.
Hallucinations looks to be unavoidable.
Verification
As we trust more of our systems to AI systems, how will we verify them ? How will we fix them (they are a black box) ? How will we be able to detect hallucinations as they will become more difficult to detect.
In the end, will we be able to trust the AI system ? Or will we trust them too much ?

Having say all that, I also think AI could be really helpful and useful on sone problems.

Parsing through scientific paper We have (as a human species) generated a vast amount of scientific papers, actually in the last few decades an increasing number. So many that most scientist are not able to read of aware of all of them. AI could help read and parse all those paper, and look for few things, in particular:
detect fraudulent paper (data manipulation for example, already being used).
find common insights and unusual new ideas.
Parsing through legal documents Again the amount of legal paper, makes the use of AI ideal, to detect logical incoherence.

Finally, some have expressed hope that universal revenue might solve the massive loss of jobs caused by AI, but: - AI ( and robots ) seems to be a promise for people owning assets (ie rich people) to rely less and less on human workers, and not for human worker to be free (well free of having a job). - Who will pay the universal revenue ? Who are we going to tax to finance it, whith more and more tax reduction and tax evasion from the companies and richest people. - I can see that the price of labour will go down, as human worker will now compete with automated systems.

A Yard Sale model.

2023-08-16T00:00:00+01:00

Using jupyter notebook, to figure out unfairness in the world.

A Yard Sale Model¶

An interesting phenoma, that even playing a "fair" game, one player ( could be any ) alsways end up with all the money. I am trying to simulate this, with python. This is very inspired by this article from the Pudding: https://pudding.cool/2022/12/yard-sale/

First creating an individual

In [1]:

import random


class Person:
  def __init__(self, name, wealth):
    self.name = name
    self.wealth = wealth
    self.win = 0
    self.gamesplayed = 0
    self.ratio = 0
    self.max_bet_ratio = 0.25
    #Maximum bet limited to 25% of the wealth
    #Rounded to
    self.max_bet = round(self.max_bet_ratio*self.wealth)
  def __str__(self):
    self.ratio = 0 if self.gamesplayed == 0 else self.win/ self.gamesplayed
    return f"{self.name}\t(Wealth:{self.wealth})\tWin Ratio:{self.ratio}\tGames Played:{self.gamesplayed}"
  def gain(self,gain):
        self.wealth += gain
        self.gamesplayed += 1
        if gain > 0:
            self.win +=1
        elif gain == 0:
            #Compensating when one player is ruined and cannot bet any more moeny
            self.gamesplayed -= 1
        self.max_bet = round(self.max_bet_ratio*self.wealth)

Now coding the game, a simple flip the coin.

In [2]:

def flip(player1, player2):
    #Making sure that player are betting affordable bets.
    bet = min(player1.max_bet,player2.max_bet)
    winner = random.choice(["H","T"])
    if winner == "H":
        player1.gain(bet)
        player2.gain(-bet)
    else:
        player2.gain(bet)
        player1.gain(-bet)

Let's test that with 2 players, and playing 50 times

In [3]:

p1 = Person('chris',1000)
p2 = Person('dave',1000)

for i in range(50):   
    flip(p1,p2)


print(p1)
print(p2)

chris  (Wealth:1890)   Win Ratio:0.56  Games Played:50
dave    (Wealth:110)    Win Ratio:0.44  Games Played:50

Note that the game is fair as each player win roughly half of the time, however all the money end up in one player hands only.

Trying with 1000 games

In [4]:

p1 = Person('chris',1000)
p2 = Person('dave',1000)

for i in range(1000):   
    flip(p1,p2)


print(p1)
print(p2)

chris  (Wealth:2)  Win Ratio:0.44537815126050423   Games Played:119
dave    (Wealth:1998)   Win Ratio:0.5546218487394958    Games Played:119

Note that the betting stopped after few iterations iteration as one player is ruined.

In [5]:

players = [Person(str(i),1000) for i in range(10)]

And let them play against each others, many many times..

In [6]:

number_players = len(players)

print('Number of players',number_players )

for g in range(10000):
    team = random.sample(players[:], 2)
    for t in team:
        flip(team[0],team[1])

for p in players:
    print(p)

Number of players 10
0   (Wealth:9982)   Win Ratio:0.5015151515151515    Games Played:660
1   (Wealth:2)  Win Ratio:0.4909090909090909    Games Played:165
2   (Wealth:2)  Win Ratio:0.47058823529411764   Games Played:221
3   (Wealth:2)  Win Ratio:0.5311143270622286    Games Played:691
4   (Wealth:2)  Win Ratio:0.5284090909090909    Games Played:352
5   (Wealth:2)  Win Ratio:0.4425531914893617    Games Played:235
6   (Wealth:2)  Win Ratio:0.4840182648401826    Games Played:219
7   (Wealth:2)  Win Ratio:0.5158371040723982    Games Played:442
8   (Wealth:2)  Win Ratio:0.46464646464646464   Games Played:198
9   (Wealth:2)  Win Ratio:0.4482758620689655    Games Played:145

One player ended up with all the money, some were ruined only after 62 games...

So looks like ultra-rich are inevitable, even without taking talent into account.

In [ ]:

Useful UNIX/Linux tips

2020-10-16T14:00:00+01:00

Find

Find all my directories not accessed in the last 30 days in tmp area.

find /tmp/ . -user $USER -type d -atime 30

Sum the size of all gds found in a given directory

find /directory/of/interest/. -type f -name '*.gds' -exec du -ch {} + | grep total$

Find broken links

find . -xtype l

Disk space

Sum available disk in human readable format

df . -h

Sum used disk in human readable format, in the current directory

du . -sh

Disk used summarized by first 3 level of hierarchy

du . -h --max-depth=3

Vi (gvim)

Go to a file under the cursor in gvim:

gf open the highlighted fie in the same window.

Preference for vi: .vimrc

color pablo
syntax on
set tabstop=4
set autoindent
set number
set showmatch
set ignorecase
set smartcase

Date

Talking To Stranger (Book Review)

2020-06-29T12:05:00+01:00

Talking To Stranger by Malcolm Gladwell

Summary: It is about miscommunication and the "wrong" assumptions one makes when talking to strangers.
My opinion: First I have been a long time big fan of Malcolm Gladwell, and I enjoy the way he can summarize deep and complicated social and scientific insight into the format of a compelling story. Sometimes it can backfire and become over-simplistic, but the story is there, and we are stories people and Mr Gladwell is a great story teller. Now, I was a bit confused by this book, and in the end it is probably the work of his at least enjoyed, the issue he is exploring is very complex and he is not quite able to condense it into a simple storyline, because it cannot. It is pardoxical, that I was annoyed by the absence of a single punchline and simple explanation, when at the same time, I have been well aware that it is the exact default of Malcolm Gladwell writtings. May be this book is more suited to the podcast format
Key points:
You need to challenge your preconceptions, your existing notions
When confronted with somebody out of the "norm",
External Links:
https://en.wikipedia.org/wiki/Truth-default_theory

Open EDA iVerilog

2020-06-21T13:20:00+01:00

Tools (MacOSX):

Installation

iVerilog You can nstall it through brew

brew install icarus-verilog

Or MacPorts:

sudo port -v install iverilog

Running a simulation

First compile Verilog using iverilog:

iverilog -o test.vvp test_tb.v

Then running the simulation:

vvp test.vvp

Which should result in the following:

VCD info: dumpfile test.vcd opened for output.
done testing case           0
done testing case           1
done testing case           2
done testing case           3
done testing case           4
done testing case           5
done testing case           6
done testing case           7

Finally, have a look at waveform:

open -a Scansion test.vcd

More test.v:

module test(A, B, C, D, E);

   output D, E;
   input  A, B, C;
   wire   w1;

   and GATE1(w1, A, B);
   not GATE2(E, C);
   or  GATE3(D, w1, E);

endmodule

And the testbench:

// Verilog test bench for test
`timescale 1ns/100ps
`include "test.v"

module test_tb;

wire A, B, C, D, E;
integer k=0;

assign {A,B,C} = k;
test the_circuit(A, B, C, D, E);

initial begin

    $dumpfile("test.vcd");
    $dumpvars(0, test_tb);

    for (k=0; k<8; k=k+1)
        #10 $display("done testing case %d", k);

    $finish;

end

endmodule

Revisiting xmas kindle draw

2019-11-18T00:00:00+00:00

Using jupyter notebook, to figure out how to draw our next christmas kindle.

A very simple kindle draw¶

Revisiting the simple kindle draw. And getting utterly confused.

In [1]:

import random

def simple_secret_santa(names,printit=False):
    n = len(names)
    s = random.sample(names,k=n)
    out = []
    for i,name in enumerate(s):
        if i == len(s)-1:
            to = s[0]
        else:
            to = s[i+1]
        out.append(name+'-'+to)
        if printit:
            print(str(i+1) + "   "+name + ' gives to ' + to)
    return out

In [2]:

names = ['Ro','Pa','Ao','HaM',
        'Fi', 'Ci',
        'Te','Ja','Ch','HaC',
        'Co','An','It',
        'To',
        'Et',
        'Ro', 'Li']

In [3]:

s = simple_secret_santa(names,printit=True)

1   Te gives to Ro
2   Ro gives to Ci
3   Ci gives to Ch
4   Ch gives to HaM
5   HaM gives to Co
6   Co gives to HaC
7   HaC gives to Pa
8   Pa gives to To
9   To gives to Ao
10   Ao gives to Ro
11   Ro gives to Li
12   Li gives to Fi
13   Fi gives to It
14   It gives to Et
15   Et gives to Ja
16   Ja gives to An
17   An gives to Te

How random is this ?¶

First lets import matplotlib package so we can do some vizualisation.

In [4]:

%matplotlib inline
import matplotlib.pyplot as plt 
import matplotlib.gridspec as gridspec 
import numpy as np
import itertools

In [5]:

def monte_carlo(names,n=1000):
    comb = {name1+'-'+name2:0 for name1,name2 in itertools.permutations(names,2)}
    for i in range(n):
        for couple in simple_secret_santa(names):
            comb[couple] += 1
    fig1 = plt.figure()
    plt.plot([comb[k] for k in comb.keys()])
    plt.xlabel('Iterations')
    plt.ylabel('Count each combo')
    #plt.axhline(y=n, color='r', linestyle='-')
    plt.title('Xmas Kindle')
    plt.show()
    return comb

In [6]:

c = monte_carlo(names,n=1000000)

That is very odd, I would have expected each combination to appear roughly the same number of times, it looks like some combination are twice as frequent ? ... Let's try with a smaller set.

In [7]:

names2 = ['Jolene','Niamh','Sophie','Lorraine','Paschal']
#only 5 people in the draw, which should make 20 possible donor combinations
comb2 = {name1+'-'+name2:0 for name1,name2 in itertools.permutations(names2,2)}
print(len(comb2))

In [8]:

c2 = monte_carlo(names2,n=50000)

This looks about right.Let's used a reduced inital set.

In [9]:

c_Reduced = monte_carlo(names[:14],n=1000000)

Again this looks more as expected, after inspecting the initial names selection, I actually realized the name were not uniques.

In [10]:

duplicates = [names.count(n)>=2 for n in names]
print(duplicates)

[True, False, False, False, False, False, False, False, False, False, False, False, False, False, False, True, False]

In [11]:

# Replicating the issue with duplicate names
names3 = ['Jolene','Niamh','Sophie','Lorraine','Jolene']
#only 5 people in the draw, which should make 20 possible donor combinations, with a duplicate
comb3 = {name1+'-'+name2:0 for name1,name2 in itertools.permutations(names3,2)}
print(len(comb2))
c3 = monte_carlo(names3,n=50000)

In [12]:

#This explain the disruption in randomness so we can easily fix that by adding a check for name uniqueness
def isThereDuplicates(names):
    dups = [names.count(n)>=2 for n in names]
    if any(dups):
        print("Duplicate names")
        print(names[dups.index(True)])
        return True
    else:
        return False

isThereDuplicates(names)

Duplicate names
Ro

Out[12]:

True

In [13]:

# Fixing the duplicates

def simple_secret_santa(names,printit=False):
    n = len(names)
    s = random.sample(names,k=n)
    out = []
    if not isThereDuplicates(names):
        for i,name in enumerate(s):
            if i == len(s)-1:
                to = s[0]
            else:
                to = s[i+1]
            out.append(name+'-'+to)
            if printit:
                print(str(i+1) + "   "+name + ' gives to ' + to)
    return out

Family draws¶

However for a family draw, you may want to avoid gift between siblings ( my particular family case ) or between partners (not considered here), so only gift to cousin are allowed, to do this I am introducing a dictionary, with the name of the parent as value and the kid name as key.

In [14]:

def simple_secret_santa_avoid_siblings(names):
    givers = list(names.keys())
    recipients = []
    s = random.sample(givers,k=len(givers))
    out = []
    if not isThereDuplicates(list(givers)):
        for i,name in enumerate(s):
            possible = [r for r in names.keys() if (names[r] != names[name] and r not in recipients)]
            recipient = random.sample(possible,k=1)[0]
            recipients.append(recipient)
            print(str(i+1) + ': ' + name + ' gives to ' + recipient + ' ::: '+ str(len(recipients))+':'+str(len(possible)))
            out.append(name+'-'+recipient)
        #Check
        print('Recipients: '+str(len(recipients))+':: names:'+str(len(names.keys())))

In [15]:

names_siblings = {'Child1':'Parent1','Child2':'Parent1','Child3':'Parent1','Child4':'Parent1',
        'Child5':'Parent2', 'Child6':'Parent2',
        'Child7':'Parent3','Child8':'Parent3','Child9':'Parent3','Child10':'Parent3',
        'Child11':'Parent4','Child12':'Parent4','Child13':'Parent4',
        'Child14':'Parent5',
        'Child15':'Parent6',
        'Child16':'Parent7', 'Child17':'Parent7'}

In [16]:

simple_secret_santa_avoid_siblings(names_siblings)

1: Child11 gives to Child14 ::: 1:14
2: Child1 gives to Child11 ::: 2:12
3: Child9 gives to Child6 ::: 3:11
4: Child14 gives to Child3 ::: 4:14
5: Child3 gives to Child5 ::: 5:10
6: Child6 gives to Child9 ::: 6:12
7: Child12 gives to Child10 ::: 7:9
8: Child8 gives to Child17 ::: 8:8
9: Child7 gives to Child15 ::: 9:7
10: Child5 gives to Child8 ::: 10:8
11: Child4 gives to Child12 ::: 11:4
12: Child17 gives to Child13 ::: 12:5
13: Child10 gives to Child1 ::: 13:4
14: Child16 gives to Child4 ::: 14:3
15: Child2 gives to Child7 ::: 15:2
16: Child15 gives to Child2 ::: 16:2
17: Child13 gives to Child16 ::: 17:1
Recipients: 17:: names:17

In [ ]:

On prime birthdays.

2019-11-10T00:00:00+00:00

Using jupyter notebook, to figure out how my next prime brithday.

Find your prime and double prime birthday¶

Prime Birthday definition, we define a prime birthday as:

when your reach an age which is a prime number (2 year old,3 year old,5th birthday,7th,11th...)
when the year of your birthday is a prime. Double Prime Birthday:
when both your age and the year are a prime

How frequent are prime birthday, in your lifetime and double prime ?

In [1]:

#Check if a number is prime
# Using Sieve of Eratosthenes ( https://en.wikipedia.org/wiki/Sieve_of_Eratosthenes )
def ListPrimes(n):
    #init
    primes = [True for i in range(0,n+1)]
    for index,value in enumerate(primes):
        if index == 1 or index == 0:
            primes[index] = False
        if primes[index] == True:
            next_index = index*2
            while next_index <= n:
                primes[next_index] = False
                next_index += index
    return primes

def isPrime(n):
    primes = ListPrimes(n)
    return primes[n]

In [2]:

#Basic test
isPrime(3)

Out[2]:

True

In [3]:

isPrime(458)

Out[3]:

False

In [4]:

#Based on year you are born, calculte all prime birthdays.
def prime_birthday(year_born,age_limit=125):
    primes = ListPrimes(year_born+age_limit)
    bday  = []
    count = 0
    for age in range(age_limit+1):
        if primes[age] or primes[year_born+age]:
            count += 1
            bday.append({'age':age,'year':year_born+age,'Prime age':primes[age],'Prime Year':primes[year_born+age]})
    return bday

In [5]:

# example for someboday born in 1910
prime_birthday(1910)

Out[5]:

[{'age': 2, 'year': 1912, 'Prime age': True, 'Prime Year': False},
 {'age': 3, 'year': 1913, 'Prime age': True, 'Prime Year': True},
 {'age': 5, 'year': 1915, 'Prime age': True, 'Prime Year': False},
 {'age': 7, 'year': 1917, 'Prime age': True, 'Prime Year': False},
 {'age': 11, 'year': 1921, 'Prime age': True, 'Prime Year': False},
 {'age': 13, 'year': 1923, 'Prime age': True, 'Prime Year': False},
 {'age': 17, 'year': 1927, 'Prime age': True, 'Prime Year': False},
 {'age': 19, 'year': 1929, 'Prime age': True, 'Prime Year': False},
 {'age': 21, 'year': 1931, 'Prime age': False, 'Prime Year': True},
 {'age': 23, 'year': 1933, 'Prime age': True, 'Prime Year': True},
 {'age': 29, 'year': 1939, 'Prime age': True, 'Prime Year': False},
 {'age': 31, 'year': 1941, 'Prime age': True, 'Prime Year': False},
 {'age': 37, 'year': 1947, 'Prime age': True, 'Prime Year': False},
 {'age': 39, 'year': 1949, 'Prime age': False, 'Prime Year': True},
 {'age': 41, 'year': 1951, 'Prime age': True, 'Prime Year': True},
 {'age': 43, 'year': 1953, 'Prime age': True, 'Prime Year': False},
 {'age': 47, 'year': 1957, 'Prime age': True, 'Prime Year': False},
 {'age': 53, 'year': 1963, 'Prime age': True, 'Prime Year': False},
 {'age': 59, 'year': 1969, 'Prime age': True, 'Prime Year': False},
 {'age': 61, 'year': 1971, 'Prime age': True, 'Prime Year': False},
 {'age': 63, 'year': 1973, 'Prime age': False, 'Prime Year': True},
 {'age': 67, 'year': 1977, 'Prime age': True, 'Prime Year': False},
 {'age': 69, 'year': 1979, 'Prime age': False, 'Prime Year': True},
 {'age': 71, 'year': 1981, 'Prime age': True, 'Prime Year': False},
 {'age': 73, 'year': 1983, 'Prime age': True, 'Prime Year': False},
 {'age': 77, 'year': 1987, 'Prime age': False, 'Prime Year': True},
 {'age': 79, 'year': 1989, 'Prime age': True, 'Prime Year': False},
 {'age': 83, 'year': 1993, 'Prime age': True, 'Prime Year': True},
 {'age': 87, 'year': 1997, 'Prime age': False, 'Prime Year': True},
 {'age': 89, 'year': 1999, 'Prime age': True, 'Prime Year': True},
 {'age': 93, 'year': 2003, 'Prime age': False, 'Prime Year': True},
 {'age': 97, 'year': 2007, 'Prime age': True, 'Prime Year': False},
 {'age': 101, 'year': 2011, 'Prime age': True, 'Prime Year': True},
 {'age': 103, 'year': 2013, 'Prime age': True, 'Prime Year': False},
 {'age': 107, 'year': 2017, 'Prime age': True, 'Prime Year': True},
 {'age': 109, 'year': 2019, 'Prime age': True, 'Prime Year': False},
 {'age': 113, 'year': 2023, 'Prime age': True, 'Prime Year': False},
 {'age': 117, 'year': 2027, 'Prime age': False, 'Prime Year': True},
 {'age': 119, 'year': 2029, 'Prime age': False, 'Prime Year': True}]

In [6]:

#Summarize
def SumPrimeBday(year_born,age_limit=125):
    bday = prime_birthday(year_born,age_limit)
    count_prime_age = 0
    count_prime_year = 0
    count_double = 0
    for b in bday:
        if b['Prime age']:
            count_prime_age += 1
        if b['Prime Year']:
            count_prime_year += 1
        if b['Prime age'] and b['Prime Year']:
            count_double += 1
    return {'Prime Ages':count_prime_age,'Prime Years':count_prime_year,'Double':count_double}

In [7]:

SumPrimeBday(1976)

Out[7]:

{'Prime Ages': 30, 'Prime Years': 19, 'Double': 8}

So the number of potential prime age birthdays will be the same for everybody, 30 if you live till 125.

In [8]:

SumPrimeBday(1970,80)

Out[8]:

{'Prime Ages': 22, 'Prime Years': 12, 'Double': 7}

But only 22 if you live till 80. So the difference depending on your age will be the birth year.

In [9]:

#Using french revolution and current year as window
pbday_by_birth_year =[SumPrimeBday(year) for year in range(1789,2020)]

In [10]:

#Plot
import matplotlib.pyplot as plt
%matplotlib inline

years = [year for year in range(1789,2020)]
yp = [y['Prime Years'] for y in pbday_by_birth_year]

plt.plot(years,yp)
plt.ylabel('Prime Years count')
plt.xlabel('Year of birth')
plt.show()

So being born in the 60/70 does provide an advantage in term of primes years you will encounter (assuming you live till 125).

In [11]:

yd = [y['Double'] for y in pbday_by_birth_year]

plt.plot(years,yd)
plt.ylabel('Double Prime Birthday count')
plt.xlabel('Year of birth')
plt.show()

In [12]:

import math
import statistics
statistics.mean(yp)

Out[12]:

16.25974025974026

In [13]:

statistics.mean(yd)

Out[13]:

3.8181818181818183

In [14]:

print(statistics.mode(yd),statistics.mode(yp))

0 16

In [15]:

print(statistics.stdev(yd),statistics.stdev(yp))

4.191276466540036 1.2792925769576962

In [16]:

print(statistics.median(yd),statistics.median(yp))

1 16

In [17]:

no_double = [years[index] for index,value in enumerate(yd) if value == 0]

In [18]:

#People born in those years will have no double prime birthday !!
print(no_double)

[1789, 1791, 1793, 1795, 1797, 1801, 1803, 1805, 1807, 1811, 1813, 1815, 1817, 1819, 1823, 1825, 1827, 1831, 1833, 1835, 1837, 1839, 1841, 1843, 1847, 1849, 1851, 1853, 1855, 1857, 1861, 1863, 1867, 1873, 1879, 1881, 1883, 1885, 1889, 1891, 1893, 1895, 1897, 1901, 1903, 1907, 1909, 1913, 1915, 1917, 1919, 1921, 1923, 1925, 1927, 1933, 1935, 1937, 1939, 1941, 1943, 1945, 1951, 1953, 1955, 1957, 1959, 1961, 1963, 1965, 1967, 1969, 1973, 1975, 1979, 1981, 1983, 1987, 1989, 1993, 1999, 2003, 2005, 2007, 2011, 2013, 2017, 2019]

In [19]:

import numpy as np
# An "interface" to matplotlib.axes.Axes.hist() method
n, bins, patches = plt.hist(x=yd, bins=[i for i in range(20)], color='#0504aa',
                            alpha=0.7, rwidth=0.85)
plt.grid(axis='y', alpha=0.75)
plt.xlabel('Double Primes Birthday')
plt.ylabel('Frequency')
plt.title('Histogram')
maxfreq = n.max()
# Set a clean upper y-axis limit.
plt.ylim(ymax=np.ceil(maxfreq / 10) * 10 if maxfreq % 10 else maxfreq + 10)

Out[19]:

(0.0, 90.0)

So it turns out having double prime birthday in your lifetime is not usual.

In [20]:

len(no_double)

Out[20]:

So between 1789 and 2020, 88 years will result in no double prime birthdays over 231 years.

Completing Panini albums.

2019-11-07T00:00:00+00:00

Using jupyter notebook, to figure out how many packet I need to buy.

Question: How many stickers do you need to buy to complete a Panini album ?¶

Panini stickers album are famous ( especially during soccer world cup ) in many school yard, where multiple exchange of duplicate are taking place in order to achieve the evasive completion of an album. I am not sure I have ever been able to complete fully an album when I was a kid. This bring the question, what does it takes to be able to complete an album. First we will look at it without the possibility of exchange between collectors. Off course now as a parent the number of packet needed to purchase reveals the sad reality of cost.

In [1]:

# N total number of stickers in the album
N = 662

import math
import matplotlib.pyplot as plt

class album:
    def __init__(self, Nstickers):
        # Album start empty. making it private
        self.__album = {i:0 for i in range(1,Nstickers+1)}
        self.__Nstickers = Nstickers
    def addSticker(self,sticker_number):
        if sticker_number <= self.__Nstickers:
            self.__album[sticker_number] += 1
        else:
            print("Not a valid number.")
    def isAlbumFull(self):
        return all([v >= 1 for v in self.__album.values()])
    def countGot(self):
        return sum([v >= 1 for v in self.__album.values()])
    def duplicates(self):
        return sum([v > 1 for v in self.__album.values()])
    def display(self):
        plt.plot([self.__album[n] for n in self.__album.keys()])
        plt.ylabel('Sticker Count')
        plt.xlabel('Sticker Number')
        plt.show()

In [2]:

import random
def packet(Sticker_in_packet = 5, Nalbum=N):
    return random.sample(range(1,Nalbum+1), k=Sticker_in_packet)

In [3]:

a = album(N)

In [4]:

print(a.isAlbumFull(),a.countGot(),a.duplicates())

False 0 0

In [5]:

def fill_album(album,limit=50):
    packet_bought = 0
    tracker = []
    while not album.isAlbumFull():
        packet_bought += 1
        for card in packet():
            album.addSticker(card)
        tracker.append([album.countGot(),album.duplicates(),album.isAlbumFull()])
        if album.countGot() - N < limit:
            #print('INFO can buy the remaining',packet_bought)
            if packet_bought > N*10:
                return tracker
    return tracker

In [6]:

t = fill_album(a)

In [7]:

t[-1]

Out[7]:

[662, 658, True]

In [8]:

def plot_tracker(tracker):
    plt.figure()
    plt.subplot(211)
    plt.plot([y[0] for y in tracker])
    plt.axhline(y=N-50,linewidth=4, color='r')
    plt.ylabel('Sticker filled')
    plt.xlabel('number of packet bought')
    plt.subplot(212)
    plt.plot([y[1] for y in tracker])
    plt.ylabel('Duplicates')
    plt.xlabel('number of packet bought')
    plt.show()

def plot_filled(tracker):
    plt.plot([y[0] for y in tracker])
    plt.axhline(y=N-50,linewidth=4, color='r')
    plt.ylabel('Sticker filled')
    plt.xlabel('number of packet bought')
    plt.show()

def plot_duplicate(tracker):
    plt.plot([y[1] for y in tracker])
    plt.ylabel('Duplicates')
    plt.xlabel('number of packet bought')
    plt.show()

In [9]:

plot_filled(t)

The red line indicate the thresold you need to reach to be able to buy directly from Panini the missing cards. The company only allow you to purchase a limited number of individually chosen cards (usually 50). That is about 320 packets in our example !!! It is true that after that the completion become really hard.

In [10]:

plot_duplicate(t)

In [11]:

plot_tracker(t)

In [12]:

a.display()

In [13]:

#Monte Carlo
n_monte_carlo = 500
y1 = []
y2 = []
for i in range(n_monte_carlo):
    am = album(N)
    t = fill_album(am)
    y1.append(len(t))
    y2.append(t[-1])

plt.plot(y1)
plt.ylabel('how many bought')
plt.xlabel('MONTE Carlo runs')
plt.show()

In [14]:

print("Average",sum(y1)/len(y1),"Max",max(y1),"Min",min(y1))

Average 938.866 Max 1870 Min 591

In [15]:

y_dup = [yy[1] for yy in y2]
print("Mean",sum(y_dup)/len(y_dup))

Mean 654.64

In [16]:

print("Max duplicates",max(y_dup),"Min duplicates",min(y_dup))

Max duplicates 661 Min duplicates 624

See below in the links few article which examined the problem.

Other references: Reference

http://wjgsp.com/number-packs-panini-album/

https://stats.stackexchange.com/questions/101255/how-many-stickers-do-i-need-to-complete-my-fifa-panini-album

shown: https://youtu.be/vuGaez6oFvo

http://images.math.cnrs.fr/Petite-collection-d-informations.html?lang=fr

https://www.unige.ch/math/folks/velenik/Vulg/Paninimania.pdf

How I work (2019)

2019-05-23T01:20:00+01:00

Location: Dublin, Ireland

Current Gig: Lead Physical Digital Implementation Engineer at Adesto/S3 semiconductors

One word that best describes how you work: juggling

Current mobile device: Sony Experia (Android 8.0)

Current personal computer: MacBook Pro (old 2012)

Work Computer: Windows 10 laptop, No machine to unix workstation (vpn) and Grid engine to submit various tools jobs.

A little about my background and how I got to where I am today. Studied Physics in Clermont-Ferrand, France, followed on by a master in Microelectronics which I finished with a work placement in small company in Grenoble (Dolphin Integration). After that I decided to go abroad, and landed in Ireland on the west coast, there I transferred from Analog Layout to digital Physical Implementation. I work in Integrated Circuit design for a couple of companies, alternating between my two areas of expertise: analog and digital physical implementation, until I join my current employer.

On a Typical Work Day

Wake up at 7:00AM:
get myself dressed.
wake up kid, get them dressed and let have them breakfast.
prepare lunch box for the kid.
Bring kid to school by car, walk in the park (unless my wife does it).
Due to covid lockdown, working from home, so setup and start logging in from my kitchen.
Once at my desk:
- List/review tasks and priorities for the day on my work notebook.
- Check email/Jira tickets and email replies
- Start on first task of the day and move on. Usually start with analysing results from overnight runs / review logs and reports
- Loop over
- Multiple breaks (lunch/coffee) so i don't spend all day sit at my desk, I actually need to stand up and move every so often.
- 10 minutes before end of the day:
- Review tasks status, write next day task in notebook.
- Email progress and issues found.
- List additional tasks for following day.
My wife usually does the after-school pickup, but I might do it.
Coloring with the kid
Prepare dinner, family time, dinner all together.
19:00 to 20:00 kid bedtime routine (brushing teeth, wash face, story time, quiet time).
Reading after that for me.

Appps, software, or tools I use

Notebook - draft paper.
Apps/Software:
- Google Keep for quick capture on my phone.
- vi/gvim still for me the best editor. NotePad++ on Windows, and been slowly converted to MS Visual Code on Mac.
- Python whenever I can.

My workspace setup

In the office i have dual screen, with one always on the linux workstation connection, and the other alternating between Chrome(online documentation),outlook(email) and Onenote. Beside this I have my earphone, and a notebook open. Reduced to single external screen at home ( dual with the laptop one )

My best time-saving shortcut

Reduce and Prioritize and focus on only one single task at a time for 10 minutes. Basically blocked focused session. Be intentional with my time and the time of others. Took me a long time to get it, and I still fall behind sometimes. Besides my phone and computer, I could not live without a notebook and pen.

On good meetings

Good meetings are rare those days, I am missing a clear outcome announced ahead of the meeting, however meetings do not tend to last too long.

Listening habits while at work

Depending on the tasks:

mostly podcasts (Freakonomics, Planet Money, Rationally Speaking, ....)
if needing more concentration:
- Online Radio: BBC 6 recommends especially Mary Ann Hobbes.
- Jazz Playlist.

My Key Strength

Most other engineer are (much) better than me on very specific tasks/areas, but I have a broad array of knowledge in IC design, more fox than hedgehog.

Relaxing and recharging

Best advice I have ever received

One received recently: Do not hesiate to voice your unhappiness about stuff at work, don't wait until you reach meltdown to ask for help or voice your concern and issues.

Random thoughts from Pycon Dublin 2014

2014-10-13T00:05:00+01:00

Over the weekend I attended my first Pycon in Dublin, Ireland. It is organised by the wonderful crowd of Python Ireland, I would actually recommend that you go to one of their monthly meetups if you have any interest in the Python language. !day 2 Main stage

I was not sure what to expect, but the experience turned out to be truly enjoyable, and I finished up the weekend, my head full of ideas to investigate with python.

Here's my favourite highlights of the weekend, in no particular order.

Ian Ozsvald @ianozsvald with its talk titled: The Real Unsolved Problems in Data Science
A lightning talk by @saoili about getting into python jobs, which comfort me in my current chosen path. See her lightning talk slides.
On the same lightning talk show, a very entertaining demonstration of failing at data science (Homeless cloud by an engineer from rehabstudios, I sadly forgot the name), but also a silent presentation by @dontusethiscode
Rachel Willmer @rwillmer in her keynote: A Tech Person's Adventures in Startup, gave me some inspiration and food for thoughts on creating your own company.
I realize that Python was widely used in the financial industry (something I did not know before ) thanks to Dr. Yves J. Hilpisch @dyjh keynote : Python and the Financial Industry.
Finally a very cool demonstration, inside iPython notebook on how to build an earth centered universe by Brandon Rhodes @brandon_rhodes.

Overall it was fun, exciting and energizing conference, I will probably be back next year, more engaged.

The conference happen every year, have a look on the python ireland website for more information on Pycon. Also they usually make the presentation available on youtube, see Python Ireland channel.

UPDATE 2019: Sadly I did not engage more nor attended since 2015.