Mathematics

doc. Dipl.-Math. Jurjen Duintjer Tebbens, Ph.D.

Department of biophysics and physical chemistry, Faculta of pharmacy in Hradec

Králové, Charles University.

Winter term 2024-2025

1
Content

Contents of the rst lecture

1 Introduction

2 Mathematics and pharmacy

3 Functions

2
Introductory remarks

Lectures and seminars take place every week but during the rst
half of the semester only!

Seminars are compulsory, the credit test consists of being present at

all 7 seminars (there is no seminar on October 28, it will be
replaced). Lectures provide complementary information.

The written exam consists of 20 questions about all the material

from all seminars. 12 points are enough to pass and for the mark
'good'. Points for selected themes can also be obtained during
10-minute tests at the beginning of the 3rd, 5th and 7th
seminar. More details will be provided at the last lecture.

4
Introductory remarks

On the intranet (see Study materials of my department) you will nd the
(in due course actualized) les:

These slides.

The conditions to pass the credit test and the exam.

Some examples to exercice.

Literature: The textbook of doc. P. Klemery. More literature is

listed at the description of the subject in the SIS system.

Last year's course is available on Moodle as well. The contents have

somewhat changed this year, though.

5
Mathematics and pharmacy

In pharmacy, mathematis are used to quantify the inuence of input

variables (factors) on output variables (outcomes).
Examples of independent input variables (usually labeled x): Drug
dose, drug lipophilicity, organ acidity, etc...

Examples of dependent output variables (usually labeled y ): Drug

concentration in an organ, absorption rate, drug half-life, etc...

We describe this inuence using a mathematical function, e.g. a linear,

quadratic or exponential function.
The precise functions are usually obtained by means of experimental
measurements and often only estimates of the desired function are
possible.

7
The calibration curve
A typical example is the detection of the inuence of substance
concentration on absorbance in photometry:
At suitable wavelengths, this dependence is linear and the
corresponding graph, a straight line, can be estimated using several
measurements.

From the graph one can nd the particular mathematical

expression of the corresponding linear function :

(source DOI:10.1016/j.jiec.2014.06.022)
8
The calibration curve
Once we know the specic mathematical function, we can predict
the absorbance for other concentrations where we did not measure.
Or, on the contrary, we can calculate the original input
concentration corresponding to a measured absorbance, which
represents one of the main uses of spectrophotometry.
A substantial part of the course Mathematics is about the properties of
functions that are used to describe processes relevant to pharmaceutical
research and we will often analyze abstract functions.
However, it should be realized that when specic functions are derived
from laboratory experiments, then there are various sources of potential
errors.
Of course, the vast majority of laboratory measurements can only be
made with some error.
Measurement error is the dierence between the actual value of the
measured quantity, which we do not know, and the value determined by
the measurement.
9
Notations for inexact numbers
For laboratory practice, no measurement is a good measurement unless
its error is determined. We will get to the determination of measurement
error in a moment. When we have found the measurement error, then the
measured inexact number can be written in dierent ways. For example,
let us have measured a value of 4 kg with a measurement error of 0.2 kg:
with a notation of the form 4 ± 0.2 kg we use the absolute error
0.2 kg. The absolute error describes a plausible distance between
the measured value and the true value. The absolute error for the
measurement of x is usually denoted as ∆x and is expressed in the
measured units.
with a notation of the form 4 kg ± 5% we use the relative error
5%. The relative error is the ratio of the absolute error to the
measured value. The relative error for the measurement of x is equal
to δx = ∆x x and is dimensionless; it is often given as a percentage.
with a notation of the form 4.0 kg the last valid digit indicates the
precision: The imprecision of the measurement is at the level of the
last digit. In this example, it is at the level of "several"decimals, but
not at the level of whole numbers.
10
Rounding and errors

In the seminars, we will discuss the notion of valid digits and decimal
places in more detail.
If we are performing calculations between numbers with dierent
levels of imprecision, then we need to round the result appropriately.
Rounding of a number to the specied decimal place is done such, that
the dierence between the original and the rounded number is minimal. If
we round the last digit being 5 to the previous decimal place, then we
round upwards.
There are dierent rules for determining the level of imprecision after
calculating with inexact numbers; these rules dier from eld to eld
(and from subject to subject within your study).
As part of the subject Mathematics, we will only show a procedure for
determining the level of imprecision after applying a function to an
imprecise input value. This procedure is based on derivatives, which we
will discuss from week 5.
11
Error theory
As for the determination of measurement errors, we distinguish:
Outliers: they arise as the result of mistakes by the worker who
performs the measurement, of imperfection of measuring devices, of
unreliability of the senses, etc. Measurements with this error can be
recognized and must be removed from the measurement results.
Systematic errors: they aect the result according to some
consistent law; they usually have a "one-sided" character. The
average value of individual measurements can deviate considerably
from the correct value. They are removed by calibrating the
instruments, careful analysis of the method used, repeated
measurements by another worker. An example is the repeated
incorrect reading of results from deection instruments, i.e. reading
"from the side", not perpendicular to the scale.
Random errors: they arise from the joined inuence of random
elementary errors, both external (changes in pressure, temperature,
electromagnetic interference, . . . ) and internal (uctuations of the
measured quantity, thermal noise, . . . ).
12
The normal distribution
Usually, the totality of random errors is the sum of a high number of
mutually independent elementary errors. It is then normally distributed:
When we repeatedly measure the values of the same phenomenon, then
the values
tend to uctuate around some mean value
this uctuation is symmetrical about the mean value
small uctuations (close to the mean) are more common than large
uctuations.

C. F. Gauss, 1777-1855

Such behavior is represented by the Gaussian curve: a graph for the

relative frequencies of innitely many normally distributed observations:
13
The Gaussian (or bell) curve

The expected value of µ is the most frequently occurring value. The

standard deviation σ gives information about the general size of the
uctuation around µ.
Normal distribution of random errors is normal (i.e. usual). So if we have
removed outliers and potential systematic errors, then the standard
deviation σ gives us information about the general size of the random
error. It is assumed that the expected value µ is the actual value of the
measured quantity.
14
Descriptive statistics
Because of the errors, laboratory measurements are usually performed
repeatedly, e.g. in duplicate, three times or even 10 times, 20 times. This
will help us estimate µ and σ :
To characterize n repeated measurements of x1 , . . . , xn , one uses the
Sample mean or average:
Pn
i=1 xi
x̄ = ,
n
which can be understood as the "center of gravity of individual
observations". It is also an estimate for the expected value µ.

Sample standard deviation:

sP
n
− xi )2
i=1 (x̄
s= ,
n−1
which we can understand as the average distance of individual
observations from the sample mean. It is an estimate for σ.
15
More about statistics
The standard error of the mean:
s
SEM = √ ,
n

gives the average distance of the sample mean to the expected

value. Demonstration in Excel . . .
If we write down the results of repeated measurements in the form
x = x̄ ± s,

where x̄ is the sample mean, s is the standard deviation and errors are
normally distributed,
then it does not mean, that the expected value lies entirely in the
interval [x̄ − s, x̄ + s],
but it means, that the expected value lies entirely in the interval
[x̄ − s, x̄ + s] with 68.2%, i.e. about 2 thirds, probability.
16
More about statistics

Statistics always try to claim something about a general situation (about

a set of innitely many measurements or patients, the so-called
population) using a small sample (a few trials, patients).
During your studies, you will probably use, e.g.,
t-tests, which nd out on the basis of two small samples for two
dierent situations, whether in general there is a dierence between
the given two situations (populations).

condence intervals for a given parameter, e.g. the expected

value. A condence interval is calculated around a sample parameter
(e.g. the sample mean) and gives the interval in which the
population parameter lies with a given condence (probability).

Statistics are indispensable for pharmaceutical research. More information

is oered within the course Applied Statistics (3rd year).

17
Linear regression
As a result of measurement errors, we often have complications in
determining the valid particular functional dependency:

(source DOI:10.1016/j.jiec.2014.06.022)

There is no straight line that passes through the measured points,

but the functional dependency should be linear according to theory.
The displayed line is an estimate of the real line, which is called
regression line, and also belongs to the eld of statistics.
The value of R2 gives information about the quality of the "t". We
will focus on the calculation of the regression line in the lectures
about derivatives.
18
Denition of functions
A function f is a mapping of numbers x from the domain Df to
the image Vf according to a unique assignment f (x):

f : Df → V f ; x → f (x).

For example for

f (x) = x2 + 1
we have

Df = R, Vf = {x ∈ R|x ≥ 1} = [1, ∞).

For
√
f (x) = x+1
we have

Df = {x ∈ R|x ≥ −1} = [−1, ∞), Vf = {x ∈ R|x ≥ 0}.

20
Important sets of numbers

N = {1, 2, 3, 4, . . . }: natural numbers,

Z = {0, 1, −1, 2, −2, 3, −3, 4, −4, . . . }: integers,

p
Q={ q | p ∈ Z, q ∈ Z, q ̸= 0}: rational numbers,

√
R \ Q = { 2, π, e, . . . }: irrational numbers (denition will be
given later),

R: real numbers (all existing" numbers, other numbers are

not needed for this subject).

√
(C: complex numbers, e. g. −1, they are not needed).

21